The forecasting model research of rural energy transformation in Henan Province based on STIRPAT model

In order to find the model of rural energy transformation in Henan Province. In this paper, Tapio decoupling model is employed to investigate the pivotal factors affecting rural power consumption (PC) and total energy consumption (TEC) in Henan Province. In addition, PSO-BP is used to predict the values of each influencing factor in 2020–2025. Last, the STIRPAT model is used to forecast TEC and PC from 2020 to 2025 based on the data of rural energy consumption in Henan Province from 2009–2019. The results show that other factors besides population promote TEC and PC to different degrees. Moreover, the influencing factors, TEC and PC, form a virtuous cycle of mutual promotion. Then, TEC and PC consumption show an increasing trend year by year in 2020–2025. It is worth noting that after 2022, the variation of PC is greater than that of TEC. To sum up, improving rural electrification level is a necessary way to realize its low-carbon energy transition.


Introduction
Energy is an important material foundation for the survival and development of human society. It is vital to the national economy and people's livelihood and the strategic competitiveness of a country (Hussain et al. 2020). In today's world, the massive use of fossil energy brings about a series of problems in the fields of environment, ecology and global climate change . Countries around the world have taken the initiative to solve the dilemma and accelerate energy transformation and development. Yet now, around 2.8 billion people worldwide still rely on traditional biomass for cooking, heating and so on, 95% of which are concentrated in rural areas in Asia and Africa (Ma et al. 2021b;Han et al. 2021).
China is a big agricultural country with vast rural area and the rural population. About half of China's population lives in the countryside. Since the reform and opening up, the traditional rural energy consumption structure has also undergone significant changes (Jin et al. 2019). Nonetheless, China's combined agricultural and rural greenhouse gas emissions account for about 15% of national emissions . Promoting rural energy revolution and realizing rural energy transformation and upgrading are crucial measures to implement the spirit of the 19th CPC National Congress and General Secretary Xi Jinping's strategic thought on energy revolution. Therefore, it is an effective way to realize the green development of rural areas to explore the path of low-carbon energy transition in China.
In China's rural areas, energy consumption is mainly used for villagers' life, such as heating, cooking, lighting, transportation, and agricultural production activities, such as irrigation, planting, harvesting, and fertilization (Ma et al. 2021a). Due to China's rapid urbanization process, energy consumption in rural areas is increasing. Moreover, a large amount of CO 2 gas is emitted, which poses a threat to the health of villagers (Amagai et al. 2014).
Technological advances in energy efficiency and replacing conventional energy with renewable energy are generally regarded as effective measures to solve the problem of harmful gas emissions (Wu 2020;Du et al. 2021;Su et al. 2022). However, the consumption of fossil fuels supports the total consumption of rural energy, and the use of clean energy is less.
Consequently, many scholars have explored the transformation mode of rural energy. Han et al. 2021 used the space-time model to study the space-time characteristics of energy ladder in agriculture and rural life of developing countries, taking China as an example, by analyzing rural life and production activities. From the perspective of information reliability and standard non-compensation, Peng et al. 2021 constructed a decision support framework for managing the choice of new energy in rural areas. Also, Yahyaoui et al. 2016 analyzed the potential of renewable energy alternatives in rural areas. This is of great significance to the realization of the national low-carbon sustainable goal (Lemence and Tamayao 2021). He et al. 2014 proposed to reduce the total energy consumption in rural areas by improving the energy efficiency of rural buildings. Jia et al. 2022 andImran et al. 2019 analyzed the impact of the energy consumption revolution on the health of the elderly in rural areas and its mechanism. And put forward, should improve the rural medical security level. Also, the energy consumption revolution has an impact on the health of the elderly in rural areas Rose et al. 2018). In addition, Ma et al. 2021a andZi et al. 2021 analyzed the current situation of energy consumption in rural China, obtained the main factors affecting rural energy consumption, identified the key factors to improve energy efficiency, and provided a meaningful reference for rural low-carbon energy transformation (Zou and Luo 2019).
Thus, only by analyzing the current situation of rural energy consumption can we find a way out of rural energy transformation. However, most of the above literature is cross-regional analysis of the path of national rural energy transformation. Hence, this paper focuses on the analysis of the current situation of rural energy in Henan Province.
Also, some scholars have studied the transformation of rural electrification level. Studies on power consumption in rural areas have found that, with the improvement of rural living standards, energy consumption increases and electricity consumption also shows a rising trend (Tesfamichael et al. 2020;Agrawal et al. 2020). The increase of per capita income makes the transformation of rural energy use to commercialization (Asmare et al. 2021). That is, the villagers will gradually shift their dependence on non-commercial energy sources such as straw, firewood and bulk coal to electricity (Riva and Colombo 2020). The improvement of rural electrification can not only improve the efficiency of people's work and study, but also diversify people's ways of entertainment and study, thus improving people's physical and mental health (Han et al. 2020). Electrification has also increased the use of household appliances in rural areas and improved the quality of life of the people , thereby changing lifestyles, improving the per capita income, and accelerating the pace of urbanization in rural areas towards a relatively well-off society (Sedai et al. 2021).
