Power grid company comprehensive plan project evaluation method and system based on machine learning

Abstract

The invention provides a power grid company comprehensive plan project evaluation method and system based on machine learning, and relates to the technical field of data processing. Investment data and actual output benefit data of a historical power grid company comprehensive plan project are obtained from a power grid enterprise internal database, then an input-output analysis model is constructed through machine learning, and key influence factors between different input factors and output benefits are determined. And according to the influence degree of the key influence factors on the investment accuracy, determining the weight of the key influence factors, obtaining a correlation analysis result, and further constructing an investment accuracy evaluation model. And after related data of the comprehensive plan project of the target power grid company is obtained, inputting the related data into the investment accuracy evaluation model to obtain investment accuracy pre-evaluation information, and when an evaluation check request instruction of the display end is received, sending the pre-evaluation information to the display end for display. According to the method, the accuracy of power grid project investment can be improved, and powerful support is provided for investment decision making.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a method and system for evaluating a comprehensive planning project of a power grid company based on machine learning. Background Art

[0002] The comprehensive plan of the power grid is based on the comprehensive balance and overall optimization of the core resources and needs of the power grid company, and is the overall annual operation and development goal of the company. It is a systematic implementation plan for the full implementation of the company's strategy and planning. From a management perspective, comprehensive plan management is the management of the entire process of the company's project reserves, plan preparation, approval, issuance, execution, inspection, adjustment, and assessment. It is an important guarantee for the smooth implementation of the company's strategic goals and development plans.

[0003] Some existing power grid companies often make investment decisions based on past experience and simple cost-benefit analysis. Usually, they collect basic input and output data from past projects, and then professionals make a rough judgment on the feasibility and expected benefits of the project based on industry experience and conventional financial analysis methods. For example, they estimate whether a new project is worth investing in based on the investment amount range, construction period, and ultimately achieved power transmission growth of similar projects in the past.

[0004] However, the existing investment forecasting and evaluation methods are difficult to fully utilize a large amount of historical project data and difficult to establish quantitative analysis of project investment, which makes investment decisions somewhat blind. This leads to an increase in the deviation between power grid investment planning and actual benefits, and reduces investment accuracy. Summary of the invention

[0005] The present application provides a method and system for evaluating comprehensive planning projects of power grid companies based on machine learning, which is used to solve the problem that existing power grid companies have difficulty in making full use of a large amount of historical project data for quantitative analysis when making investment decisions, resulting in blind investment decisions, increased deviation between investment plans and actual benefits, and further reduced investment accuracy. The application can improve the accuracy of power grid investment, provide a more scientific and accurate basis for investment decisions, and improve the efficiency and quality of power grid investment.

[0006] In the first aspect, the present application provides a method for evaluating a comprehensive plan project of a power grid company based on machine learning, which is applied to an evaluation system. The method includes: obtaining multiple historical project input data and corresponding historical actual output benefit data of historical power grid company comprehensive plan projects through the internal database of the power grid enterprise; performing machine learning on the historical project input data and the historical actual output benefit data to construct an input-output analysis model, and the input-output analysis model is used to determine the key influencing factors between different input factors and output benefits; determining the influence weights of different key influencing factors according to the degree of influence of the key influencing factors on investment accuracy, and obtaining an association analysis result with the influence weights, and the investment accuracy The degree of influence is determined based on the degree of deviation between the actual output benefit data and the expected output benefit data; an investment accuracy evaluation model is constructed based on the key influencing factors and the results of the correlation analysis, and the investment accuracy evaluation model is used to determine the evaluation information of the key influencing factors on investment accuracy; after obtaining the project input data and the expected output benefit data of the target power grid company's comprehensive plan project to be carried out, the project input data and the expected output benefit data are input into the investment accuracy evaluation model to obtain the investment accuracy preliminary evaluation information of the target power grid company's comprehensive plan project; after receiving the evaluation view request instruction sent by the display end, the investment accuracy preliminary evaluation information is sent to the display end for display.

[0007] By adopting the above technical solution, the problem that the existing technology lacks full use of historical data for quantitative analysis in power grid investment decisions is effectively solved, thereby achieving the technical effect of improving the scientificity and accuracy of power grid investment precision evaluation, providing a more reliable basis for investment decisions, reducing blindness, and improving investment benefits.

[0008] In combination with some embodiments of the first aspect, in some embodiments, after the step of constructing an investment accuracy assessment model based on the key influencing factors and the correlation analysis results, it also includes: encrypting the calculation process of the investment accuracy assessment model; setting access rights for the investment accuracy assessment model, and the access rights are used to limit authorized personnel to operate the investment accuracy assessment model.

[0009] By adopting the above technical solution, the calculation process of the investment accuracy assessment model is encrypted and access rights are set. Encryption processing can prevent the model's calculation logic and key data from being leaked, ensuring the security of the model; setting access rights ensures that only authorized personnel can operate the model, preventing the model from being misoperated or maliciously tampered with, thereby ensuring the reliability and accuracy of the assessment results.

[0010] In combination with some embodiments of the first aspect, in some embodiments, after obtaining the project input data and the expected output benefit data of the target power grid company's comprehensive plan project to be carried out, the project input data and the expected output benefit data are input into the investment accuracy evaluation model to obtain the investment accuracy pre-evaluation information of the target power grid company's comprehensive plan project. It also includes: comparing and analyzing the investment accuracy pre-evaluation information with the actual investment accuracy data; calculating the error rate between the investment accuracy pre-evaluation information and the actual investment accuracy data; if the error rate is greater than a preset threshold, adjusting the investment accuracy evaluation model until the error rate is less than or equal to the preset threshold.

[0011] By adopting the above technical solutions, the model can be continuously optimized to make it more in line with the actual situation, improve the accuracy of investment precision pre-assessment, and thus provide a more reliable reference for investment decisions and reduce investment risks.

