Photovoltaic power station operation and maintenance cost prediction method and device and storage medium
By establishing an operation and maintenance cost database and prediction model based on machine learning algorithms, the problem of lack of systematicity and accuracy of operation and maintenance cost prediction in the existing technology is solved, and high-accuracy operation and maintenance cost prediction is achieved worldwide.
Patent Information
- Application Number
- CN202510112967.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-01-22
- Publication Date
- 2025-05-16
AI Technical Summary
The existing photovoltaic power station operation and maintenance cost prediction methods are difficult to consider a variety of variables globally, and lack systematicity and accuracy, making it difficult to quickly and accurately predict operation and maintenance costs in different regions.
By collecting historical operation and maintenance data of photovoltaic power stations, extracting characteristic variables and data, establishing an operation and maintenance cost database, building a prediction model based on machine learning algorithms, training and optimizing the model, and finally correcting predictions based on the data to be predicted to obtain accurate operation and maintenance cost prediction values.
It has achieved high accuracy and adaptability in the global operation and maintenance cost forecast, which can accurately predict operation and maintenance costs and supports the long-term economic planning of photovoltaic power plants.
Smart Images
Figure CN120013006A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of photovoltaic power station operation and maintenance cost prediction, and in particular to a photovoltaic power station operation and maintenance cost prediction method, device and storage medium. Background Art
[0002] With the rapid development of photovoltaic power station construction around the world, the operation and maintenance costs of photovoltaic power stations have become an important factor affecting the long-term economic viability of the project. Due to differences in economic conditions, environmental conditions, technical characteristics, market maturity, etc. in different countries and regions, the operation and maintenance costs of photovoltaic power stations have significant regional differences. Most of the existing photovoltaic power station operation and maintenance cost prediction methods are based on single factors or local regional data analysis, lacking a systematic method that considers multiple variables on a global scale.
[0003] Traditional operation and maintenance cost forecasting methods are difficult to fully consider the actual conditions in different regions. In the global operation of photovoltaic projects, how to quickly and accurately predict the operation and maintenance costs of photovoltaic power stations in different regions has become a technical problem that needs to be solved urgently.
[0004] Therefore, there is an urgent need for a photovoltaic power station operation and maintenance cost prediction method that can be applied globally, with high accuracy and adaptability. Summary of the invention
[0005] The technical problem to be solved by the present application is to provide a method, device and storage medium for predicting the operation and maintenance cost of a photovoltaic power station.
[0006] The technical solution adopted by the present application to solve the technical problem is to provide a photovoltaic power station operation and maintenance cost prediction method, including:
[0007] Collect historical operation and maintenance data of several photovoltaic power plants, extract characteristic variables and corresponding characteristic variable data based on the historical operation and maintenance data, and establish an operation and maintenance cost database;
[0008] Build a prediction model based on the machine learning algorithm, and train the prediction model through the operation and maintenance cost database to obtain a trained prediction model;
[0009] Based on the trained prediction model, a basic prediction value is obtained according to the predicted operation and maintenance data of the photovoltaic power station to be predicted, and the basic prediction value is corrected to obtain a corrected prediction value.
[0010] Optionally, the characteristic variables include regional environment, per capita installed capacity, project technology type, operation and maintenance mode, equipment operation time and historical operation and maintenance costs.
[0011] Optionally, before the step of building a prediction model based on a machine learning algorithm, the step further includes:
[0012] Analyze the correlation between characteristic variables in historical operation and maintenance data;
[0013] Determine the machine learning algorithm based on the correlation analysis results.
[0014] Optionally, the machine learning algorithm may be a random forest algorithm.
[0015] Optionally, the steps of building a prediction model based on a machine learning algorithm include:
[0016] According to the prediction formula of the random forest regression model: Construct a prediction model, where M is the basic prediction value, N is the number of decision trees, X represents the feature vector, which consists of feature variables, and f i (X) represents the prediction result of the i-th decision tree for the feature vector.
