Challenges, strategies and opportunities for wind farm incorporated power systems: a review with bibliographic coupling analysis

Wind power is a rapidly developing energy source. Many nations use wind power to meet a considerable amount of their energy needs. Moreover, the technology of wind power has evolved over the period of time. As a result, the wind farm-incorporated power system has received more attention for its outstanding contributions. The purpose of this study is to review the research works published on four key topics within the theme of wind farm-incorporated power systems. We survey the research papers that are featured in the Web of Science database. We employ an approach called Methodi Ordinatio to filter the papers. The publication of papers related to wind farm-incorporated power system has increased significantly, especially between 2018 and 2022. Therefore, we conduct a database search during this period and select important papers. Then we review and describe the technical challenges and solutions of these papers. Furthermore, a bibliographic coupling analysis is presented. The analysis shows that the journals such as Energy, Energies, and Renewable Energy are the leading journals publishing papers on all four key topics. The analysis further demonstrates that the focus of the researchers is on wind power forecasting, followed by energy storage systems, and wind farm layout optimization. The least focus is on optimal power flow.


Introduction
Wind power technology has been rapidly developed in recent years. Moreover, the cost of wind power has decreased. In many nations, wind power has emerged as a potential alternative to meet the increasing cost of fossil fuels and increasing energy demand. Wind power is also useful in dealing with environmental challenges. In the future, wind power will be one of the primary sources of electricity. This is due to the stringent emission regulations (Ongsakul and Dieu 2016), and advancements in wind power technology (Huh and Seo 2015). The incorporation of wind power into the energy network results in the transition of a passive energy network into an active energy network (Georgilakis Pavlos 2008). Because of the fluctuating nature of wind power, all the generated electricity may not be accommodated within the network. This excess of power must be curtailed. In other words, the entire energy output of the wind farm is not always used.
Hence, research works have been conducted in the past to lower the curtailment level in energy networks (Li et al. 2015). The battery storage system is a promising solution for reducing curtailment. When there is a surplus of energy, it is stored in a battery storage system. This surplus energy is then delivered to the grid when the demand is high. Nevertheless, the integration of the battery storage system into the power grid results in dynamic state operations. Hence, the power flow is modeled using dynamic model equations. It increases the complexity of the problem (Naidu 2002).
The uncertainty of wind power is another issue. Wind power is renewable, and it lowers pollution levels. These are the main benefits of wind power. Therefore, wind power is rapidly expanding. (Zhao et al. 2016). Because of wind speed variations, wind power generation is uncertain. This unpredictable nature of wind creates significant challenges in the operation and management of electric power systems. If wind power is accurately predicted, it can improve energy market efficiency, reduce operation costs, and increase the utilization of wind power. Therefore, accurate wind power prediction is important to operate the wind farm incorporated power system.
Another issue is wind farm layout optimization. In the early years, the layouts were organized based on empirical data. As a result of this, the materials and costs were wasted. The significance of wind farm layout optimization has been recognized with the advancement of wind power utilization technologies. Research scholars have carried out a number of studies on wind farm layout optimization (Tsvetkova and Ouarda 2021).
The optimal power flow plays a significant role in incorporating wind farms into the power grid. When it is incorporated, the power systems encounter operation and planning challenges. The primary objectives of optimal power flow are minimizing transmission loss, distributing the demands, and managing the dispatching process. In addition, it eases the operation process and reduces the overall cost of power generation Niknam and Golestaneh 2012).
This paper presents the key topics within the theme of wind farm-incorporated power systems. These topics include the effect of the energy storage system, wind power forecasting, wind farm layout optimization, and optimal power flow.
The paper is organized in the following manner: The paper selection methodology is described in "Methodology" section. "Challenges, strategies, and opportunities" section presents the review of significant papers on each topic and describes the challenges, strategies, and opportunities. "Bibliographic coupling analysis" section provides a bibliographic coupling analysis. "Conclusion" section concludes the paper.

Methodology
Based on the scientific significance, the papers are selected and reviewed. The papers are filtered using the Methodi Ordinatio (Pagani et al. 2015). Methodi Ordinatio is the paper selection methodology, which allots an Index Ordinatio (InOrdinatio) to each paper. This index considers the most important aspects of a scientific paper, such as citation count, journal impact factor, and publication year.
Prior to the systematic review, the paper selection and ranking tasks are completed. Figure 1 describes the search methodology followed in this paper.
Stage 1: Identifying the topic of review. This paper focuses on four key topics such as the energy storage system, wind power forecasting, wind farm layout optimization, and optimal power flow. Stage 2: Selection of keywords. The databases are searched for familiar keywords that are commonly used by the researchers. Stage 3: Selection of database. Scopus, Google Scholar, and Web of Science are the commonly used databases. In this paper, the Web of Science database is used. Stage 4: Searching the database. The papers are searched in the database using the keywords. Stage 5: Filtering process. The following filters are applied. The conference paper, book, and book chapters  (Pagani et al. 2015) are omitted. In order to review the most recent research works, the papers that have been published between 2018 and 2022 are taken into consideration. Stage 6: Identifying the number of citations, publication year, and journal impact factor. From the Web of Science database, information about the citation count of the paper, its publication year, and the impact factor of the journal are collected. Stage 7: Ranking the papers. The InOrdinatio value is calculated using the equation given below.
In the above equation, the impact factor of the journal is denoted by IF. The term represents the weightage factor. It varies from 1 to 10. When the value of is closer to 10, more weightage is given to the publication year, and vice versa. In this work, the value is taken as 10. The term Y r represents the research year, whereas the term Y p represents the publication year. The citation count of the paper is denoted by ∑ C i . The papers are then sorted in descending order. The top ten papers are chosen for review.