[0012] In combination with some embodiments of the first aspect, in some embodiments, after obtaining the project input data and the expected output benefit data of the target power grid company's comprehensive plan project to be carried out, the project input data and the expected output benefit data are input into the investment accuracy evaluation model to obtain the investment accuracy preliminary evaluation information of the target power grid company's comprehensive plan project, and it also includes: collecting historical power grid company comprehensive plan project data under different investment environments, and constructing a learning model according to different investment environments, the learning model is used to determine the weight adjustment parameters corresponding to the impact weight according to different investment environments, and the investment environment includes the economic environment and the policy environment; extracting the existing data features of the target power grid company's comprehensive plan project, and inputting the existing data features into the learning model to obtain the corresponding investment environment type; determining the corresponding weight dynamic adjustment parameters according to the investment environment type; and dynamically adjusting the impact weight of the key influencing factor in the investment accuracy evaluation model according to the weight dynamic adjustment parameters.

[0013] By adopting the above technical solutions, considering investment environment factors and dynamically adjusting weights, the investment accuracy evaluation model can be made more adaptable to actual conditions, improve the accuracy and flexibility of the evaluation, and provide more accurate guidance for power grid investment in different environments.

[0014] In combination with some embodiments of the first aspect, in some embodiments, after receiving the evaluation view request instruction sent by the display end, after the step of sending the investment accuracy pre-evaluation information to the display end for display, it also includes: determining the investment strategy based on the investment accuracy pre-evaluation information and a preset investment strategy library mapping relationship library; after receiving the investment strategy view instruction sent by the display end, sending the investment strategy to the display end.

[0015] By adopting the above technical solutions, specific strategic support is provided for investment decisions, enabling decision makers to quickly formulate reasonable investment strategies based on the evaluation results, improve the efficiency and scientific nature of investment decisions, and enhance the overall benefits of power grid investment.

[0016] In combination with some embodiments of the first aspect, in some embodiments, after receiving the evaluation view request instruction sent by the display end, the step of sending the investment accuracy pre-evaluation information to the display end for display specifically includes: after receiving the evaluation view request instruction sent by the display end, extracting key display data in the investment accuracy pre-evaluation information; based on a pre-set natural language template, filling the extracted key display data into a preset position in the natural language template, and generating a preliminary text description; performing semantic optimization on the preliminary text description to obtain a text interpretation report; and sending the text interpretation report to the display end for display.

[0017] By adopting the above technical solution, the evaluation information is presented in the form of a text report, which is more intuitive and easy to understand, making it easier for decision makers to quickly understand the evaluation results, reduce understanding costs, and improve decision-making efficiency.

[0018] In combination with some embodiments of the first aspect, in some embodiments, before the step of performing machine learning on the historical project input data and the historical actual output benefit data to construct an input-output analysis model, it also includes: obtaining project attribute information of the target power grid company's comprehensive plan project; determining the project category based on the project attribute information; based on the project category, selecting historical project sample data of the corresponding category as input data of the input-output analysis model to obtain the category-customized input-output analysis model.

[0019] By adopting the above technical solutions, more targeted models can be built for different types of projects, improving the accuracy and applicability of the models, thereby better serving the investment accuracy assessment.

[0020] In a second aspect, the present application provides an evaluation system, comprising: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code includes computer instructions, and the one or more processors call the computer instructions to enable the evaluation system to perform the method described in the first aspect and any possible implementation of the first aspect.

[0021] In a third aspect, the present application provides a computer-readable storage medium comprising instructions, which, when executed on an evaluation system, causes the evaluation system to execute the method described in the first aspect and any possible implementation of the first aspect.

[0022] In a fourth aspect, the present application provides a computer program product. When the computer program product is run on an evaluation system, the evaluation system executes the method described in the first aspect and any possible implementation manner of the first aspect.

[0023] One or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages: 1. Due to the use of machine learning to construct an input-output analysis model, determine key influencing factors and influence weights, and construct an investment accuracy evaluation model, it effectively solves the technical problem that it is difficult to make full use of a large amount of historical project data for quantitative analysis in existing technologies, resulting in blind investment decisions. It then realizes the quantitative evaluation of the accuracy of power grid investment, provides a scientific basis for investment decisions, and improves the technical effect of investment efficiency.

[0024] 2. Since the learning model is constructed by collecting historical power grid company comprehensive plan project data under different investment environments, and the investment environment type is determined according to the existing data characteristics of the target power grid company's comprehensive plan project to dynamically adjust the impact weight, the technical problem that the investment environment factors are not considered in the existing technology and the investment accuracy evaluation model lacks adaptability is effectively solved, thereby achieving the technical effect of improving the accuracy and flexibility of investment accuracy evaluation and providing more accurate guidance for power grid investment under different environments.

[0025] 3. After receiving the evaluation view request instruction, the technical means of extracting key display data, filling in natural language templates, semantically optimizing the investment accuracy pre-evaluation information to generate a text interpretation report and sending it for display is adopted. Therefore, the technical problem that the evaluation information is presented in a single form and is difficult to understand in the existing technology is effectively solved, and the evaluation information is presented in the form of an intuitive and easy-to-understand text report, which facilitates decision makers to quickly understand the evaluation results and improves the technical effect of decision-making efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 It is a flow chart of a method for evaluating a comprehensive plan project of a power grid company based on machine learning in an embodiment of the present application; Figure 2 is another flow chart of the method for evaluating a comprehensive plan project of a power grid company based on machine learning in an embodiment of the present application; Figure 3 It is a schematic diagram of a physical device structure of the evaluation system in the embodiment of the present application. DETAILED DESCRIPTION

[0027] The terms used in the following embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to be used as limitations to the present application. As used in the specification and appended claims of the present application, the singular expressions "one", "a kind of", "said", "above", "the" and "this" are intended to also include plural expressions, unless there is a clear indication to the contrary in the context. It should also be understood that the term "and / or" used in the present application refers to and includes any or all possible combinations of one or more listed items.