[0017] Optionally, after the step of building a prediction model based on a machine learning algorithm and training the prediction model through the operation and maintenance cost database to obtain a trained prediction model, the step further includes:
[0018] The cross-validation method is used to optimize the model parameters of the prediction model, including the number of decision trees, the maximum depth of the tree, the minimum number of sample classifications, the minimum number of sample leaves, and the maximum number of features.
[0019] Optionally, the step of correcting the basic predicted value to obtain a corrected predicted value includes:
[0020] Calculate labor cost adjustment factor and equipment cost adjustment factor;
[0021] The revised forecast value is revised based on the labor cost adjustment factor and the equipment cost adjustment factor.
[0022] Optionally, the step of correcting the corrected forecast value in combination with the labor cost adjustment coefficient and the equipment cost adjustment coefficient includes:
[0023] According to the revised formula: M 区域修正值 =M*C GNI *C CAPEX , where M 基础预测值 is the basic prediction value obtained based on the prediction model, C GNI is the labor cost adjustment coefficient, C CAPEX is the equipment cost adjustment factor.
[0024] In addition, the present application also provides a device, which includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the above-mentioned photovoltaic power station operation and maintenance cost prediction method.
[0025] In addition, the present application also provides a storage medium, which is a computer-readable storage medium. A computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the above-mentioned photovoltaic power station operation and maintenance cost prediction method are implemented.
[0026] The technical solution implemented in the present application has at least the following beneficial effects: first, historical operation and maintenance data of several photovoltaic power stations are collected, characteristic variables and corresponding characteristic variable data are extracted based on the historical operation and maintenance data, and an operation and maintenance cost database is established; then, a prediction model is constructed based on a machine learning algorithm, and the prediction model is trained through the operation and maintenance cost database to obtain a trained prediction model; finally, based on the trained prediction model, a basic prediction value is obtained according to the predicted operation and maintenance data of the photovoltaic power station to be predicted, and the basic prediction value is corrected to obtain a corrected prediction value, which has the beneficial effect of accurately predicting the operation and maintenance cost of the photovoltaic power station. BRIEF DESCRIPTION OF THE DRAWINGS
[0027] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0028] Figure 1 It is a flow chart of the first embodiment of the photovoltaic power station operation and maintenance cost prediction method provided in the present application. DETAILED DESCRIPTION
[0029] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0030] The present application provides a method for predicting the operation and maintenance cost of a photovoltaic power station. Figure 1 , Figure 1 This is a flow chart of the first embodiment of the photovoltaic power station operation and maintenance cost prediction method of the present application.
[0031] In this embodiment, the photovoltaic power station operation and maintenance cost prediction method includes steps S10 to S30:
[0032] Step S10, collecting historical operation and maintenance data of several photovoltaic power plants, extracting characteristic variables and corresponding characteristic variable data according to the historical operation and maintenance data, and establishing an operation and maintenance cost database;
[0033] It should be noted that the operation and maintenance data of different photovoltaic power stations are generally stored in the project's data management system. The historical operation and maintenance data can be manually extracted or automatically collected from the databases of different projects. The historical operation and maintenance data of each power station is extracted and aggregated into a database to facilitate subsequent data extraction.
[0034] Specifically, operation and maintenance data are extracted manually or automatically in the data management systems of photovoltaic power stations in different regions, and characteristic variables and corresponding characteristic variable data are extracted based on the operation and maintenance data, a data table containing key operation and maintenance data is created, and an operation and maintenance cost database is built.
[0035] Exemplarily, historical operation and maintenance data are manually recorded and extracted from the project management systems of photovoltaic power stations in different regions, the SCADA system (Supervisory Control And Data Acquisition) of the centralized control center, and the financial reporting system. Feature variables and feature variable data are extracted from the historical operation and maintenance data, a data table containing the feature variables is created, and an operation and maintenance cost database is built to facilitate the extraction of operation and maintenance data directly from the operation and maintenance cost database.
[0036] Optionally, the characteristic variables include regional environment, per capita installed capacity, project technology type, operation and maintenance mode, equipment operation time and historical operation and maintenance costs.