Challenges, strategies, and opportunities
In this paper, the information is retrieved from the Web of Science database. Several authors recognized this database as a complete and reliable database (Meho and Yang 2007;Waltman et al. 2010;Bornmann and Mutz 2015;García-Orozco et al. 2022). The Core Collection of Web of Science (Clarivate) features more than 21,100 journals. They are dispersed across more than 250 scientific disciplines. Furthermore, these journals are peer-reviewed and published worldwide. In total, this database contains 74.8 million records and 1500 million cited references from the year 1900 to date.
In the Web of Science, we use the basic search function. We conducted the data retrieval and search in April 2022. The search terms are as follows. For battery energy storage (1) system, the search term is "wind farm + energy storage system". For forecasting wind power, the search term is "wind power forecasting". For wind farm layout optimization, the search term is "wind farm + layout". For optimal power flow, the search term is "OPF + wind farm". Table 1 lists the search terms used and their corresponding Web of Science query links. We then apply the filters. Since 2010, there has been an upward research trend on wind farm-incorporated power system, and from 2018 onwards, there has been a sharp increase. Therefore, we consider the papers published from January 2018 to April 2022. We exclude the editorial, notes, letters, conference papers, and reviews in order to narrow down to search only the most important scientific contributions (Merigo et al. 2015). In this search, the total number of documents found for the search term "wind farm + energy storage system" is 8629. From these documents, 604 papers are filtered and ranked. The top 10 papers are selected and reviewed. For the search term "wind power forecasting", 10,683 documents are retrieved. We filter 2417 papers and rank them. We review the top 10 papers. For the search term "wind farm + layout", 5701 documents are fetched. From these documents, 434 papers are filtered and ranked. The top 10 papers are selected and reviewed. For the search term "OPF + wind farm", 973 documents were found, and 42 papers were filtered and ranked. We review the top 10 papers.
The following subsections explain the four key topics with the illustrations showing the basic functioning of each of the topics. The influencing factors, challenges, strategies, and opportunities are also described.

Energy storage system
Energy storage system plays a significant role in modern power system. It has the potential to handle intermittent renewable energy (Luo et al. 2015). Other significant applications of energy storage system are the expansion of transmission networks (Yekini Suberu et al. 2014), frequency and voltage regulation (Del Rosso and Eckroad 2014), and reduction of transmission network congestion (Divya and Østergaard 2009;Das et al. 2015). Wind farm + energy storage system https:// www. webof scien ce. com/ wos/ woscc/ summa ry/ 99eff 7c0-20a1-4d04-a26b-1f9c0 66b50 39-334bf a1a/ relev ance/1 Wind power forecasting https:// www. webof scien ce. com/ wos/ woscc/ summa ry/ 686d6 8c7-5fdf-4014-9b73-f8bcc 4c52c 2c-334c0 1e2/ relev ance/1 Wind farm + layout https:// www. webof scien ce. com/ wos/ woscc/ summa ry/ 6b768 3c1-e59f-4ef8-a1ad-6463b 23b05 1e-334c0 c5e/ relev ance/1 OPF + wind farm https:// www. webof scien ce. com/ wos/ woscc/ summa ry/ d7287 4a0-58d2-4d99-81d3-56fab 15d52 b8-334c1 145/ relev ance/1 The energy storage system is incorporated into the wind farm as shown in Fig. 2a. With the help of the energy storage system, the wind farm controls the power passing through the electrical cables and regulates the quantity of power. The storage system does not require any additional power electronics. As shown in Fig. 2b, it would work even with the variable speed drive (VSD) (Fingersh, 2004;Qu and Qiao 2011). Therefore, the DC power would flow between the battery and the wind turbine, whereas the AC power would flow from the VSD to the grid. The battery is charged when the wind farm produces excess power. This extra power is curtailed once the battery is full. When the wind farm produces less power, the battery provides additional power through discharging. This strategy makes an effort to maintain the power supply at a steady level. New methods to integrate energy storage system into power system have sparked a lot of interest. The primary goals of these methods include the following: (1) Ensuring the stable and safe power system operation; (2) Preventing the frequent starting and stopping of thermal power generation plants; (3) Increasing the performance of power generation equipment; (4) Minimizing the gap between the demand and supply of power (Aneke and Wang 2016; Aziz et al. 2022). The important factors influencing energy storage system are storage capacity, design of energy storage system, compression strategy, and optimal sizing. Over the last few decades, many researchers have analyzed energy storage systems. The important research works carried out on the topic of the energy storage system are ranked in Table 2.
The citations, year of publication, and journal impact factor decide the rank of a paper. The paper by Dui et al. (2018) is ranked first. Compared to other papers, this paper has received more citations. The journal impact factor is also high. Because of the recent publication and high impact factor, the paper by Razmi et al. (2021) is ranked second. The paper by Cao et al. (2020) is ranked third, as it has received  (Simpson et al. 2021) reasonably good citations and impact factor. Because of the recent publication, the papers by Huang and Khajepour (2022), Khosravi et al. (2021), and Du et al. (2021) find a place in the list. The papers by Yazdani et al. (2019), Baros et al. (2019), Moghaddam et al. (2018), andPaul et al. (2020) find a place in the list, as they have a reasonably good impact factor. These ranked papers are comprehensively reviewed and presented in the following paragraphs.
In recent years, wind energy has made a significant contribution to the power system. On the other hand, wind power curtailment has become an issue. The impact of wind power curtailment can be reduced through the use of battery energy storage. To optimize the battery energy storage capacity, it is important to consider the wind power uncertainty as well as the operational constraints of the power system. Dui et al. (2018) analyzed a system consisting of a thermal plant, battery energy storage, and wind farm. The authors presented a technique for determining the capacity of battery energy storage and optimal power flow in the system. In this research work, there is an opportunity to improve the optimization method for cooperating with multiple wind and battery energy storage system. Moreover, it is important to establish novel strategies as the number of battery energy storage systems installed in wind farms continues to expand. This will help wind farms to operate more efficiently.