[0028] In the following, the terms "first" and "second" are used for descriptive purposes only and are not to be understood as suggesting or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features, and in the description of the embodiments of the present application, unless otherwise specified, "plurality" means two or more.

[0029] For ease of understanding, the following is a description of the process of the method provided by this implementation. Figure 1 , which is a flow chart of a method for evaluating a comprehensive planning project of a power grid company based on machine learning in an embodiment of the present application.

[0030] S101. Obtain multiple historical project input data and corresponding historical actual output benefit data of historical power grid company comprehensive planning projects through the internal database of the power grid enterprise; In this step, the evaluation system first needs to establish a connection with the internal database of the power grid company. This connection process involves the application of network communication protocols to ensure that the evaluation system can access the database safely and stably. The evaluation system will send a request instruction to request relevant data of the historical power grid company's comprehensive plan projects. For historical project investment data, it covers multiple aspects of information. Including but not limited to the capital investment of the project, which is further subdivided into equipment procurement costs, engineering construction costs, labor costs, etc. In terms of equipment procurement costs, the evaluation system needs to obtain the purchase price, quantity and supplier information of different types of equipment, so as to fully understand the specific investment of funds in equipment. Engineering construction costs involve various expenses in the construction process, such as building materials costs, construction personnel salaries, machinery and equipment rental costs, etc. Human costs are not only the wages of construction personnel, but also include the salary expenditures of project managers, technicians, etc., as well as related training costs.

[0031] In addition to capital investment, there may also be resource input data, such as the use of land resources and the amount of raw materials consumed. The use of land resources includes the land area occupied by the project construction, the nature of land use, and the cost of land leasing or purchasing. The amount of raw materials consumed needs to record the use of different types of raw materials, as well as their procurement costs and transportation costs. When obtaining historical actual output benefit data, the evaluation system should consider indicators in multiple dimensions. Electricity transmission growth is an important indicator. The evaluation system needs to extract electricity transmission data before and after the project implementation from the database and conduct comparative analysis to determine the actual electricity growth. This may involve the collection and collation of electricity data in different time periods and different regions. In addition, it also includes the improvement of power supply reliability. The evaluation system will obtain relevant data such as the number of power outages and power outage time, and evaluate the impact of the project on power supply reliability through statistical analysis. At the same time, economic benefits are also aspects that cannot be ignored, such as the additional income and cost savings brought by the project. The additional income may come from the increase in electricity sales and related service charges; cost savings may be reflected in the reduction of equipment maintenance costs and energy consumption. In order to improve the efficiency and flexibility of data acquisition, the evaluation system can adopt distributed data storage and computing technology. By distributing and storing data on multiple nodes and using a distributed computing framework for parallel processing, the speed of data acquisition and processing can be greatly accelerated.

[0032] In some embodiments, before this step, the project attribute information of the comprehensive plan project of the target power grid company is obtained from the internal database of the power grid enterprise. The project attribute information includes: project scale, project type, stage, location, etc. According to the project attribute information, the category of the target project is determined by matching the project attribute rule library. The project attribute rule library will clearly define the categories of projects corresponding to different attribute combinations, such as 500kV ultra-high voltage transmission projects, urban distribution network transformation projects, etc. According to the matched project category, sample data of matching categories are selected from the historical project sample data. For example, if the target project is determined to be a 500kV ultra-high voltage transmission type, samples marked as this category are selected from the historical project data, and the historical project sample data corresponding to the category is used as the training input of the input-output analysis model, and the category-customized input-output analysis model is obtained through machine learning training. Compared with the overall model, this customized model can more accurately characterize the characteristics of the category project. When the data of the comprehensive plan project of the target power grid company is obtained, the first step is to extract its project attribute information and determine the category. Then, the data is input into the corresponding customized model for input-output benefit analysis to obtain more accurate evaluation results. In the future, the sample data of different project categories can also be continuously expanded to continuously optimize the corresponding customized model so that it can more accurately evaluate new projects belonging to this category. When users want to know the investment accuracy of a specific project category in a comprehensive plan project of a power grid company, they can customize the input-output analysis model for different categories to obtain a more targeted evaluation effect of the input-output analysis model.

[0033] S102, performing machine learning on the historical project input data and the historical actual output benefit data to construct an input-output analysis model, where the input-output analysis model is used to determine key influencing factors between different input factors and output benefits; The evaluation system will use a large amount of historical project input data and the corresponding historical actual output benefit data to build an input-output analysis model. First, the evaluation system needs to preprocess the data, including data cleaning, data standardization, and data normalization. Data cleaning is to remove noise and erroneous data in the data. For example, some obviously unreasonable input or output data are checked and corrected. For example, if it is found that the equipment procurement cost of a project is too high or too low, exceeding the normal range, the evaluation system will verify and adjust the data by comparing with historical data and referring to industry standards. For missing data, the evaluation system can use mean filling, median filling or model-based prediction methods to supplement it to ensure the integrity of the data. Data standardization is to convert data of different dimensions into values ​​with the same scale. For example, for capital input data and resource input data, their units and magnitudes may be different. Through standardization, they can be made comparable in subsequent model training.