[0037] It should be noted that the reason for obtaining the regional environment is that the regional environment (such as climate conditions, topography, etc.) directly affects the equipment operation and maintenance requirements of photovoltaic power stations. For example, arid areas may cause dust accumulation and increase the workload of cleaning, while cold areas may increase the risk of equipment damage. The differences in climate conditions and topography in different regions will directly affect the maintenance frequency and cost of power stations. The reason for obtaining per capita installed capacity is that per capita installed capacity reflects the scale of the power station and the workload of each worker or equipment. The reason for obtaining project technology is that different project technology types determine the complexity and failure rate of the equipment. Equipment of different technical types has different maintenance methods and costs. The reason for obtaining the operation and maintenance mode is that different operation and maintenance modes affect labor costs. The reason for obtaining the equipment operation time is that the equipment operation time determines the degree of equipment aging. After a certain period of time, the equipment failure rate will increase, and the annual operation and maintenance costs will also increase.
[0038] Specifically, the regional environment, per capita installed capacity, project technology type, operation and maintenance mode, equipment operation time and historical operation and maintenance costs of photovoltaic power stations in different regions of the world are obtained, and characteristic variables and characteristic variable data corresponding to the characteristic variables are extracted according to the historical operation and maintenance data. For example, the extracted regional environmental characteristic variables include temperature, precipitation, terrain, etc. Assuming that power station 1 is located in a "warm and dry plain", "the temperature is 25°C", "the annual precipitation is 100mm", and "the terrain is a plain", the regional environmental characteristics are temperature, precipitation and terrain, etc., then the regional environmental characteristic data can be: [temperature = 25, precipitation = 100, terrain = plain].
[0039] Optionally, the acquired operation and maintenance data can also use dimensionality reduction techniques such as principal component analysis and factor analysis to optimize the characteristic dimensions of characteristic variables.
[0040] Step S20, building a prediction model based on a machine learning algorithm, and training the prediction model through an operation and maintenance cost database to obtain a trained prediction model;
[0041] Specifically, the feature variables extracted from the operation and maintenance cost database and the historical operation and maintenance cost data are used as model inputs to train the machine learning algorithm model.
[0042] Optionally, after step S20, step S21 is further included:
[0043] Step S21, using a cross-validation method to optimize the model parameters of the prediction model, the model parameters include the number of decision trees, the maximum depth of the tree, the minimum number of sample classifications, the minimum number of sample leaves and the maximum number of features.
[0044] It should be noted that the cross-validation method evaluates the performance of the model by dividing the data set into multiple subsets (usually k subsets) and using one of the subsets as the validation set and the other subsets as the training set in turn. In this way, the adaptability of the trained prediction model to different data sets can be effectively detected, the deviation caused by imbalanced training sets or accidental factors can be reduced, and each parameter setting can be ensured to have a positive impact on the performance of the trained prediction model, while preventing overfitting and underfitting.
[0045] Step S30, based on the trained prediction model, a basic prediction value is obtained according to the predicted operation and maintenance data of the photovoltaic power station to be predicted, and the basic prediction value is corrected to obtain a corrected prediction value.
[0046] Specifically, the operation and maintenance data of photovoltaic power stations in a certain area are input into the trained prediction model to predict the operation and maintenance cost, and a basic prediction value is obtained. The basic prediction value is corrected to obtain an accurate operation and maintenance cost prediction value.
[0047] First, historical operation and maintenance data of several photovoltaic power stations are collected, and characteristic variables and corresponding characteristic variable data are extracted based on the historical operation and maintenance data to establish an operation and maintenance cost database. Then, a prediction model is constructed based on a machine learning algorithm, and the prediction model is trained through the operation and maintenance cost database to obtain a trained prediction model. Finally, based on the trained prediction model, a basic prediction value is obtained according to the predicted operation and maintenance data of the photovoltaic power station to be predicted, and the basic prediction value is corrected to obtain a corrected prediction value, which has the beneficial effect of accurately predicting the operation and maintenance cost of the photovoltaic power station.