Yet 1.2 billion people still do not use electricity, and most of them live in rural areas (Robert and Gopalan 2018;Vinicius et al. 2021). Therefore, the driving factors for improving rural electrification level should be explored while improving grid coverage. Ma et al. 2018 proposed that the purchase rate of energysaving household appliances is greatly affected by the education level of rural households, per capita income and government subsidies, as well as the influence of place and custom (Zou and Mishra 2020). Rahman et al. (2013) analyzed that the success of rural electrification program in Bangladesh is significantly related to the driving factors of system investment, community participation, anti-corruption characteristics, standardized practices and performance incentives. It provides a meaningful reference for rural electrification in other countries. Minaei et al. 2021 proposed to improve rural electrification by improving the quality of life in rural and remote areas through clean electricity.
Nonetheless, these studies only focus on the influencing factors of rural electrification from a single perspective. Correspondingly, the task of this paper is to explore the driving factors affecting rural electricity consumption from various aspects.
Henan is the largest agricultural province in central China, with a population of 96.4 million. Residents of Henan Province are heavily concentrated in the countryside. As an important province with a large population and large grain and agricultural production, Henan is the main battlefield of the rural revitalization strategy (Wu et al. 2015;Zi et al. 2021).
In January 2017, the People's Government of Henan Province issued the 13th Five-Year Energy Development Plan of Henan Province, which put forward eight key tasks, including accelerating the development of non-fossil energy, strengthening rural energy construction and building a smart energy system. The supply-side structural reform in the energy sector has achieved remarkable results. The average annual concentration of PM2.5 and PM10 has dropped by more than 30%, and all environmental indicators have reached the best level in the past 5 years. The 14th Five-Year Plan points out that it is necessary to build a low-carbon and efficient rural energy support system in Henan Province, continue to promote the energy revolution, and actively develop new and renewable energy.
In addition, the strategic Plan for Rural Revitalization of Henan Province (2018-2022) specifies the key tasks of "promoting the rural energy revolution." It is mainly reflected in optimizing the rural energy supply structure and constructing a clean, low-carbon, safe, and efficient modern rural energy system.
According to previous research literature, STIRPAT model has been successfully applied to study the environmental pollution caused by various factors. Based on STIR-PAT model, the scenario of peak household carbon dioxide emissions at provincial level in China was simulated by Zhao et al. 2021Ghazali and Ali 2019 used the extended STIRPAT model to analyze the drivers of the downward trend of CO 2 emissions in developed countries. Using the extended STIRPAT model, Lin et al. 2017 explores the impact of urbanization and real economy development on CO 2 emissions in developing countries.
Moreover, some literatures also use STIRPAT model to explore the driving factors of energy consumption. Gani 2021 researched the impact of fossil fuel power generation on environmental quality, based on a STIRPAT model that incorporated economic, social and institutional factors. Using STIRPAT model, Shahbaz et al. 2017 re-investigated the relationship between urbanization and energy consumption in Pakistan during 1972Q1-2011Q4. And it confirmed that urbanization increases energy consumption. Employing the extended STIRPAT model, Wang et al. 2013 inspected the effects of population, economic level, technological level, urbanization level, industrialization level, service level, energy consumption structure, and foreign trade level on energy consumption in Guangdong from 1980 to 2010.
Whereas, there is no literature to assess the main drivers of rural energy in China applying STIRPAT model. Therefore, exploring the driving factors of rural energy consumption in Henan Province and predicting the trend of total energy consumption (TEC) and power consumption (PC) in Henan Province are crucial steps to explore the mode of rural energy transformation and upgrading in Henan Province.
For these reasons, the main tasks of this paper are to analyze the current situation of rural energy consumption in Henan Province, explore the driving factors affecting TEC and PC in rural areas of Henan Province, and forecast TEC and PC to analyze their changing trend. So as to explore the low-carbon energy transformation model of Henan Province. In addition, the theoretical framework for this article is shown in Fig. 1. In addition, the nomenclature in appendix Table 1.
The contributions and originalities of this paper can be summarized as follows: (1) In this paper, the decoupling states between rural PC and TEC and its influencing factors in Henan Province are analyzed by using decoupling principle creatively.
(2) Since there is multicollinearity among the 6 influencing factors of TEC, and the same problem exists among the 6 influencing factors of PC. Therefore, based on STIR-PAT model, this paper uses ridge regression method to fit the linear equations of TEC and PC originally, and the validity of the fitted equations are verified. Rather than simply summing up the primary energy consumption of rural Henan Province. (3) In addition, this paper optimizes the BP neural network by PSO, and innovatively predicts the values of influencing factors of PC and TEC. (4) Finally, the consumption of TEC and PC in 2020-2025 are obtained by STIRPAT using ridge regression. Also, the forecast results of TEC and PC are analyzed to find the mode of rural energy transformation in Henan Province.