[0034] After completing data preprocessing, it is necessary to select a suitable machine learning algorithm to build an input-output analysis model. Common machine learning algorithms include support vector machines and neural networks. Taking neural networks as an example, the process of building an input-output analysis model is as follows: First, determine the structure of the neural network, including the number of nodes in the input layer, hidden layer, and output layer. The number of nodes in the input layer corresponds to the number of features of the historical project input data, and the number of nodes in the output layer is usually 1, representing the output benefit data. The number of nodes in the hidden layer can be determined based on experience or through experiments. Then, initialize the parameters of the neural network, including weights and biases. A random initialization method can be used to assign an initial value to each parameter. Next, use the training data to train the neural network. During the training process, the output of the neural network is calculated through forward propagation, and compared with the actual output benefit data to calculate the loss function. Common loss functions include mean square error, cross entropy, etc. Then, the gradient of the loss function to each parameter is calculated through the back propagation algorithm, and the parameters are updated according to the gradient descent method. Repeat the above process until the loss function converges or the preset number of training times is reached.

[0035] After the training is completed, the model needs to be evaluated and verified. The performance of the model can be evaluated using test data. Common evaluation indicators include mean square error, mean absolute error, coefficient of determination, etc. If the performance of the model does not meet the requirements, the model can be adjusted and optimized, such as adjusting the structure of the neural network, increasing training data, adjusting the learning rate, etc. Through the above steps, an input-output analysis model is finally constructed. The model can determine the key influencing factors between different input factors and output benefits, providing a basis for the subsequent construction of an investment accuracy evaluation model.

[0036] S103. Determine the influence weights of different key influencing factors according to the influence degree of the key influencing factors on investment accuracy, and obtain the correlation analysis results with the influence weights. The influence degree of investment accuracy is determined according to the deviation degree between the actual output benefit data and the expected output benefit data. Investment accuracy is obtained through in-depth analysis of the key influencing factors of power grid investment projects, and the use of machine learning to predict and evaluate the degree of match between the actual output benefits of investment projects and the expected output benefits. It is an important indicator to measure the scientificity and accuracy of power grid investment decisions, which helps to reduce investment risks and improve investment benefits. In this step, the evaluation system first needs to calculate the degree of deviation between the actual output benefit data and the expected output benefit data corresponding to each key influencing factor. A variety of methods can be used to measure this degree of deviation, such as mean square error, mean absolute error, etc. Taking the mean square error as an example, for each key influencing factor, the evaluation system calculates the square of the difference between the actual output benefit data and the expected output benefit data, and then calculates the average. Assume that for key influencing factor A, there are data from n historical projects, and the actual output benefit data is , the expected output benefit data is , then the mean square error . By calculating the mean square error of each key influencing factor, the evaluation system can obtain an indicator that reflects the degree of deviation. The greater the degree of deviation, the greater the impact of the key influencing factor on investment accuracy. Next, the evaluation system needs to determine the impact weight of each key influencing factor based on these degrees of deviation. A common method is to use normalization to convert the degree of deviation of each key influencing factor into a relative weight. Specifically, the evaluation system first calculates the sum of the degrees of deviation of all key influencing factors, S, , where m is the number of key influencing factors, then, for each key influencing factor j, its influence weight It can be calculated by the following formula: ,In this way, the evaluation system can obtain the impact weight of each key ,influencing factor, and thus obtain the correlation analysis results with ,impact weights.

[0037] In addition to the above methods, other technical means can also be considered to determine the impact weights. For example, the analytic hierarchy process can be used to construct a hierarchical model, with investment accuracy as the target layer and key influencing factors as the criterion layer. Then, the relative importance of each factor is determined by pairwise comparison, and the impact weight is calculated. This method can comprehensively consider the relationship between multiple factors and more comprehensively evaluate the impact of each factor on investment accuracy. In addition, machine learning algorithms can be introduced to automatically determine the impact weights. For example, the random forest algorithm can be used, with key influencing factors as input features and the degree of deviation between actual output benefit data and expected output benefit data as the target variable. By training the random forest model, the importance score of each key influencing factor is obtained, and its impact weight is determined. This method can use the powerful learning ability of machine learning algorithms to automatically mine potential relationships in data and improve the accuracy of weight determination.

[0038] S104: constructing an investment accuracy evaluation model according to the key influencing factors and the correlation analysis results, wherein the investment accuracy evaluation model is used to determine evaluation information of the key influencing factors on investment accuracy; The core goal of the investment accuracy assessment model is to quantitatively assess the impact of different key influencing factors on the accuracy of power grid investment, so as to provide a scientific basis for investment decisions. The specific operation process is as follows: First, the model receives key influencing factors as input data. These key influencing factors are accurately determined in the previous steps through in-depth machine learning analysis of the input data and actual output benefit data of a large number of historical power grid companies' comprehensive planning projects. Then, the investment accuracy assessment model processes and analyzes these key influencing factors based on its specific algorithms and structures. Different types of models use different algorithms. For example, the linear regression model will try to find the linear relationship between key influencing factors and investment accuracy; the decision tree model will gradually divide and judge the key influencing factors to construct a decision tree to determine the investment accuracy under different combinations of factors; the neural network model can automatically learn and mine the deep, nonlinear relationship between key influencing factors and investment accuracy through the complex connection and nonlinear operation of a large number of neurons. In the process of the model processing key influencing factors, the interaction and correlation between various factors will also be fully considered. For example, the scale of capital investment and the construction period may affect each other and jointly affect the investment accuracy. The model will accurately capture these relationships through complex calculations and analysis, and incorporate them into the evaluation process. Then, based on the results of processing and analysis, the model outputs evaluation information about the accuracy of the investment. This evaluation information can be a specific numerical indicator, such as return on investment, net present value, internal rate of return, etc., which can intuitively reflect the economic benefits of the investment project; it can also be a comprehensive evaluation score used to compare and rank the accuracy of different investment projects; or it can be a classification result, such as dividing investment projects into different categories such as high accuracy, medium accuracy and low accuracy, so that decision makers can quickly understand the general situation of the investment project.

