Project engineering management evaluation based on GABP neural network and artificial intelligence

BP neural network is the most representative computational and research method of current machine learning, BP neural network has gradually developed into the most widely used and now the most widely used in the industry with its powerful nonlinear feature function mapping computing capabilities, good data induction and feature recognition computing capabilities. This paper closely links the actual activities of project engineering technology management with modern machine learning network technology, and proposes a new type of engineering management activity model based on machine-integrated GABP neural network. Based on artificial intelligence technology, this paper discusses the theory of analytical neural learning in depth, discusses the theory of analytical neural learning in combination with genetic algorithms, and systematically discusses the models commonly used in this study based on theoretical foundations such as genetic algorithms. And points out the defects and defects of the neural network structure itself. Good generalization ability, in order to minimize the impact of subjective factors on the valuation results. Although there are many management software specializing in construction project engineering currently on the market, the service scope and demand range of these management software specializing in construction project engineering are relatively narrow. For example, construction project management software can only be used manage a construction project. In the actual operation process, it is also found that unreasonable time arrangements often occur in various links, such as personnel scheduling and material distribution of engineering projects. It will not only cause a large waste of material resources, but also affect the effect of labor. To this end, starting from the government’s desire to improve the efficiency of engineering management projects, this paper designs an engineering management system for general projects based on artificial intelligence technology and neural networks, and after a large number of practical analysis.


Introduction
In our real life, project engineering often has a large amount of investment, and it is difficult to coordinate the investment of other projects. The characteristics of these technologies make the business process of our project engineering management cumbersome and the amount of information generated by the project is huge (Sharon et al. 2011). Traditional engineering management software and extensive management models have become increasingly unable to meet our engineering construction and management needs. More and more enterprise project managers are beginning to need a more scientific and effective solution that is closer to social reality (Wasiak et al. 2011). The existing engineering project management software has been functionally supplemented and systematically optimized to fully meet the needs of enterprises (Lechler and Dvir 2010). Various information management requirements in the project management process, collect and process various project engineering data with more reasonable methods and methods, and provide enterprises and project engineering decision-makers with more effective information and auxiliary solutions to improve Management efficiency of enterprises and engineering projects  (Chou 2011). This article mainly studies the following content of design work: first, on the basis of extensive research and reference to relevant types of design experience at home and abroad, the specific design work methods and implementation ideas of the implementation of the key project management engineering system of this article are proposed (Spalek 2013). Next, when designing the system, this paper expands the application scope of the system, improves the accuracy of the system model, and combines the problems encountered in the experiment, the technical framework of this paper is elaborated in detail in order to achieve better experimental results (Hu et al. 2015). Third, detailed analysis and in-depth discussion of non-functional application requirements of different needs faced by relevant Chinese departments in the implementation of key project management engineering applications, mainly through detailed analysis of various forms, such as UML case diagrams and actual use demand case tables (Memon et al. 2010). Among the application requirements, fourth, this article will use various research results in actual applications as the theoretical basis, and carry out detailed functional planning and specific design for related functional requirements. In the process of the system design, it mainly introduces the overall functional software architecture, functional overall structure, characteristic design description, detailed structure design of main functions and overall design of the system database of the operating system (Gong et al. 2012). Adopting a three-tier B/S network technology management model, the main function of the entire management system is to realize a variety of business functions, such as system management, analysis and evaluation, tax-related key financial statements and other key project management (Bailing et al. 2016). The author of this article focuses on the management functions. The detailed design of different types of system management application functions and application timings. Finally, this article elaborates on the specific design and application of the management system functions, and shows the operation interfaces that need to be implemented simultaneously for various management functions.