Analysis of the current situation of rural energy in Henan Province
This section introduces the decoupling principle to analyze the influencing factors of rural energy consumption and electricity consumption in Henan Province.

Variable selection
The decoupling theory The word "decoupling" originated in physics. Decoupling means that environmental pressure and economic growth are no longer interdependent, and changes in environmental pressure indicators such as carbon emissions are no longer synchronized with changes in economic development, and do not affect each other (Gong et al. 2021). The Tapio decoupling model analyzes the decoupling relationship between variables through the concept of elasticity. Tapio decoupling model is not only not affected by the change of statistical dimension, but also can decompose the causal chain of the decoupling index by using the identity, so as to realize the in-depth analysis and research on the influencing factors behind the decoupling state (Duan et al. 2021). This paper studies the decoupling relationship between rural TEC and its influencing factors in Henan Province and the decoupling relationship between rural PC and its influencing factors. The expression is as follows: where, D is decoupling elasticity index; Q is rural TEC or EC; Y is the influencing factor. Consequently, the equation of energy influencing factors based on decoupling theory can be written as: where, Wi represents the ownership of various factors affecting rural TEC in Henan Province; ΔWi represents its increment; E and ΔE represent the rural TEC and its increment in Henan Province in the current year. Consequently, the equation of electricity influencing factors based on decoupling theory can be written as: where, Ni represents the ownership of various factors affecting rural PC in Henan Province; ΔNi represents its increment; C and ΔC represent the rural PC and its increment in Henan Province in the current year.
Tapio decoupling indicators take the positive and negative values of ΔQ and ΔY , as well as the elastic values of 0.8 and 1.2 as critical values, respectively, to judge the state and degree of decoupling. Eight decoupling states are defined according to the value of decoupling elasticity (Wenbo and Yan 2018), as shown in Table 1.