[0039] In addition, in order to ensure the accuracy and reliability of the evaluation information, a variety of technical means are usually used in the process of model construction and training. For example, the generalization ability of the model is evaluated through cross-validation to avoid overfitting; feature selection technology is used to screen out the key factors that have the most significant impact on investment accuracy and reduce the interference of irrelevant factors; the parameters of the model are continuously optimized to improve the performance and accuracy of the model. Through the investment accuracy evaluation model, comprehensive, accurate and quantitative evaluation information can be provided for power grid investment decisions, helping decision makers to formulate investment strategies more scientifically and reasonably, reduce investment risks, and improve the accuracy and benefits of power grid investment, so as to better achieve the strategic goals and development plans of power grid enterprises.

[0040] S105. After obtaining the project input data and the expected output benefit data of the target power grid company's comprehensive plan project to be carried out, input the project input data and the expected output benefit data into the investment accuracy evaluation model to obtain the investment accuracy pre-evaluation information of the target power grid company's comprehensive plan project; In this step, the evaluation system needs to obtain the project input data and expected output benefit data of the comprehensive plan project of the target power grid company as the input of the model. The specific technical means are as follows: 1) Connect to the planning database to query the relevant data of the comprehensive plan project of the target power grid company. Specifically, the evaluation system needs to connect to the comprehensive plan management database within the power grid enterprise, which stores the comprehensive plan project data of various power grid companies throughout the company. The evaluation system can obtain the relevant data of the target project in a programmatic way by configuring the database connection parameters and writing SQL query statements. The query content includes: the basic information table of the project, which records the basic information such as the project name, planned investment amount, and project establishment background; the project input parameter table, which records the various types of capital investment, equipment procurement and resource planning data of the project in detail; the project output forecast table, which gives the output benefit forecast data of the project after professional analysis. After obtaining all the original data, the evaluation system needs to clean, filter and check the data to ensure the data quality. 2) Connect to the sales forecast system to obtain the expected output benefit data after the project goes online, such as power growth forecast and revenue increase forecast. The evaluation system needs to be connected to the company's sales forecasting system. The system collects and predicts information such as power consumption, electricity price adjustments, and electricity market planning of power grid companies, and can provide forecast data support for the expected output benefits of the project. The evaluation system can obtain the forecast of electricity growth and electricity fee revenue growth in the next 3-5 years in the project area by calling the open application programming interface of the sales forecasting system. 3) Clean and format the acquired data to ensure data quality and convert it into a format that the model can recognize. Such as processing missing values ​​and outliers, and performing regularization processing. The evaluation system needs to pre-process the raw data obtained from the database and forecasting system, including: checking and filtering out records with too many missing values; detecting and correcting data anomalies, such as negative values ​​and maximum values; performing format verification and conversion to ensure that the data conforms to the format required by the model input; performing regularization and standardization processing to improve data consistency. The processed data needs to be manually reviewed to ensure that the data is accurate and reliable. 4) Input the processed project input data and expected output benefit data into the investment accuracy evaluation model for prediction to obtain the investment accuracy pre-evaluation results of the project, including investment risk rating, probability of expected benefits, and other information. The evaluation system needs to load the pre-built evaluation model, then use the processed project data as the model input, execute the model prediction function, and obtain the evaluation results output by the model. The results will give the investment risk level of the project, such as A-level and B-level risks, and will also predict the probability of the project achieving the expected benefits, as well as other important evaluation indicators. 5) Check the pre-evaluation results, such as whether the results are within a reasonable range and whether there are obvious anomalies. Perform manual verification when necessary to ensure the accuracy of the results. After obtaining the pre-evaluation results output by the model, the evaluation system needs to verify the results to avoid obvious unreasonable situations.For example, you can configure a reasonable range of investment risk levels based on the project type and scale to determine whether the result is within the range. If there is an abnormality, you can send abnormal information to the display end to prompt the manual review process and conduct a second evaluation. 6) Store the investment accuracy pre-evaluation results and related intermediate data for model optimization and indicator assessment. The evaluation system will store the final results of the pre-evaluation in the result database, and will also store the data of the intermediate process, such as the original project data, pre-processing results, etc. These data can be used to continuously improve the training of the model and conduct backtracking audits.

[0041] In some embodiments, actual investment accuracy data of the target power grid company's comprehensive plan project that has been actually completed are collected. These data are derived from various output benefit indicators obtained after the project is actually built and operated. These actual output benefit data are compared and analyzed with the expected output benefit data generated by the initial investment accuracy pre-assessment. The actual differences between the two in terms of power growth, power supply reliability, economic benefits, etc. are analyzed. The error rate between the pre-assessment information and the actual data is calculated. For example, taking power growth as an example, if the pre-assessment predicts that the power growth is 1 million kWh, and the actual growth is only 800,000 kWh, the error rate is (pre-assessment value-actual value) / pre-assessment value, that is, (100-80) / 100=20%. The error rates of all investment accuracy-related indicators are calculated. If there are one or more error rates greater than a preset threshold (such as more than 10%), it is determined that the model needs to be adjusted and optimized. Project feature data and actual output data are extracted from projects with large errors and added to the training sample data set of the model. Use the new training sample set to retrain the investment accuracy assessment model to improve the fit of such projects and reduce the pre-assessment error. Repeat the comparison between the project pre-assessment and actual data until the retrained model keeps the error rate of all projects within the threshold. Continue to track the actual benefit data of newly completed projects to determine whether new samples need to be added to further optimize the model, achieve continuous optimization and iteration of the model, and ensure the accuracy of the forecast assessment.

[0042] S106. After receiving the evaluation viewing request instruction sent by the display terminal, the investment accuracy pre-evaluation information is sent to the display terminal for display.