Related work
The literature introduces the main background and significance of GABP neural network recognition technology and the development trend of actual research applications in related research fields (Feng 2019). It introduces the development trends of related research applications at home and abroad based on systems, such as LPR and their actual research development trends in related research fields. The literature focuses on the detection model based on the gray-scale integrated GABP model neural network (Wang 2021). First, it focuses on the edge location detection electronic algorithm in the gray-scale positioning image, the wavelet transform detection algorithm, the color image segmentation detection algorithm, and the grayscale object image in the texture. The feature analysis detection algorithm and the edge location detection algorithm based on modern mathematical image morphology, etc., determine that the key experiment of this article mainly uses the Roberts edge detection operator algorithm to directly detect the edge operator in the grayscale image (Yang and Xu 2011). The literature briefly introduces the software experimental system environment of this article, including the experimental system software platform, processor, installation package and memory, experimental software runner and operating system, the software platform needed for the experiment, and the system experiment in this article (Hundhausen et al. 2011).
Using software examples and videos, system personnel analyzed in detail the specific process and use method of the software in this paper for system experiments (Eghtedari and Mahravan 2021). According to the actual business conditions and work needs of the local taxation business departments at all levels in a certain city and the staff in related positions, the literature first designs and analyzes the business process of the key project engineering design system of the city's local taxation business departments at all levels, and then analyzes the management system business functions were designed and analyzed according to actual requirements (Gavin 2011). At the same time, according to the system function design requirements, the management system functions were divided into three functions: performance evaluation of business database, business statistical reports of tax-related departments, and project management of key projects in the city. The literature uses the company's relevant technology and the national key construction project management experience as the research basis, and integrates business requirements and analysis and research results to design the system in detail (Barrett et al. 2016). First, the system architecture, functional structure, etc. are designed. And then use various functional structure designs as the research basis, and use various types of designs and time periods to integrate with each other to carry out detailed designs for each sub-function in the system, and complete the operation in accordance with the specified time. The literature uses Java, Mysql and other technologies to conduct experiments on the core modules of the entire system. The test results show that the functions of the designed national key project management system are basically to meet the needs of current enterprises, and its characteristics are systematic The performance is good, the interface is simple and clear, and it is convenient for practical operation (Bony 2010). The literature first designed the software architecture of the system, and also designed the overall functional structure diagram and overall type diagram of the system. And researched and realized each main function. In terms of application database system design, a variety of E-R diagram file formats are used to directly carry out system design. And according to the actual application situation on site, three processes are introduced in detail, mainly the software application layer, the view management layer and the data center management layer.

GABP neural network
The basic classification method of the average filter application template is to simulate the filter data obtained after frequency filtering and radio frequency processing are performed on a pixel point filter, and a template used as the average frequency filter is selected. The template of the average frequency filter is determined by A template filtered by radio frequency and several adjacent filter points of the pixel to be filtered are combined, and the average frequency of the average frequency filter to the template is used to calculate the filter threshold of the original pixel as a filter substitute. The template structure of the mean filter is expressed in the form (1): Formula (2) shows the data matrix block diagram of this paper in detail. The average value of the central coordinate node is obtained, and the coefficient equation is further obtained.
The output of the median filter in the two-dimensional sliding template is: Gaussian filter is used to smoothly process various types of signals and is frequently used in science and experiments. This paper also uses Gaussian filter when processing the image of license plate characters.
Then its probability density is: The value of the standard expectation l of the Gaussian function determines the position of the Gaussian function in the image node of the distribution. When l = 0 and r 2 = 1, the image position of the Gaussian function is called a standard normal Gaussian distribution.
Various quasi-number combinations of Gaussian functions, the following will focus on the two fitting forms of one-dimensional Gaussian fitting function and two-dimensional Gaussian function as simpler fitting examples in the actual function application process, and used to measure the Gaussian function Perform fitting and calculation analysis: (1) One-dimensional Gaussian function There are two main methods for fitting expressions of various functions of one-dimensional Gaussian functions: Among them: An anomalous Gaussian integral calculation method of Gaussian function, that is, the use of error Gaussian function, any inverse and time anomalies on an inverted real line in the entire Gaussian space, Gaussian integral can be accurately calculated, and It is concluded that using such an integration method can obtain a Gaussian integral such as (7): The same can be obtained: In this case, Gauss is the probability density function of a normally distributed random variable, and the expected value l = b, namely: (2) Two-dimensional Gaussian function The function expression of the two-dimensional Gaussian function is shown in (10): The bilateral signal filter in bilateral filtering is mainly a nonlinear digital image edge signal processor, which is similar to the basic working principle of Gaussian filtering. One outstanding advantage is that its bilateral signal filtering makes it better than Gaussian filters. It is more efficient to realize the processing effect of edge signal filtering in digital images.
The two-point spatial Euclidean distance error in the bilateral filtering equation can be accurately defined because the Euclidean distance between the current point of the bilateral filtering of two different images and a certain place in the filter center is expressed as: The Gaussian function of the range is defined as: Savitzky-Golay filter is a special audio filter design method that uses least squares smoothing to perform polynomial or analog combination through filtering in a certain space or time domain. The biggest technical advantage of this kind of filter is reflected in their originality. In the process of noise processing, under various conditions of effectively filtering out the original noise, it can also effectively maintain the filtering form and time width of the original digital image coding signal. The least square smoothing of the signal is illustrated by Fig. 1.