The variable selection
In this paper, EIA, PS, PAM, V, POP, and EE are taken as the influencing factors of TEC in Henan Province. Based on the data of TEC and influencing factors in rural areas of Henan Province from 2009 to 2019, the decoupling analysis is conducted based on Eq. (2). The results are shown in Table 2 and Fig. 2.
According to the Table 2, there is a negative decoupling of growth between EIA and total energy consumption TEC. The association between PAM and TEC is negative decoupling of growth and growing concatenate. Also, the relationship between PAM and TEC is negative decoupling of growth and growing concatenate. The relationship between V and TEC shows weak decoupling. POP and EE show strong negative decoupling relationship with TEC.
As shown in Fig. 2, EIA, PS, PAM, and V all show positive decoupling effects on TEC. Among them, the most influential factor is the EIA, which shows a trend of wave rise. In addition, the decoupling index of PS, PAM, and V have approximately the same trend. However, POP and EE have a negative decoupling effect on TEC, and the decoupling elasticity index of POP is large. In terms of PC, V, EEP, POP, PI, PS, and TP are taken as the influencing factors of PC in Henan Province. Based on the data of PC and influencing factors in rural areas of Henan Province from 2009 to 2019, the decoupling analysis is conducted based on Eq. (3). The results are shown in Table 3 and Fig. 3. Table 3 shows that the variables V, PI, and TP show weak decoupling with rural electricity consumption PC in Henan Province. In addition, the states of PS and EEP are negative decoupling with PC. However, there is a strong negative decoupling between POP and PC.
As shown in Fig. 3, the analysis of V, PI, PS, EEP, and TP from 2009 to 2019 shows a positive decoupling effect on PC. Among them, EEP has the most significant impact on PC, the second is the PS, and V, PI and TP have the same influence on PC.

Theoretical model construction and data
In this study, to fully investigate the characteristics of rural energy usage of Henan Province in total energy consumption (TEC) and power consumption (PC). Firstly, STIRPAT method is used to fit and forecast the total amount of rural energy consumption and electricity consumption in Henan Province, and the model is verified. The next, PSO-BP is involved to predict the change of the influencing factors of TEC and PC in Henan Province.

Theoretical model construction
The STIRPAT theory The STIRPAT model originated from IPAT (Impact by Population Affluence and Technology), first proposed by Ehrlich and Holdren (1971)  where, I is the impact, which is usually measured by the emission level of pollutants; P is the size of the population; A is the wealth of a country; T is the technical index. In order to fully study the factors affecting environmental change, IPAT model is too simple and has its limitations. Therefore, using this model as the basis, Dietz and Rosa proposed the STIRPAT model as follows (Huo et al. 2020;Lin and Li 2020): where, a is the intercept term; P, A, and T are the same as in Eq. (4); B, C and D represent the elasticity of environmental impact on P, A and T respectively. t is a random disturbance, and the subscript t indicates the year.
STIRPAT model has been widely used to analyze the factors affecting environmental pollution (Li et al. 2011;Wang et al. 2013).To eliminate possible heteroscedasticity, all variables are in logarithmic form. Thus, the equation can be rewritten as: where, P represents population size (10E4 persons); A is per capita GDP (yuan); T is a technical indicator, measured by energy efficiency (energy consumption in ISI/its actual output − ENE).
To further investigate the driving forces of TEC in rural areas of Henan Province. Based on the specific situation of Henan Province, total rural machinery power, effective irrigation area, energy intensity, rural population, total agricultural output value and per capita housing area of farmers are incorporated into STIRPAT model for improvement and expansion. The rewritten expression is: where, TEC t represents the rural energy consumption of Henan Province in t years (10E4▪ tons); PAM represents the total power of agricultural machinery; EIA stands for effective irrigation area; EE stands for energy intensity; POP stands for rural population; V represents the total value of agricultural output; PS represents per capita residential area; t is the year.
In addition, total agricultural output value, electric motor power, rural population, per capita income, per capita housing area of farmers and total power of rural durable goods are incorporated into STIRPAT model for improvement and expansion. The econometric model and equation of rural PC in Henan Province is established: where, P represents the rural PC of Henan Province (100 million kw▪h); V represents the total agricultural output value; EEP stands for electric motor power; POP stands for rural population; PI stands for per capita income; PS represents per capita housing area; TP represents the total power of rural durable goods.
The relevant variables in Eqs. (7) and (8) are defined in Table Nomenclature.