Because of the fluctuation in wind speed, unstable and intermittent wind power generation occurs. One of the important energy storage technologies is compressed air energy storage (CAES). It is capable of dealing with the stochastic nature of wind farm power output. Furthermore, it contributes to peak shaving as well as auxiliary grid services. An environment-friendly CAES system was proposed by Razmi et al. (2021). Two nearby Iranian wind farms (Kahak and Abhar) were considered. During off-peak hours, the fluctuating and excess wind power was transmitted to the high-temperature thermal energy storage and compressor. The balance of the power was transformed into heat. This research can be extended by adding the amount of electricity available on the grid during peak hours. Cao et al. (2020) presented a control strategy for sizing energy storage system to integrate with a wind farm. For various energy storage system configurations, the fluctuation rates of the wind storage systems were calculated. A sizing decision map presented the results in a visual manner. There is scope for further research. A method for coordinating the various types of energy storage devices can be developed. Another area of investigation is the determination of the optimal set-point power. Huang and Khajepour (2022) proposed a new Adiabatic Compressed Air Energy Storage (ACAES) system. The system was based on a rotary valve design and a compression strategy, which stored and released energy as and when it was required. Thus, the performance of wind farms was improved. Lower cost, ease of use, and less space are the advantages of this system. If this system is further optimized, we can increase its efficiency. The battery management system and control methodology have a significant impact on the efficiency of the energy storage system. It is always recommended to have a minimum battery size and maximum battery efficiency. Moghaddam et al. (2018) proposed a controller method called "predictive controller" for a battery energy storage system. It improved efficiency and reduced the size of the energy storage system. Khosravi et al. (2021) proposed a hybrid energy storage system. A power management method was introduced for wind-hybrid energy storage system, power dispatching, technology selection, and optimal sizing of the hybrid energy storage system. The prescheduled power delivery and the wind-hybrid energy storage system performance were investigated in short-term power management. This method can be further developed for long-term power management. Yazdani et al. (2019) investigated three energy storage systems, such as liquid air energy storage, compressed air energy storage, and hydrogen energy storage. There are various other modern energy storage systems. Future research could study these systems. Meanwhile, the materials, energy, environmental and economic impacts can also be studied.
To coordinate the operations of battery energy storage systems and large-scale wind farms, Paul et al. (2020) determined optimal battery energy storage capacity by developing a multi-objective planning framework. Multiple objectives such as the curtailment of wind energy, loss of load hour, expected energy not supplied, wind turbine availability, battery cost, and battery lifetime were considered. This research can be further improved by integrating complex structures such as electricity market-oriented control strategy and dynamic degradation of battery capacity.
A double-fed induction generator is currently used in the most advanced wind generator. It integrates storage devices into the system. Baros and Ilic (2019) developed a consensus-based approach for energy storage system control. This approach was used to regulate the power output of wind farms in real-time, with sharing of power among the storage devices. A more complex communication network topology may be explored in future work. Du et al. (2021) established a smoothing technique, which is based on adaptive wavelet decomposition. It matches wind fluctuations to the grid-connected power quality criteria. The strategy and uncertainty can be examined further in future research.
Based on the above explanation and analysis, the challenges, strategies, and opportunities identified in the ranked research works for energy storage system are summarized in Table 3.

Wind power forecasting
The purpose of wind power forecasting is to give accurate wind power information for the upcoming day, hour, or minute (Zhao et al. 2022). We can divide the forecast into four horizons: long-term (1-7 days), medium-term (6-24 h), and short-term (30 min to 6 h) (Zhang et al.  (Ribeiro et al. 2022) for wind power forecasting. The statistical approach and physical approach are the two common techniques. The hybrid method is used in some models to combine the benefits of both statistical and physical approaches. The important factors influencing wind power forecasting are error factor, residual, and ramp event. The papers related to wind power forecasting are ranked in Table 4.
The paper by Hao and Tian (2019) receives the first rank. Compared to other papers, this paper has more citations. The paper by Dhiman et al. (2019) gets the second rank, as the journal impact factor is very high. Because of recent publication and journal impact factor, the paper by Yildiz et al. (2021) is ranked third. The papers by Sun et al. (2018) and Kim and Hur (2018) find a place in the list, as they have more citations. Because of the recent publication, the papers by Duan et al. (2022) and Von Krannichfeldt et al. (2022) are found the place on the list. The papers by Wang et al. (2020), Abedinia et al. (2020), and Dong et al. (2021) are listed as they have a high impact factor. The following paragraphs review the top 10 papers in wind power forecasting.
Hao and Tian (2019) proposed a nonlinear ensemble method for forecasting wind power. The authors considered the error factor and designed a two-stage forecasting framework. A multi-objective grey wolf optimization algorithm was developed to forecast wind power. Dhiman et al. (2019) developed a machine-intelligent wind forecasting model. It was constructed using wavelet transform and various types of support vector regression. The forecasting of wind farm was improved through this machine-intelligent hybrid methodology even in uncertain wind conditions such as ramp events.  (Niu et al., 2022) Yildiz et al. (2021) designed a two-step deep learning model for wind power forecasting. In the model, the spatial attention and spectral modules could be investigated to further increase the learning capability of the model. Moreover, the forecast accuracy could also be improved by examining different decomposition approaches. Wang et al. (2020) presented a wind power forecasting method to improve forecasting accuracy. This method was based on a hybrid Laguerre neural network and singular spectrum analysis. Sun et al. (2018) introduced a short-term wind power forecasting method. This method consisted of a waveletbased neural network and a hybrid clustering method. Variable features such as pressure and wind direction could also be investigated in the future. Abedinia et al. (2020) developed an improved empirical mode decomposition to decompose wind measurements. The application of more effective feature selection and improved forecasting engines are recommended as further works. Additionally, the researchers can also extend this forecasting model to a probabilistic version. Dong et al. (2021) proposed a hybrid forecasting model to improve the stability and accuracy of wind power forecasting. This hybrid model could be further studied in the future for probabilistic energy forecasting and online wind power forecasting. Duan et al. (2022) developed a hybrid forecasting model. In order to improve wind power forecasting stability and accuracy, the nonlinear weighted combination and decomposition strategy were used. To tackle the disadvantages of the linear weighted combination, two deep learning models were employed. In the future, the effect of non-Gaussian noise, reducing the complexity of the model and eliminating the delay characteristics could be considered.