Artificial intelligence
The neuron model consists of an external offset. According to whether they are positive or negative, they are used to increase or decrease the network output when the activation function is activated. Two formulas such as (12) and (13) can be used to express a neuron K: The function of the threshold is to affine the output of the linear combiner in the model: Bias is an external parameter of artificial neuron K. The same can be obtained by combining formula (13) and formula (14) as follows: That is the step function. The corresponding output at this time Commonly used nonlinear functions are S function and radial basis function as shown in Fig. 2: The graph of the sigmoid function is sigmoid, which is a strictly increasing function. The following two forms are commonly used: (1) Logistic function (2) Hyperbolic tangent function Such functions are smooth and asymptotic, and maintain monotonicity.
(3) Radial basis function Corresponding to different laws of the basic learning structure and processing model of neural networks, there are different laws in the comprehensive learning process of different neural networks, through these neural networks according to the comprehensive operation principle of the learning process rules to accurately control the different neurons The interconnection and identification weights of, realize the comprehensive learning of different neural networks.
The summary is due to the American nervous system psychologist Donal Hebb, based on various conditioned reflex learning mechanisms in neurophysiological theory, proposed a learning rule about the changes in the interconnection and strength of various neurons. Expressed in a form of geometric algebra as: It is also called the law of error correction learning. This rule depends on the feedback outside the output node to change the weight coefficient, that is, it can be derived by using the condition of the lowest square error between the expected output and the actual output. The adjustment formula for the connection weight is: 4 Design, implementation and testing of project engineering management evaluation system

System function requirement analysis
In the overall construction of a large city, the municipal government needs to directly assume responsibility for organizing the overall construction process supervision and operation management of the entire city. The economic construction of a large and medium-sized city is inseparable from all kinds of construction projects, and the management personnel of government departments also need to strengthen the management and operation supervision of these large-scale construction projects in a timely and effective manner. In fact, because there are many types of construction projects, there are also many development and construction units, so the staff needs a special construction project management system to manage each construction project and its related construction units and other personnel. For a large-scale construction institution, they have jurisdiction over many large-scale engineering projects. For the management personnel of a large-scale construction institution, they need to accurately grasp and use all the basic information for managing these engineering projects, and at the same time, they need to accurately grasp To control the specific work schedule of each large-scale engineering project that they need to be responsible for, then managers also need a complete largescale engineering project schedule management information system to scientifically and effectively manage the speed of these engineering projects. Among the actual construction projects, large-scale engineering projects have a large number of personnel, heavy tasks, complicated documents, and a large amount of various funds. The systems they need not only need to understand these people, funds, technologies, and basic information about the project. Documents, etc. are effectively managed, and reasonable personnel deployment plans, material delivery plans, and solutions for fire monitoring must also be provided. Starting from the actual business situation of Project engineering management evaluation based on GABP neural network and artificial… 6881 engineering projects, this paper summarizes the general and general needs of various engineering projects, and studies and formulates a set of systems suitable for the management of various engineering projects. This article divides the system into seven main functional modules, as shown in Fig. 3.