The BP theory
BP network (Back Propagation) was proposed by a group of scientists headed by Rumelhart andMcCelland in 1986 (Wen andYuan 2020a;Kim et al. 2020). BP is a kind of multilayer feed-forward network trained by error backpropagation algorithm, which is one of the most widely used neural network models. The learning process of BP includes the forward propagation of information and the back propagation of error. After repeated training, the network weight and deviation changes are calculated continuously in the (6) LnI t = a + b p t + c A t + d T t + t (7) LnPC t = Lna + 1 LnV t + 2 LnEEP t + 3 LnPOP t + 4 LnPI t + 5 LnPS t + 6 LnTP t + t Fig. 3 Decoupling elasticity indexes between PC and influencing factors direction of relative error function gradient descent, gradually approaching the target. A typical three-layer BP neural network structure is shown in Fig. 4 (Keshtkarbanaeemoghadam et al. 2018).

The particle swarm optimization (PSO) theory
Particle swarm optimization (PSO) was proposed by Dr. Eberhart and Dr. Kennedy in 1995 inspired by artificial life. It simulates the random foraging behavior of a flock of birds in space (Suman et al. 2021;Malik et al. 2021).
Assuming that none of the birds know exactly where the food is, but they do know roughly how far it is, the simplest and most effective method is to search the area around the bird that is currently closest to the food. Hence PSO treats each bird as a particle with position and velocity. Then, the particle uses the closest location it has found to the food and the closest location the group has found so far to change its direction of flight. Finally, the whole population is directed to the same place, and that place is the area closest to the food.
In the process of optimization, the velocity and position of each particle are updated according to the following formulas: where, d = 1, …,n; x T i and v T i represent the position and velocity of the ith particle in the T iteration respectively. c1 represents self-cognition coefficient; c2 represents social cognition coefficient; r1 and r2 are random numbers between [0,1]; Ṗl best represents the individual optimal position found by the ith particle until the T iteration; g best represents the global optimal position of the whole population until iteration T; x T+1 i,d and V T+1 i,d represent the position and velocity of the ith particle obtained after the T + 1 iteration of the particle d, respectively.

The PSO-BP model
Based on the introduction of BP and PSO above, this paper uses PSO to improve BP to predict the changes of various influencing factors (Wen and Yuan 2020b;He et al. 2020;Wu 2021). The specific steps are as follows: Step 1: The fitness function of PSO is set as shown in formula (11) where e is the error function in formula (12): Fig. 4 Schematic diagram of typical N-dimensional input neuron model. Where, i is from 1 to n. θ is the weight between the input layer and the hidden layer. And, Ψ is the weight between the hidden layer and the output layer where λ and μ are non-negative weights. Also, the expressions of MAE and RMSE are shown below: where, a i and â i are the actual value and predicted value respectively, and n is the number of samples.
Step 2: Initialize the position and velocity of each particle, initialize parameters c1,c2, maximum velocity Vmax, and minimum velocity Vmin, population size N, and maximum iteration Tmax.
Step 3: The fitness function of each particle in the population is calculated, and the initial individual optimum and the total optimum are obtained.
Step 5: If the iteration number is equal to Tmax, the algorithm ends. And output the optimal value and the optimal particle.
Step 8: The value of the optimal particle is taken as the weights and thresholds of BP.
Step 9: Finally gets a trained network of BP.
The flow chart of PSO-BP is shown in Fig. 5. In the figure, E is the error of backpropagation, and ε is the number that is infinitely close to 0.
where, O is the output vector of the output layer; d is the expected output vector; x is the expected input vector; l is the dimension of the output vector of the output layer; w is the connection weight between the output layer and the hidden layer; v is the connection weight between the input

Data
Data were mainly drawn from government statistical yearbooks. The energy consumption and power consumption of Henan Province are from China Energy Statistical Yearbook (National Bureau of Statistics, 2020). The data of influencing factors total power of agricultural machinery, effective irrigation area, energy intensity, rural population, per capita housing area, total value of agricultural output, electric motor power, per capita income, and total power of rural durable goods are from the Statistical Yearbook of Henan Province (Henan Bureau of Statistics, 2020).