The power grid with high wind power penetration requires probabilistic wind power forecasting in the decision-making process. Before being used to make predictions online, offline training is given to train the conventional probabilistic wind power forecasting models. However, the most recent information cannot be fully used by these models during the prediction process. To address this problem, Von Krannichfeldt et al. (2022) designed an online ensemble approach considering the most recent information for probabilistic wind power forecasting. Missing and faulty data is a significant challenge for online forecasting approaches. Therefore, this problem could be addressed in future work. Kim and Hur (2018) developed an enhanced ensemble method for short-term probabilistic wind power forecasting. Better results may be attained in future research by improving the model and selecting a probabilistic forecasting range.
Based on the above explanation and analysis, Table 5 summarizes the challenges, strategies, and opportunities identified in the ranked research works for wind power forecasting.

Wind farm layout optimization
In the wind farm layout design, the wind turbines are optimally placed within the wind farm. A typical design problem is shown in Fig. 4. Because of turbulence, transmission line loss, and wake decay, a significant amount of power is lost. In order to reduce this power loss, the optimal wind farm layout design is necessary. In addition to this, the wind farm layout design significantly impacts the costs related to the maintenance, function, and installation of wind turbines. Previous research works concentrated on various streamlined details, such as wind farm site selection (Mokarram et al. 2022), stunned or adjusted rowcolumn (grid-like) design (Sorensen and Nielsen 2006), discrete pre-isolated grid points (Mosetti et al. 1994), continuous search space (Rivas et al. 2009), inconsistent separated turbine array (Patel 2005), and similarly dispersed turbine cluster (Crosby 1987). During the course of these investigations, a number of different objectives were considered. These objectives included levelized production cost (Elkinton et al. 2008), minimizing the cost of energy (Wan et al. 2010), annual energy production (Wagner et al. 2013), profit (Aytun Ozturk and Norman 2004), net present value (Song et al. 2017), and maximizing the total power output (Kaminsky et al. 1987). The important factors that influence wind farm layout optimization are spacing, turbine cooperation, wake effect, design methodology, cable routing, fatigue distribution, and atmospheric stability. It is possible to significantly reduce the costs by determining the most optimal placement and configuration of wind turbines. In the long run, it also increases the power production of wind farms. In addition to better space fixation, the installation of optimally located wind turbines is another important parameter in the optimization process.
The wind farm power output would be less than its real potential power when the wind turbines are not optimally allocated in the wind farm. It will have an impact on cost-benefit feasibility of the wind farm.
The wind farm layout design has several important aspects. The majority of previous studies have concentrated on minimizing the initial investment and maximizing the overall energy yield. Heuristic-based optimization methods have been developed to determine the optimal layout design. The wind farm layout-related papers are ranked in Table 6. The paper by Sun et al. (2019a) gets the first rank. This paper has more citations when compared to other papers. Because of the recent publication, the paper by Tang et al. (2022) is ranked second. The paper by Yang et al. (2019) is ranked third because of citations and impact factor. The papers by Liu et al. (2021), Moreno et al. (2021b), Guo et al. (2021) are found a place on the list because of recent publication as well as journal impact factor. The papers by Antonini et al. (2018) and Parada et al. (2018) are ranked in the list as they have more citations. Because of the high impact factor, the papers by Wu et al. (2020) and Huang et al. (2020) are ranked. Sun et al. (2019a) reduced the gap between wind turbines by developing a directional restriction approach. The influence of wind directions was additionally considered when compared to existing restrictions. Therefore, this approach was particularly efficient for sites with obvious prevailing wind directions. The   multi-population genetic algorithm was implemented for the wind farm layout optimization process. In the future, problems such as economic cost, power transmission, and cable layout could be further investigated.
In the traditional methods for wind farm micro-siting, it is commonly assumed that individual maximum power generation is obtained by each wind turbine. However, farm-level control action is usually implemented during daily operations so that the profit of wind plant is improved. Tang et al. (2022) considered the coordination of turbines as well as farm-level control operations. These authors achieved the optimum cost per unit of energy by investigating the wind farm layout optimization. In the future, detailed control actions for layout optimization and improvement of the algorithm efficiency could be studied.
Maximizing wind farm energy production is the primary goal of wind farm layout optimization. However, the wind turbine may be exposed to higher wake exposure when there is limited space for wind turbine installations. This effect is prominent in a mixed layout with wind turbines of various hub heights and capacities. Yang et al. (2019) ensured uniform wake loss of wind turbines and increased wind farm energy output. However, the minimization of wake loss standard deviation and maximization of energy were not simultaneously optimized in this study. In future studies, multi-objective optimization could be implemented to achieve these conflicting objectives.
The seabed terrain was not considered in many studies. It would raise the electricity cost of wind farms as it could impact the turbine supporting structure length. Liu et al. (2021) considered different seabed terrains and optimized the offshore wind farm layouts using a genetic algorithm. The authors made recommendations for future investigation. The yaw angle and the hub height of the wind turbine could be considered in the offshore wind farm optimization. The computational time cost will be high when the real application has more number of wind turbines. The researchers could develop a graphic processing unit for this optimization method. Future studies may also consider the wind farm maintenance cost and develop a model for evaluating wind farm maintenance cost. Other optimization algorithms, such as random search, local search, greedy search, particle swarm optimization, and extended pattern search can be used in future work.