Design principles
When designing the project management system, the following design principles were followed. The scalability of the project management system is mainly manifested in the presentation layer, application service layer and back-end data layer.
(1) Application layer: in the application layer, the network client provides services for users. The corresponding specialized special service equipment is the network server. When the number of internet requests made by a user rises sharply, it is that we need to increase some web servers to cope with this sharp increase in customer requests. The biggest advantage of a network load balancing cluster is that even if a host in the cluster fails, the cluster can still provide users with uninterrupted services, and the (2) Back-end data layer: in the back-end data layer, it provides users with good data scalability in three forms: widening inward, extending upward, and widening outward. The inward extension reduces the hardware conditions required during the operation of the software, which means that it can provide a higher degree of scalability under the same hardware environment. Extending the system means that one or more computers need to be added to the system. With the increase in the number of system software and hardware, the system will be quickly deployed and run on the existing software needed by itself to meet the needs of customers who continue to expand outwards and expand inwards.
The fault tolerance of construction project management system is mainly embodied in the design of data fault tolerance and the design of software fault tolerance.
(1) Data fault tolerance design: the project management system uses message middleware technology to improve the data fault tolerance. The message dissemination middleware is to ensure the effective use of network bandwidth resources by improving the sliding window technology and other methods. It not only greatly improves the efficiency of information transmission. It also enhances the stability of the network. If a message is broken during transmission, the middleware in the message will be dropped from this transmission line and then connected to it, and then continue to publish.
(2) Software fault tolerance design: code error detection, code error recording, and code error recovery. The system records the error information in the Log file, and the system operator can directly view the log files of these errors.
Good system security is very important for an operating system. During the normal use, maintenance and system update of the entire system, it is necessary to ensure that the entire system has high system stability in time. The network security of the engineering project safety management system mainly includes changes to the current engineering user identification, access control, security risk audit, data security and confidentiality, session security management, real-time alarm of abnormal events, file backup and data recovery. The main aspect.
Realize user identification: the project management system can provide a system login supervision module for identification and verification to ensure user safety in the project.
Access control: the engineering project sets up different project user roles in the system, which can mainly be divided into managers, government personnel, senior leaders of institutions, project leaders and general project constructors. A user can only have one role. You can view the pages you own on the system and do this part of the operation.
Security audit: the project management system provides security audit services covering all users. The review event includes the retrieval event date, time, retrieval type, access IP, retrieval time stamp, event description and retrieval event result. The user's privacy information involved in the review items needs to be displayed in an encrypted manner.
Information and data confidentiality: MD5 encryption algorithm is used to ensure the security of business data. Confidentiality mainly includes the confidentiality of the file system and the storage of the database.
Session management: as the system already has an automatic interrupt function, when the user does not perform any operation on the system, the server will automatically terminate the current session. The Internet time system limits the maximum number of concurrent computer sessions that users can use.
From the perspective of system architecture, it is mainly composed of four parts: server, database, PC-side project management platform, and computer client, using B/S architecture. The web background module software uses an overlay background framework. Cache storage service files use Redis caching technology. The computer mobile terminal software project management system and all computer client interfaces are implemented in VUE mode, respectively.

Database design
In the design of the information management system of this large-scale engineering project, since each entity has more attributes, the attributes that each entity needs are described in the design of the database logic structure. The specific entity association graph structure of each database that the system needs to implement is shown in Fig. 4.
Through the logical structure analysis of the database information of the project management system of China's large-scale enterprises, the specific design of the field names, data types and total length of project data in each table is described by charts one by one: It is mainly used to display and store the basic information of each engineering project, including the project name, project year, etc. The specific table framework is shown in Table 1.
The personnel information list is mainly used to display and store information related to the project, including the name of the project-related personnel, related roles, and  Table 2.
As an information list of the construction unit, it is used to record and manage the relevant information of the construction unit, including the name of the unit, the contact person of the construction unit, etc. The specific list format structure is shown in Table 3.
As a progress record data file, its main function is to display and manage project information related to the storage progress. The specific structure format is shown in Table 4.
Project history table, which is used to store related information about a project history, including items related to the project history, personnel related to the history, etc. The specific list format is shown in Table 5.