[0043] In this step, the evaluation system needs to send the investment accuracy pre-evaluation results to the display end for display, as follows: An encrypted network communication connection needs to be established between the evaluation system and the display end in advance to ensure data security. Receive the evaluation viewing request sent by the display end, authenticate and authorize the request, and ensure that only authorized users can view the results. When requesting to view the evaluation results, the display end needs to bring the user's identity information. The evaluation system needs to verify the user's identity and confirm whether he has the access right to view the results. Only authorized users can view the results normally, and unauthorized requests need to be rejected. Extract the corresponding evaluation result data according to the request and keep it confidential. For example, use obfuscation for key parameters, replace sensitive actual values, etc. The evaluation system will parse the evaluation results that the user needs to view according to the request, and then keep the result data confidential to avoid leaking sensitive information. The key parameter values ​​can be transformed using an obfuscation algorithm, or they can be replaced with symbolic representation or classification intervals for display. The processed evaluation results are assembled into a format that meets display requirements. For example, to generate a report describing the project evaluation results, the evaluation system will assemble the confidential result data into the format required by the user. The evaluation system will immediately transmit the assembled evaluation result report to the display end through push technology. The display end can be updated in real time, greatly improving the interactive experience.

[0044] In the embodiment of the present application, since the input data and actual output benefit data of the historical power grid company's comprehensive plan projects are obtained from the internal database of the power grid enterprise, an input-output analysis model is constructed through machine learning to determine the key influencing factors, and the weights are determined based on the key influencing factors to construct an investment accuracy evaluation model, and then the target project data is input into the model to obtain pre-evaluation information and send it for display as needed. This effectively solves the problem that the existing power grid enterprise investment decisions are difficult to make full use of a large amount of historical project data for quantitative analysis, resulting in greater blindness in investment decisions, increased deviation between investment plans and actual benefits, and reduced investment accuracy, thereby achieving the goal of improving the investment accuracy of power grid projects.

[0045] In some embodiments, the evaluation system can realize the construction of an investment strategy library, which stores investment strategy sets corresponding to different investment accuracy levels. The investment strategy includes the recommended investment scale, investment progress control, risk control measures, etc. Then establish a mapping relationship between the investment strategy library and the investment accuracy evaluation results. According to the investment accuracy evaluation results, the corresponding investment strategy can be uniquely determined. If the evaluation result is a grade A investment risk, then match investment strategy 1. When the display terminal requests to view the investment evaluation results, the evaluation system will determine the only investment strategy based on the evaluation results and record the corresponding relationship. When the display terminal further sends an investment strategy viewing request, the evaluation system extracts the corresponding strategy content from the strategy library according to the pre-recorded corresponding relationship. The investment strategy content is processed and organized to generate a strategy report for intuitive display. The key points of the strategy are emphasized in the report, and the investment strategy report is sent to the display terminal, and the user can browse the investment risk evaluation corresponding to the recommended investment strategy. When the investment accuracy pre-evaluation results are sent to the display terminal, the evaluation system will extract key data in the results in advance, such as investment risk level, benefit compliance, etc. According to the preset natural language template, key data is inserted into the variable position of the template to generate a fluent language description, i.e., a preliminary text interpretation report. The report is optimized using a natural language generation algorithm to process text syntax and semantics to enhance readability. Finally, a natural language interpretation report that can intuitively reflect the results of the investment accuracy assessment is obtained and sent to the display end to assist users in quick understanding and decision-making.

[0046] In some embodiments, the usage habits of users of different professions can also be obtained in the form of electronic questionnaires, and the specific information needs and display preferences corresponding to each profession can be summarized. For example, managers pay more attention to reports and overall trends, while technicians need detailed data and indicator comparisons. Build a profession-feature model to represent the characteristics of different professions. The investment evaluation result report can be designed in a modular manner, and the key modules include a summary module, a trend module, a detail module, etc. The evaluation system can control the display end to display the occupation options. When the user wants to query the investment strategy or the pre-assessment information of the investment accuracy, he can select his own occupation in advance on the display end connected to the communication of the evaluation system. According to the user's occupational characteristics, the module combination and module content are dynamically selected to generate a personalized result template. The occupation uses specific industry vocabulary, and a vocabulary mapping table can be constructed to convert professional vocabulary into more easy-to-understand expressions before display to assist understanding. Different professions have different information interaction preferences, such as managers focus on reports, and technicians focus on visualization and interaction. The interactive format can be customized for the profession, and the feedback from users of different professions can be collected to identify their display preferences and needs, which can be used to continuously optimize the occupational feature model and continuously improve the result presentation.

[0047] After combining the above content, the following is a more detailed description of the process of the method provided by this implementation. Figure 2 , is another flow chart of the comprehensive planning project evaluation method of a power grid company based on machine learning in an embodiment of the present application.

[0048] S201, collecting historical power grid company comprehensive plan project data under different investment environments, and constructing a learning model according to different investment environments, wherein the learning model is used to determine weight adjustment parameters corresponding to impact weights according to different investment environments, wherein the investment environment includes an economic environment and a policy environment; In this step, the evaluation system needs to collect historical project data under different investment environments and build a learning model based on this data. The specific technical means are as follows: 1) Connect to the internal database of the enterprise and query historical project data. 2) Extract the macroeconomic indicator data and industry policy data of the period of the project in the database to determine the investment environment at that time, such as GDP growth rate represents the economic environment, and power market reform represents the policy environment. 3) Call the external financial database interface to obtain financial market indicators during the project period, and supplement the characteristics of the investment environment at that time, such as exchange rate level, loan interest rate, etc. 4) Collect project data at different stages of the same project to determine the impact of changes in the investment environment on the project, such as changes in output benefits. 5) Label the project data and mark the type of investment environment to which it belongs, such as "rapid economic growth" and "tightening policy", etc., to correspond to different environments. 6) Build a machine learning model, use project data and corresponding investment environment as samples, and train the model. The goal is to train a classification model to identify the type of investment environment to which the new project belongs. 7) Collect the adjustment parameter data set of the weights of key influencing factors under different investment environments, use the collected adjustment parameters of the weights of key influencing factors as labels, and train the machine learning model in combination with the investment environment type so that it can learn to generate weight adjustment plans under different environments.