The maximization of energy efficiency is the commonly used objective function in wind farm layout optimization. Other objective functions might also be considered. Moreno et al. (2021b) considered three minimization objective functions such as the losses of wake effect, the area of the overall wind farm, and the annual energy production cost. The authors designed a multi-objective lightning search algorithm for solving this multi-objective optimization problem. Wu et al. (2020) studied a wind farm layout optimization problem. The authors proposed a new method to optimize the wind farm layout. In this method, wake loss, wind distribution, and power production were taken into consideration. The placements of wind turbines in the wind farms were optimized using a genetic algorithm. The layout of cables had a significant impact on the transmission of power. The annual economic benefit was defined as the objective function. It was made up of the following components: land cost, cable cost, energy cost, and annual production benefit. There is space for additional investigation. The wind farm topographic condition was not taken into consideration in this problem. The wake loss method used in this work was unable to resolve the problems when applied to different types of wind turbines. Because wind turbines may have different rotor diameters. In the future, the wake loss method can be improved. It is possible to solve the layout problem with different wind turbine rotors. The cable route in wind farms with complex topography is another topic for further investigation.
Due to the fast expansion of wind energy, both land resources and wind resources are being placed under increasing strain. In order to address this problem, the wind industry has become increasingly concerned about the issue of obtaining better micro-sitting for wind turbines. Huang et al. (2020) proposed a method for designing three-dimensional wind turbine layouts. The authors simultaneously optimized the vertical hub height and horizontal layout. The micrositting and low-speed wind turbine technology were considered in this work. The layout optimization was performed by a harmony search algorithm. The multi-objective functions for turbine selection were solved by a sorting genetic Algorithm-II. The wake effect model based on a polar coordinate transformation was implemented in the optimization process. When the wind direction changes, it simplified the calculations of wake effects among the multiple wind turbines with different rotor diameters and hub heights.
In recent years, atmospheric stability has received a lot of attention. It influences the performance of wind farm. Nevertheless, most of the previous research works assumed that atmospheric stability is neutral. Guo et al. (2021) established a framework for wind farm layout optimization considering the impact of atmospheric stability. It was developed with the help of an improved Gaussian wake model. One of the important input parameters for this framework was the local atmospheric stability. Consequently, the method of calculating wind power generation was modified. This method was applied to optimize the wind farm layout of one real wind farm and three ideal wind farms. The universality of complex terrain problems can be considered for future investigation. Study on more common complex terrain appears to be a promising opportunity for improvement. Antonini et al. (2018) presented a continuous adjoint formulation to calculate the gradients in the wind farm layout optimization. The high-fidelity computational fluid dynamics (CFD) models were integrated into the system using the developed optimization methodology. It was successful in overcoming the high computational cost associated with the CFD-based optimization technique. Prior to applying any discretization, the general adjoint equations were derived by the proposed continuous adjoint formulation. Because of this, flexible implementation in the CFD software package was possible. The adjoint formulation has been presented for different conditions of the flow equation, such as turbulent flow, frozen turbulence, and laminar. The developed formulation was tested in a two-dimensional domain. The results were compared to those obtained by gradient calculations through finite-difference approximations. The developed adjoint method based on the gradient calculations was implemented in the gradient-based optimization method. Under a diverse range of wind resource scenarios, a two-dimensional wind farm layout optimization problem was solved. The application of this continuous adjoint formulation for two-dimensional wind farm layout optimization could be expanded to a more general three-dimensional formulation. In addition, it has the potential to solve the optimization problem of wind farm layout in complex terrain. Parada et al. (2018) proposed a method for designing highly efficient wind farms. The number of wind turbines and different spacings were considered. This method efficiently handled the use of real wind data and irregular terrain boundaries. It also maximized wind farm power. In the future, more realistic cases with different wind turbine types and irregular terrain may be considered. Additional practical objectives could also be considered.
Based on the above explanation and analysis, Table 7 summarizes the challenges, strategies, and opportunities identified in the ranked research works for wind farm layout optimization.

Optimal power flow
The wind farm-incorporated power system includes thermal power generation units, wind farms, transmission networks, distribution networks, and loads (Duman et al. 2020). Figure 5 illustrates a typical power system that is incorporated with a wind farm. In this wind farm incorporated power system, the electrical power is produced by the generators installed in the thermal power plants. The voltage is stepped up to a high voltage level by means of step-up transformers. It is sent out through the transmission network.
Nowadays, potential wind power sites are identified, and wind farms are installed widely. These wind farms are incorporated into the power system. The electrical power generated by the wind farm is transported to the nearby substations. Within the substation, the voltage is stepped down to a low level. The electrical power can be dispersed to the loads either through the primary distribution feeders or the secondary distribution feeders, depending on the requirements of the consumers .
In recent years, there has been substantial growth in the penetration of wind energy which has led to an increase in the complexity of power system networks. Therefore, in addition to the unpredictability of load demand, wind power also  Parada et al. 2018 Designing highly efficient wind farms Regularly arrayed layouts Consider more realistic cases and additional practical objectives introduces additional randomness and volatility to the network (Ida Evangeline and Rathika 2019). The presence of wind power fluctuation greatly impacts the tap changing, frequency control, voltage control, power control, and power system protection. The power system operator needs to make realtime adjustments to the parameters so that they can respond appropriately to the fluctuations. The optimal power flow (OPF) is used to obtain the optimal control parameters in each time period, and these parameters are subsequently corrected in real time based on actual measurements. The major goal of OPF is to choose the appropriate solution for the fluctuating and unpredictable characteristics of the power system network. This response will be either optimal or sub-optimal, depending on the circumstances ). In the traditional OPF, the optimal operation parameters are determined once. This is in contrast to the OPF for the wind farm incorporated power system, where the parameters are calculated multiple times (Ida Evangeline and Rathika 2021a). The important factors influencing the wind farm incorporated optimal power flow are wind power uncertainty, sizing and placement of FACTS devices, voltage stability, power system security, and wind energy planning.