Multicollinearity analysis
The multicollinearity test is performed based on the data of each influencing factor and rural TEC and rural PC from 2009 to 2019. The multicollinearity test results of various influencing factors of rural TEC are shown in Tables 4 and  5. The multicollinearity test results of various influencing factors of rural PC are shown in Tables 6 and 7. According to the results shown in Table 4, it is clear that more than half of the variables are correlated, among which the relationship values of total agricultural machinery power, energy efficiency, population, gross output value and housing area per capita are all high.
In addition, VIF is the most common collinearity evaluation standard. The VIF values of agricultural machinery power, population, total agricultural output value and housing area per capita, energy efficiency are all well above 10 in Table 5. That is, there is a serious multicollinearity between these variables.
In the same way, there is a serious multicollinearity between PC and its related variables.

Construction of regression equations based on STIRPAT
In this paper, ridge regression is used to estimate the coefficients in STIRPAT model for TEC and PC.
In the aspect of TEC, according to the ridge regression equation, Figs. 6 and 7 respectively show the relationship between ridge trace and R 2 and k. Each independent variable first changes rapidly with the increase of k value, and then becomes rapidly stable after k = 0.67. Figure 6 shows that R 2 of ridge regression changes at a high rate of change before K = 0.67, and the change rate becomes much lower when the value of k is 0.67. Thus, the minimum value of k (k = 0.67) can be performed with a high adjustment R 2 of 0.941. Therefore, it is reasonable to choose k as 0.67 in this paper.
F test can be passed and the result of F-Sig is 0.000 0049 < 0.05. This implies that there is a linear relationship between the independent variable TEC and the dependent variables. At the same time, the t-sig of constant term and regression coefficient is less than 0.1, in Table 8. This indicates that all independent variables should be introduced into the regression equation. Finally, the fitting ridge regression equation of LnTEC is as follows:  In the aspect of PC, ridge regression is used to estimate the coefficients in the STIRPAT model (Figs. 8 and 9), and k = 0.66. F test can be passed and the result of F-Sig is 0.0001732 < 0.05. This implies that there is a linear relationship between the independent variable TEC and the dependent variables. At the same time, the t-sig of constant term and regression coefficient is less than 0.1, in Table 9. This indicates that all independent variables should be introduced into the regression equation. Finally, the fitting ridge regression equation of LnPC is as follows:

Prediction results of influencing factors
In this paper, PSO-BP model is used to predict the factors affecting rural TEC and PC in Henan Province. In addition, the rolling method of the prediction of the first three years and the next year is adopted to obtain the predicted values of influencing factors affecting TEC and PC in Henan Province during 2020-2025, as shown in Tables 10 and 11.
From Table 10, it can be seen that PAM, V, and PS increase to varying degrees over time from 2020 to 2025, POP shows a decreasing trend, also EIA and EE have changed very little. This means that with the gradual increase of V and PS, the use of coal, oil, natural gas and other primary energy will inevitably increase in rural areas of Henan Province. Figure 10 shows the trends of the influencing factors of TEC from 2009 to 2025.   This means that the increase of EEP, PS and TP will directly lead to the increase of rural electricity consumption. The increase of V and PI improves the living standard of villagers and indirectly leads to the increase of PC. Figure 11 shows the trends of the influencing factors of PC from 2009 to 2025.

Prediction results of TEC and PC
In order to predict rural TEC and PC in Henan Province, the predicted values of influencing factors in 2020-2025 are put into Eqs. (17) and (18) respectively.
Thus, the predicted TEC and PC in rural areas of Henan Province from 2020 to 2025 can be calculated, as shown in Tables 12 and 13.
As shown in Table 12, from 2020 to 2025, the TEC in Henan Province keeps a steady growth, and the predicted values of TEC in 2020 and 2025 are 24.84 million tons and 26.02 million tons, respectively. The TEC in rural areas of Henan Province in 2025 is 4.72% higher than that in 2020.
As shown in Table 13, the total rural PC in Henan Province keeps a steady growth from 2020 to 2025, and the predicted values of PC in 2020 and 2025 are 39.078 billion kwh and 43.034 billion kwh, respectively. In 2025, rural PC in Henan Province will increase by 10.12% compared with 2020.
As can be seen from Fig. 12, both PC and TEC in Henan Province are increasing year by year. However, the growth range of TEC will decrease significantly from 2022, while that of PC will increase significantly, which means that the implementation of Henan Province's rural revitalization strategic plan has achieved significant results in the key task of "promoting rural energy revolution." Nonetheless, after converting the annual PC in rural areas of Henan Province into standard coal, it accounts for less than 30% of TEC. This shows that rural electrification in Henan Province is at a low level and relies more on fossil energy such as coal, oil and natural gas. According to the forecast values of TEC and PC in Henan Province from 2020  to 2025, in the future, rural TEC in Henan Province will still rely more on fossil energy without transformation. This will undermine the goal of carbon peaking by 2030.
Consequently, the realization of Henan Province's rural energy transformation needs the substitution of low-carbon energy and the improvement of electrification level.