[0049] S202, extracting existing data features of the target power grid company's comprehensive plan project, and inputting the existing data features into the learning model to obtain a corresponding investment environment type; In this step, the evaluation system needs to extract the existing data features of the comprehensive plan project of the target power grid company and input them into the previously trained learning model for investment environment type identification. Specifically, the evaluation system connects to the planning database, queries the basic information table, investment budget table, benefit forecast table, etc. of the comprehensive plan project of the target power grid company, and obtains the basic data of the project. Natural language processing technology is used to parse the unstructured text information in the project overview document and extract keywords and semantic features in the text. Then the regional statistical information interface is called to obtain the economic development statistics of the project area in the past 3-5 years, such as GDP growth rate, fiscal revenue growth rate, etc., to reflect the regional economic environment. The industry policy database is called to query the policy changes of the target industry in recent years. If there are major policy adjustments and regulations, the features are recorded. Then the extracted structured and unstructured information is processed, new features are constructed, normalized and standardized, and a standardized feature data set is obtained. Then the feature data set is input into the trained learning model to run the model judgment program. The learning model analyzes and judges the feature set, and judges the investment environment type that the target project is most likely to belong to based on the different feature values ​​in the feature data set.

[0050] S203, determining a corresponding weight dynamic adjustment parameter according to the investment environment type; Query the weight adjustment parameter data set, determine the corresponding historical weight adjustment plan according to the investment environment type, calculate the average historical adjustment ratio of the weights of each key influencing factor under this type of environment, and generate the weight adjustment benchmark parameters. Combined with the specific data characteristics of the target power grid company's comprehensive plan project, conduct secondary training and adjustment of the benchmark parameters to make them more in line with the actual situation of the project. Use the weight adjustment range verification mechanism to check whether the adjustment parameters are within the predetermined reasonable range. If they exceed, truncate them to generate the final weight adjustment parameters, and record the parameter name, adjustment target, adjustment range and other elements.

[0051] S204. Dynamically adjust the influence weight of the key influencing factor in the investment accuracy evaluation model according to the dynamic weight adjustment parameter.

[0052] Load the weight module of the investment accuracy assessment model to determine the default static weights of each key influencing factor in the current investment accuracy assessment model. Input dynamic adjustment parameters to parse out the target factors that need to be adjusted and the positive and negative adjustment ranges. According to the adjustment targets and ranges, automatically calculate and generate new dynamic weight values ​​to replace the original static weights. Update the new weights to the model weight module, overwrite the original default static weight values, and complete the dynamic adjustment. Run the adjusted model to generate a new assessment report that takes dynamic factors into account, compare the reports before and after the dynamic adjustment, analyze the impact of weight changes on the results, and continuously improve the dynamic weight adjustment plan through the feedback regression mechanism.

[0053] In the embodiment of the present application, since the impact of different investment environments on the investment accuracy assessment is taken into account, a learning model is constructed by collecting historical power grid company comprehensive plan project data under different investment environments, extracting existing data characteristics of the target power grid company comprehensive plan project to determine the investment environment type, and dynamically adjusting the influence weights of key influencing factors according to the investment environment type. This effectively solves the problem that the investment environment factors are not considered in the prior art and the investment accuracy assessment model lacks adaptability, thereby achieving the technical effect of improving the accuracy and flexibility of the investment accuracy assessment and providing more accurate guidance for power grid investment under different environments.

[0054] The following describes the evaluation system in the embodiment of the present invention from the perspective of hardware processing. Figure 3 , is a schematic diagram of a physical device structure of an evaluation system in an embodiment of the present application.

[0055] It should be noted that Figure 3 The structure of the evaluation system shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0056] like Figure 3 As shown, the evaluation system includes a central processing unit (CPU) 301, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 302 or the program loaded from the storage part 308 to the random access memory (RAM) 303, such as executing the method described in the above embodiment. In the RAM 303, various programs and data required for system operation are also stored. The CPU 301, the ROM 302 and the RAM 303 are connected to each other through the bus 304. The input / output (I / O) interface 305 is also connected to the bus 304.

[0057] The following components are connected to the I / O interface 305: an input section 306 including an audio input device, a button switch, etc.; an output section 307 including a liquid crystal display (LCD) and an audio output device, an indicator light, etc.; a storage section 308 including a hard disk, etc.; and a communication section 309 including a network interface card such as a LAN (Local Area Network) card, a modem, etc. The communication section 309 performs communication processing via a network such as the Internet. A drive 310 is also connected to the I / O interface 305 as needed. A removable medium 311, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 310 as needed so that a computer program read therefrom is installed into the storage section 308 as needed.

[0058] In particular, according to an embodiment of the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present invention includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a computer program for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through the communication part 309, and / or installed from a removable medium 311. When the computer program is executed by the central processing unit (CPU) 301, various functions defined in the present invention are performed.

[0059] It should be noted that specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a flash memory, an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, apparatus, or device.

[0060] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present invention. Each box in the flowchart or block diagram may represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of a code contains one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions marked in the box may also occur in an order different from that marked in the accompanying drawings.

[0061] Specifically, the evaluation system of this embodiment includes a processor and a memory. The memory stores a computer program. When the computer program is executed by the processor, the machine learning-based comprehensive planning project evaluation method for power grid companies provided in the above embodiment is implemented.

[0062] As another aspect, the present invention further provides a computer-readable storage medium, which may be included in the evaluation system described in the above embodiment; or may exist independently without being assembled into the evaluation system. The above storage medium carries one or more computer programs, and when the above one or more computer programs are executed by a processor of the evaluation system, the evaluation system implements the power grid company comprehensive plan project evaluation method based on machine learning provided in the above embodiment.