Recently several investigations have been conducted to study wind power impact on optimal power flow. These studies are ranked in Table 8. Because of recent publication and high impact factor, the papers by You et al. (2022) and Ida Evangeline and Rathika (2022) are ranked first and third, respectively. The paper by Quan et al. (2021) is ranked second, as this paper has more citations. The papers by Sun et al. (2019b), Ida Evangeline and Rathika (2021b), Shilaja and Arunprasath (2019), Yao et al. (2020) are ranked as they have a high impact factor. Because of more citations, the papers by Rabiee et al. (2018), Salkuti (2019), and Biswas et al. (2020) are listed. You et al. (2022) proposed an optimal power flow model with risk-based contingency constraints. The robust optimization and an adjustable uncertainty were included in the model. With the incorporation of network contingencies, an adjustable uncertainty wind power set was created. On the basis of this uncertainty set, this model was able to secure the network against contingencies in a probabilistic manner. Hence, the optimal balance between risk and operation cost was maintained. In the meantime, a datadriven ambiguity set based on the L1-norm was employed. Therefore, this model was robust in terms of ambiguous wind power probability distribution. Moreover, when the available wind power data was increasing, the model size was unchanged. The authors also derived a decompositionbased algorithm. The performance of this model was verified by conducting experiments on IEEE 14 bus system and 118 bus system. It is possible to expand the scope of this model in order to overcome the limitations of the current model. The computation time required to solve the problem was significantly longer when more creditable contingencies were considered. Thus, some acceleration techniques such as umbrella contingency and parallel-computing framework may be incorporated. It is also possible to extend the scope of this model to take into account the distributional ambiguity of contingencies. Prior to the optimization process, a high occurrence probability may be assigned to more severe contingencies to model such ambiguity.
Because of the uncertainties associated with wind energy, its integration into the power system leads to difficulty in system operation. One of the aspects of uncertainty is a spatial correlation. Its origins can be traced back to a geographically dispersed distribution. Nevertheless, the majority of studies did not take into consideration of spatial correlation, which decreases the accuracy of the model. To preserve spatial correlation characteristics, Quan et al. (2021) proposed the high-dimensional t-copula model. A sufficient amount of simulated data was sampled through this model to form scenario generation. To decrease the number of scenarios, the K-means cluster algorithm was further employed. From the Australian Energy Market Operator (AEMO) dataset, four wind farms were selected to demonstrate the formation of the spatial correlation model. Finally, optimal power flow was conducted on an IEEE 24 bus system with and without considering spatial dependence. In the future, more efficient methods for planning and operation of superconducting wind farms could be investigated.
When wind power is incorporated into electric power grids, optimal power flow plays a significant role. In the present scenario, the conventional equations are not sufficient. Ida Evangeline and Rathika (2022) studied wind farm integrated power systems and explored the multi-objective optimal power flow problems. The objective functions such as power generation cost, voltage deviation, power loss, and carbon emission were considered. The authors solved the multiple-objective optimal power flow problem using the multi-objective horse herd optimization algorithm. The algorithm was enhanced by the addition of a decomposition strategy. On the other hand, the prior research works did not have efficient techniques to handle the constraints. To manage the variables out of bounds, a constraint handling technique was presented. Seven case studies were examined to verify the suitability and performance of the proposed algorithm. Experiments were carried out on wind farms integrated with IEEE 30, IEEE 57, and IEEE 118 bus systems. Shilaja and Arunprasath (2019) combined the gravitational search algorithm and the moth swarm algorithm to develop a hybrid approach. The authors applied this hybrid approach to solve optimal power flow problems in the power system with the wind farm. The fluctuating nature of the wind farm was demonstrated using the Weibull distribution function. Minimization of fuel cost and minimization of power loss are the objective functions. The authors carried out case studies and evaluated the efficiency of the hybrid approach. In order to mitigate congestion from the network and manage growing demand, the Flexible AC transmission system (FACTS) is commonly used in modern power systems. Biswas et al. (2020) examined stochastic wind power incorporated power systems with different FACTS devices such as thyristor-controlled phase shifter, thyristorcontrolled series compensator, and static VAR compensator. The authors performed test cases with fixed and uncertain load demands. Appropriate probability density functions were used to model the load demand and stochastic wind energy. The objective functions, such as, overestimation cost of the wind power, penalty cost for underestimation, the direct cost of wind power, and cost of thermal generation were considered. To reduce the total generation cost of the power systems, ratings and locations of the FACTS devices were optimized. The optimization task was performed by success history-based adaptive differential evolutionary algorithm. The superiority of the feasible solution method was used to handle the constraints. In the future, along with FACTS devices, solar power can be considered. Furthermore, the complex power systems incorporating the FACTS devices, such as static synchronous compensators and thyristor-controlled series reactors, could also be studied.
The inconsistency in the nature of wind power is a main problem in the wind farm-incorporated power system. The operator is required to select an operation strategy on a regular basis, as the wind power varies from time to time. To solve this issue, Ida Evangeline and Rathika (2021b) developed a real-time framework for multiple objective optimal power flow. The control variables were simultaneously optimized, and the dynamic optimization problem associated with it was solved in real-time. A decomposition-based multi-objective particle swarm optimization was developed to assess the feasibility and performance of the algorithm. Experiments were carried out on wind farm-integrated IEEE 30 and IEEE 57 bus systems.