Design and implementation of key modules
When the user clicks on a project list module, it will directly enter a main page about the project list. If it is the person in charge of a single project, he can only directly see the basic information of a project bundled with him. There are many interfaces included in this module, and the selection provides an in-depth explanation of the project list search interface, the item catalog entry interface, and the project statistics interface.
(1) Item list search interface: the search interface is jointly implemented by the search method and the busSearch method. The search method can be used by administrators to search, and the bussearch method can be used by grassroots users to search. The search query interface takes the basic information fields of the construction project, the construction project unit, the personnel, and the construction workflow obtained by the project receiver as the main parameters. The search function in the search interface is mainly realized through the process of storing the database. (2) Item catalog entry interface: this entry interface can not only realize the batch entry of separate different projects at the same time, but also realize the batch entry of multiple different projects at the same time. This input interface can receive the main input data category of a project, the input data source of the project, etc. as an input parameter. The type of project entry mainly includes the type of manual entry, the type of pictures and documents, the type of excel files, and the data source of the database.   (3) Project statistics interface: the data interface takes a project parameter D as the input value parameter, and returns the entire relevant industry statistical data information of the project. In the actual project management project, a construction project with a particularly large actual engineering volume, often someone will be forced to re-decompose it into several individual sub-construction projects, if redecomposed several sub-construction projects. If the actual project volume is not small, some people may be forced to re-decompose these new sub-projects. So when we count the information and data of each project, we may find a prominent problem, because the parent project actually does not have real information and data at all, and the information and data of the parent project itself should also be controlled by its children. Project information and data are accumulated. A tree structure can be formed between parent and child projects. The structure of this tree structure is shown in Fig. 5.
The main function of an intelligent module is to read the message of the sensor device, analyze the message data of the change monitor, and obtain a temperature change data that can be automatically displayed in real time in a new project. When the temperature of an object in a specific area exceeds a certain measured value, it is likely to cause a fire. In order to effectively avoid the recurrence of fire accidents, it is necessary to install temperature sensors in each specific area of the project at the same time. The device in temperature change sensing can record the temperature change of the object in any specific area in real time. When the temperature in a certain place is high, the project leader can easily view the landforms around the project through the map, so as to determine the route of its transfer. All monitored data are displayed in the form of a histogram. The intelligent monitoring module mainly provides several interfaces: (1) Read sensor automatic message interface: this automatic communication data interface can automatically communicate with the equipment in the sensor. After the communication is successful, it will read some messages and send these messages directly to the system database. In this timing interface, we can respectively set a message time point and data time period for reading data messages and saving data.
(2) Parsing message data interface: the parsing interface directly reads the content of the message from the database and parses it, and writes the result directly into the pm_fire table. The pm_fire table can store temperature and time value.
(3) Modeling interface: the main purpose of this interface is to establish a linear regression equation by using the temperature and time data that need to be read, and use the linear regression equation

Project engineering management evaluation system test
For a system to run stably and safely, a detailed test process is essential. In order to provide a reliable and safe system for the users and technicians of the entire project, the test mainly starts from the two dimensions of the function of the equipment and the performance of the system. Unit testing is to check and verify the smallest functional unit in the software. This test is mainly to test various function values of the source code in the form of writing test functions. In the test function, by calling the function in the source code, check whether the consistency of the realized function meets the expected function. The background software of this system is developed using SpringBoot framework, and the interface test is performed using SpringBoot integrated SwagerUi tool. The environment configuration formula of the test software is shown in Table 6.
The test hardware environment configuration is shown in Table 7 The test project practice proved that the system safety function test and key function design in the system project management have basically been able to achieve the

Conclusion
This article carries on the function and performance test to the project management system. In the whole process of technical research, it can be clearly seen that there are many shortcomings in itself: (1) in the intelligent fire monitoring system module of the engineering project, the linear temperature regression relationship model used is useful for predicting the fire temperature. Automatic prediction is carried out. There are many actual situations in which the predicted temperature coefficient T and the actual temperature t do not completely satisfy this linear regression relationship. Then we need to predict a lot, and there may be some mistakes due to this.
(2) In the enterprise material circulation and transportation management module, when calculating the minimum travel distance between two different projects, generally use Baidu search map to call the project place name of the project location to find a corresponding project location, so that The shortest travel distance can be obtained directly, but the location of the project site may still be inaccurate. In this case, the minimum distance of the project site is not a direct representation of the shortest travel distance between the two places in reality. Therefore, it becomes inaccurate when calculating the distance between two engineering projects.
(3) Estimate the working ability value of each technician in the dispatch system module of the professional and technical personnel. These may not be very accurate, and in the actual situation, if there is a large deviation in the working date of the technical personnel, the optimal The solution is not necessarily the only one, so we need to consider all the events that may occur in these situations as a whole, and it is really reasonable to choose the best solution for different situations.
Funding This paper was supported by Natural Science Basic Research Program of Shaanxi: research on the causes of public rejection in reclaimed water reusing projects based on EEG experiment (Program No.2022JM-429).
Data availability Data will be made available on request.

Declarations
Conflict of interest The authors declare that they have no conflict of interests.
Ethical approval This article does not contain any studies with human participants performed by any of the authors.