[0063] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

[0064] As used in the above embodiments, the term "when..." may be interpreted to mean "if..." or "after..." or "in response to determining..." or "in response to detecting...", depending on the context. Similarly, the phrases "upon determining..." or "if (the stated condition or event) is detected" may be interpreted to mean "if determining..." or "in response to determining..." or "upon detecting (the stated condition or event)" or "in response to detecting (the stated condition or event)", depending on the context.

[0065] Those skilled in the art can understand that to implement all or part of the processes in the above-mentioned embodiments, the processes can be completed by computer programs to instruct related hardware, and the programs can be stored in computer-readable storage media. When the programs are executed, they can include the processes of the above-mentioned method embodiments. The aforementioned storage media include: ROM or random access memory RAM, magnetic disk or optical disk and other media that can store program codes.

Claims

1. A method for evaluating a comprehensive planning project of a power grid company based on machine learning, applied to an evaluation system, characterized in that: The method comprises: Obtain multiple historical project input data and corresponding historical actual output benefit data of historical power grid company comprehensive planning projects through the internal database of the power grid enterprise; Performing machine learning on the historical project input data and the historical actual output benefit data to construct an input-output analysis model, wherein the input-output analysis model is used to determine key influencing factors between different input factors and output benefits; According to the influence degree of the key influencing factors on investment accuracy, the influence weights of different key influencing factors are determined to obtain the correlation analysis results with the influence weights, and the influence degree of investment accuracy is determined according to the deviation degree between the actual output benefit data and the expected output benefit data; Constructing an investment accuracy evaluation model according to the key influencing factors and the correlation analysis results, wherein the investment accuracy evaluation model is used to determine evaluation information of the key influencing factors on investment accuracy; After obtaining the project input data and the expected output benefit data of the target power grid company's comprehensive plan project to be carried out, the project input data and the expected output benefit data are input into the investment accuracy evaluation model to obtain the investment accuracy pre-evaluation information of the target power grid company's comprehensive plan project; After receiving the evaluation viewing request instruction sent by the display terminal, the investment accuracy pre-evaluation information is sent to the display terminal for display.

2. The method according to claim 1, characterized in that After the step of constructing an investment accuracy evaluation model according to the key influencing factors and the correlation analysis results, the method further includes: Encrypting the calculation process of the investment accuracy assessment model; An access permission is set for the investment accuracy assessment model, and the access permission is used to limit authorized personnel to operate the investment accuracy assessment model.

3. The method according to claim 1, characterized in that After obtaining the project input data and the expected output benefit data of the target power grid company's comprehensive plan project to be carried out, the project input data and the expected output benefit data are input into the investment accuracy evaluation model to obtain the investment accuracy pre-evaluation information of the target power grid company's comprehensive plan project, and the step also includes: Comparing and analyzing the investment accuracy pre-assessment information with the actual investment accuracy data; Calculating the error rate between the investment accuracy pre-assessment information and the actual investment accuracy data; If the error rate is greater than a preset threshold, the investment accuracy assessment model is adjusted until the error rate is less than or equal to the preset threshold.

4. The method according to claim 1, characterized in that: After obtaining the project input data and the expected output benefit data of the target power grid company's comprehensive plan project to be carried out, the project input data and the expected output benefit data are input into the investment accuracy evaluation model to obtain the investment accuracy pre-evaluation information of the target power grid company's comprehensive plan project, and the step also includes: Collecting historical power grid company comprehensive plan project data under different investment environments, and constructing a learning model according to different investment environments, wherein the learning model is used to determine weight adjustment parameters corresponding to the impact weights according to different investment environments, wherein the investment environment includes an economic environment and a policy environment; Extracting existing data features of the target power grid company's comprehensive plan project, and inputting the existing data features into the learning model to obtain a corresponding investment environment type; Determine the corresponding weight dynamic adjustment parameter according to the investment environment type; The influence weights of the key influencing factors in the investment accuracy evaluation model are dynamically adjusted according to the weight dynamic adjustment parameters.

5. The method according to claim 1, characterized in that After receiving the evaluation viewing request instruction sent by the display terminal, after sending the investment accuracy pre-evaluation information to the display terminal for display, the method further includes: Determine the investment strategy based on the investment accuracy pre-assessment information and the preset investment strategy library mapping relationship library; After receiving the investment strategy viewing instruction sent by the display end, the investment strategy is sent to the display end.

6. The method according to claim 5, characterized in that After receiving the evaluation viewing request instruction sent by the display terminal, the step of sending the investment accuracy pre-evaluation information to the display terminal for display specifically includes: After receiving the evaluation viewing request instruction sent by the display terminal, extracting key display data in the investment accuracy pre-evaluation information; Based on a preset natural language template, filling the extracted key display data into a preset position in the natural language template and generating a preliminary text description; Performing semantic optimization on the preliminary text description to obtain a text interpretation report; The text interpretation report is sent to the display terminal for display.

7. The method according to claim 1, characterized in that Before the step of performing machine learning on the historical project input data and the historical actual output benefit data to construct an input-output analysis model, the step further includes: Obtain project attribute information of the comprehensive planning project of the target power grid company; Determine the project category according to the project attribute information; According to the project category, historical project sample data of the corresponding category is selected as input data of the input-output analysis model to obtain the category-customized input-output analysis model.

8. An evaluation system, characterized in that: The evaluation system comprises: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code comprises computer instructions, and the one or more processors call the computer instructions so that the evaluation system executes the method as described in any one of claims 1-7.

9. A computer-readable storage medium comprising instructions, characterized in that: When the instructions are executed on an evaluation system, the evaluation system is caused to execute the method according to any one of claims 1 to 7.

10. A computer program product, characterized in that When the computer program product is run on an evaluation system, the evaluation system is caused to perform the method according to any one of claims 1 to 7.

Images