The wind power from far-distance offshore is typically collected using the multiple terminal system. It is the efficient transmission approach for renewable energy integration. Despite this, the system operation faces a great challenge due to the lack of power flow regulation, especially in a multiple terminal system with conventional fixed DC droop control. Yao et al. (2020) developed a multiple terminal model to tackle this issue. The linear function of the control variable was used to represent the output power of the onshore converter. For this hybrid system, the authors developed a robust distribution chance-constrained model considering wind power uncertainty. Based on the Wasserstein-Moment criterion, this model enforced chance constraints on an ambiguity set using the worst-case probability distribution.
The model also ensured a secure operation without making any assumptions about the probability distribution of uncertainty. The 14-bus system and 1354-bus system were used to demonstrate the performance of the model.
The power system operation is influenced by the truncated probabilistic and irregular characteristics of wind power uncertainty. Sun et al. (2019b) proposed a probabilistic optimal power flow framework to tackle wind power uncertainties. The correlation between the wind power generated by multiple wind farms was taken into account. To illustrate the multimodal and irregular wind power distribution, a truncated multi-variate Gaussian mixture model (Trun-MultiGMM) was developed. To produce the wind power sample from the Trun-MultiGMM, a Markov chain quasi-Monte-Carlo model was designed. Experiments were carried out on multiple benchmark power systems. This research has the potential to be expanded in the future. The Trun-MultiGMM can be enhanced to obtain optimum solutions for extremely large power systems. It is possible to develop TrunMultiGMM-based load/wind forecasting model. Rabiee et al. (2018) considered the inherent uncertainty of wind energy and proposed a method for long-term wind energy planning. The information gap decision theory (IGDT) model was employed to manage the uncertainty of wind power. In addition, the power system security was ensured by applying the load margin as an index for voltage stability. In the initial operation point, the power flow equations (operation constraints) were considered in addition to the voltage collapse point. Thus, the safe operation of the power network was ensured via the voltage stability-constrained wind power planning method. The New England 39 bus system was used to demonstrate the performance of the method.
Salkuti (2019) studied a thermal-wind power system and solved the multi-objective optimal power flow problems. The output of wind energy was assumed to be a schedulable resource. Hence, the system operator can determine the wind power penetration limits. The Weibull probability density function was used to model the stochastic behavior of wind power. The objective functions, such as voltage stability enhancement index, transmission loss, and total generation cost, were optimized. The total generation cost was the summation of the wind power over-estimating cost, wind power underestimating cost, wind power cost, and thermal power cost. The Glow-worm swarm optimization algorithm was used to solve optimal power flow problems. The experiments were carried out on the wind farm-incorporated IEEE 30 bus and IEEE 300 bus systems. The scope for future research includes receiving subsidies/credits for local pollutant emission reduction, meeting the greenhouse gas target, and reducing carbon emission.
Based on the above explanation and analysis, the challenges, strategies, and opportunities identified in the ranked research works for optimal power flow are summarized in Table 9.

Bibliographic coupling analysis
Bibliographic coupling analysis is the systematic treatment of information, which enables the deep mining of scientific data through the use of computer algorithms . It facilitates the impact of the elements, understanding of the relationships, and connections assessed in the field of study (Gu et al. 2017). In addition, it helps to track and classify recent trends. It can be applied to organizations, journals, authors, productivity, distribution, periodicity, etc. (García-Orozco et al. 2020). This analysis is extremely useful as it encourages synergy, enables the development of new research, and provides the present state of scientific research (Bonilla et al. 2015;Alfaro-García et al. 2020).

VOSviewer software
We construct a bibliographic coupling network using VOSviewer software. This network consists of circles and arcs. The circles represent the journals. The larger circle represents a journal that has published more papers. The smaller circle represents a journal with less number of paper publications. The arc indicates the reference relation between the journals. Based on the shared references, the relationships between the journals are determined (Freire and Veríssimo 2020). For example, when journal A is cited by journal B and journal C, the relationship is measured (Kleminski et al. 2020). The journals that do not have the same references are located far away, while journals that cite the same references are closely located. In other words, when the journals have more common references, they are closely located and strongly related (more arcs) in the visualization. Color indicates a cluster of strongly related journals (Gracio 2016).
The important journals, number of papers, and number of citations related to the search term "wind farm + energy storage system" are shown in Fig. 6. The ENS journal has published more papers. The AE journal has received more citations. The ER journal has published less number of papers and received less number of citations. The bibliographic coupling network for this search term is given in Fig. 7. The journals ENS, IJEPES, ENY, and IACC have larger circles as they have published more papers. Among these four main journals, IJEPES and IACC are closely located as they have strong reference relations. In other words, these journals share more common references. The VOSviewer software has identified three clusters of journals. The pink cluster mainly consists of ENY, IRPG, RE, and ITSE journals. The green cluster mainly consists of ENS, AE, ITPS, and ECM journals. The blue cluster mainly consists of IJEPES and IACC journals. Table 10 lists the impact factors of the journals. The impact factor is high for AE journal. The SCImago Journal Rank is high for ITPS journal. The impact factor is low for JRSE journal. The SCImago Journal Rank is low for ITEES journal.
The significant journals, number of papers, and number of citations related to the search term "wind power forecasting" are shown in Fig. 8. The ENS journal has published more papers. The AE journal has received more citations. The ITEES journal has published less number of papers and received less number of citations.
The bibliographic coupling network of journals is given in Fig. 9. The journals ENS, ENY, AE, RE, and  IACC have larger circles as they have published more papers. Among these five main journals, ENS, AE, and IACC are closely located as they are strongly related to one another. The ENY and RE are strongly related to each other. So, they are closely located. The VOSviewer software has identified four clusters of journals. Two clusters are very small and can be ignored. The green cluster mainly consists of ENY, ECM, RE, and IJEPES journals. The pink cluster mainly consists of ENS, AE, IACC, and ITPS journals. Table 11 lists the impact factors of the journals. The AE journal has a high impact factor. The ITSG journal has a high SCImago Journal Rank. The JRSE journal has a low impact factor. The ITEES journal has a low SCImago Journal Rank.