Conclusions
Rural areas account for 12% of global carbon emissions. China, as a major agricultural country, is exploring a low-carbon transformation model of rural energy as a necessary path to achieve carbon neutrality by 2060. This paper takes rural areas of Henan Province as an example to explore the path suitable for the low-carbon transformation of rural energy in Henan Province. The research results of this paper can be summarized into the following aspects.
(1) In this paper, the decoupling states between rural PC and TEC and its influencing factors in Henan Province are analyzed by using decoupling principle. The results show that there are different degrees of decoupling between TEC and PC.
(2) In addition, this paper optimizes the BP neural network by PSO, and innovatively predicts the values of influ-  Which also means that each influencing factor is complementary to each other. (3) Finally, the consumption of TEC and PC in 2020-2025 is obtained by STIRPAT using ridge regression. According to the forecast results, the consumption of TEC and PC will continue to increase year by year in 2020-2025. However, from 2022, the growth rate of PC will be greater than that of TEC. This is a good beginning of the low-carbon energy transition in Henan Province. Nonetheless, rural TEC in Henan Province will still be more dependent on fossil fuels in 2020-2025.

Policy implications
These conclusions can provide the following applicable policy implications for rural energy upgrading of Henan Province and other rural areas in China. First, rural power grid upgrading should be accelerated. (1) Unified planning. The long-term goal of the unified construction of power grid, unified power grid equipment sequence, integrated distribution of power grid services, narrowing the gap between urban and rural power supply services, and promoting the coordinated development of urban and rural power grids. (2) Classified development. First, in light of different regions, strengthen power grids in cities and industrial clusters. Second, according to different voltage levels, the development of strong simplification and orderly, speed up the construction of 110 kv and 10 kv and below power grid, according to the principle of orderly advance and backward optimization of 35 kv power grid.
Second, promote renewable energy generation. According to the characteristics of rural regional distribution and resource endowment, different power supply should be adopted. Promoting the development of small hydropower, photovoltaic, wind power, methane, and other renewable energy sources in rural areas, implement the substitution of electric energy and multi-energy complementarity, improve energy efficiency, and eliminate prominent problems such as high pollution and low energy efficiency in rural development, in Henan Province under the "dual carbon" target.
Third, promote electric energy substitution projects. To improve the quality of agriculture and promote the electrification of agricultural production. First, centering on the goal of high-standard farmland in Henan, the government should promote comprehensive matching of farmland, water, forest, roads and electricity, and turn more "wang-tian farmland" into "high-yield farmland." Second, promoting the application of technologies such as electric flue-cured tobacco, electric tea production, electric grain drying, heat pump drying of wood, and cold storage of fruits and vegetables, so as to ensure power consumption of intelligent planting and breeding bases and enhance agricultural electrification.
Fourth, the establishment of national subsidy policies should be based on the level of agricultural development in the region. The upgrade of rural energy can improve the utilization efficiency of rural energy, promote the great development of rural economy, improve the income of rural areas and offset the high cost of clean energy. Therefore, on the one hand, attention should be paid to the collaborative promotion of TEC and PC, while improving the use of PC in rural areas, the use quality of TEC should be improved, so as to reduce CO 2 emissions in rural areas. On the other hand, agricultural mechanization can increase V and PI, thereby boosting the rural economy and thus offsetting the additional costs of renewable energy.

Appendix
The appendix