The important journals, number of papers, and number of citations related to the search term "wind farm layout optimization" are shown in Fig. 10. The ENS journal has published more papers. The RE journal has received more citations. The IJGE journal has published less number of papers and received less number of citations.
The bibliographic coupling network is shown in Fig. 11. The journals ENS, RE, ENY, AE, and WE have larger circles as they have published more papers. Among these five main journals, ENS and RE have strong reference relations. So, they are closely located. The ENY and WE journals are closely located as they are strongly related to each other. The VOSviewer software has identified seven clusters of journals. The smaller clusters can be ignored. The pink cluster mainly consists of ENS and ASB journals. The green cluster mainly consists of ITSE and AE journals. The blue cluster consists of WE, ENY, and JWEIE. Table 12 lists the impact factors of the journals. The impact factor and SCImago Journal Rank are high for AE journal. The impact factor is low for JRSE journal. The SCImago Journal Rank is low for WEG.
The significant journals, number of papers, and number of citations related to the search term "OPF + wind farm" are shown in Fig. 12. The ENS journal has published more papers. The IRPG journal has received more citations. As shown in the figure, there is less number of papers and citations on this topic.
The bibliographic coupling network is given in Fig. 13. The journals ENS, ITEEE, IJEPES, ITII, IRPG, and RE have larger circles as they have published more papers. Among these six main journals, ENS, IRPG, and IJEPES are strongly related to each other. So, they are closely located. The VOSviewer software has identified five clusters of journals. The smaller clusters can be ignored. The pinky-brown cluster mainly consists of ENS, IRPG, and IJEPES journals. The green cluster consists of ITEEE, ITII, and IJGE journals. The cyan cluster consists of ITEES, ITSE, and ENY journals. Table 13 lists the impact factors of the journals. The ITII journal has a high impact factor. The ITPS journal has a high SCImago Journal Rank. The ITEE journal has a low impact factor and a low SCImago Journal Rank.

Conclusion
This paper has attempted to review four key topics within the theme of wind farm-incorporated power system. These topics include the effect of the energy storage system, wind power forecasting, optimization of wind farm layout, and impact on optimal power flow. The Methodi Ordinatio was applied to select the papers. After  applying the exclusion and inclusion criteria, the search was conducted on the Web of Science dataset. From the search results, it was determined that the Methodi Ordinatio filtered more recent papers, which have more citations, from the journals, which have high impact factors.
Since 2010, there has been an upward trend in academic research on wind farm-incorporated power system, and from 2018 onwards, there has been a steep rise. In the year 2021, more papers were produced. It highlights the growing involvement of academic institutions in the process of determining solutions for achieving sustainability in the power system. More than 60% of papers relating to wind farm-incorporated power system were published during the period of 2018-2022. Hence, a review of the papers that were published during In addition, the journals are evaluated both quantitatively and visually using Bibliographic coupling analysis. The results also highlight that the number of papers and number of citations of high-impact factored journals have been more than the low-impact factored journals. In terms of paper publications, the journals such as Energy, Energies, and Renewable Energy are the leading journals publishing papers on all four key topics related to wind farm-incorporated power system.
The number of papers on key topics reflects the emergence of wind farm-incorporated power system. It demonstrates that the main focus has been on wind power forecasting, followed by energy storage systems, and wind farm layout optimization. The least focus has been on optimal power flow.

Limitations and challenges
We integrated the classical content review and advanced bibliometric analysis. This paper helps to reduce some of the bias in the existing method. Even though this study has made  a significant contribution in terms of the scientific process and fundamental features related to the wind farm incorporated power system, its limitations cannot be ignored. Firstly, only the Web of Science database was used in this analysis. Additional research with a wider variety of integrated databases may give more light on a global viewpoint. For example, Scopus is a prominent webbased bibliometric database, which can also be taken into account. Therefore, we can effectively reduce the possibility of significant relevant research works being omitted. Additionally, we can further validate the robustness of bibliometric review.
Secondly, the bibliometric analysis is based on data. The formulation of the search methodology has a direct correlation with the accuracy of the data. The terms "wind farm + energy storage system", "wind power forecasting", "wind farm + layout", and "OPF + wind farm" were used in this study to describe the domain boundaries. It ensures that the obtained final data will have a high degree of relevance. At the same time, the search pattern comprehensively considers the title, abstract, and keyword information. However, there is still a very low chance that the significant paper may be omitted when an author does not use these terms in the keywords, abstract, and title. Hence, more relevant data may be obtained by experimenting with different retrieval string combinations. It does not mean that this paper has an inefficient retrieval strategy. More relevant data may be obtained when we search with different retrieval combinations.
Finally, there are some limitations in the bibliometric analysis. Despite the fact that they have impressive visualization mapping and direct statistical data, the bibliometric analysis only provides a macroscopic view of the topic. Therefore, we suggest a more in-depth textual analysis that places a greater emphasis on the topic. This review relies on the database that is updated in real-time. As a consequence of this, the bibliometric analysis results may change significantly over the course of several years. In spite of these limitations, this review paper gives a complete picture of wind farm-incorporated power system research.

Future research directions
This review paper provides a foundation for future research. Further research to be conducted includes the following: • Multiple databases such as Scopus, Google Scholar, and Web of Science can be used. • Country-specific primary data analysis and empirical analysis can be studied. • Various other key topics can be considered. • The performance of influential authors in the field may be studied. • Other renewable sources can be reviewed. • The dataset of the field between 1980 and 2022 can be reviewed. • The contributions of the research institutions may be analyzed.