[0048] The present application also provides a second embodiment, and the contents in the second embodiment that are the same or similar to the first embodiment are not repeated here. The machine learning algorithm may be a random forest algorithm, and step S20, the step of building a prediction model based on the machine learning algorithm may include:
[0049] The prediction model is built based on the random forest algorithm and trained based on the feature variables.
[0050] The random forest algorithm model formula is:
[0051]
[0052] Where M is the base prediction value, N is the number of decision trees in the random forest algorithm, and f i (X) represents the prediction result of the i-th decision tree for the input feature vector X, X represents the feature vector, including characteristic variables such as regional environmental characteristics (C_env), per capita installed capacity (C_cap), project technology type (C_dev), operation and maintenance mode (C_mode), equipment operation time (C_time), etc., where regional environmental characteristics include temperature, precipitation, etc. In addition, the machine learning algorithm can also be a support vector machine, a neural network, etc.
[0053] Optionally, before the step S20 of building a prediction model based on a machine learning algorithm, the step S22 to S23 are also included:
[0054] Step S22, analyzing the correlation between characteristic variables in the historical operation and maintenance data; Step S23, determining a machine learning algorithm based on the correlation analysis results.
[0055] It should be noted that the reason for the correlation analysis is that some characteristic variables in the historical operation and maintenance data are associated and will affect each other. For example, there is a correlation between historical operation and maintenance costs and other characteristic variables in the historical operation and maintenance data. The characteristic variables of non-historical operation and maintenance costs in the historical operation and maintenance data include but are not limited to regional environment, per capita installed capacity, project technology type, operation and maintenance mode, equipment operation time, equipment failure rate, component type, number of operation and maintenance personnel, etc. However, some key characteristic variables in the characteristic variables of non-historical operation and maintenance costs have a large correlation with historical operation and maintenance costs. For another example, in addition to historical operation and maintenance costs, there will be correlations between characteristic variables in historical operation and maintenance data and they will affect each other. Therefore, the appropriate machine learning algorithm can be determined by analyzing the correlation between the characteristic variables in the historical operation and maintenance data, so that the subsequent training model will be more accurate.
[0056] Specifically, first calculate the correlation coefficient between each feature variable, such as the Pearson Correlation Coefficient, which is used to measure the linear relationship between data, and the Spearman Rank Correlation, which is used to evaluate the nonlinear relationship between data and quantify the degree of correlation between different feature variables. Then, through feature selection algorithms, such as Lasso regression (L1 regularization) or feature importance evaluation algorithms based on tree models, select the feature variables that have the greatest impact on historical operation and maintenance costs from the historical operation and maintenance data. In addition, there will be correlations between data other than historical operation and maintenance costs, which will in turn affect the operation and maintenance costs. For example, there is a correlation between the operation and maintenance mode and the number of operation and maintenance personnel, and the component type affects the equipment failure rate. The relationship between these feature variables can be determined through multivariate regression analysis, thereby further determining the impact of the relationship between feature variables on the operation and maintenance costs. Finally, determine the machine learning algorithm based on the correlation between each feature variable. If most feature variables have a strong correlation and the relationship is relatively linear or monotonic, then linear regression or support vector machine (SVM) can be selected. However, if the relationship between feature variables is complex, contains nonlinearity or has strong interactions, the random forest algorithm is more suitable. The random forest algorithm can automatically discover complex nonlinear relationships between variables by training different decision trees multiple times and evaluating the importance of features. It can also process high-dimensional data and avoid feature omissions.
[0057] In this embodiment, before determining the machine learning algorithm, the correlation analysis is performed on each characteristic variable of the historical operation and maintenance data to analyze the key features that affect the operation and maintenance costs and the correlation between the key features. By analyzing the correlation of each characteristic variable in the historical operation and maintenance data, the appropriate machine learning algorithm is accurately determined to improve the accuracy of subsequent model training.
[0058] Optionally, the machine learning algorithm may be determined by analyzing the equipment operating time and regional environment.
[0059] It should be noted that the two characteristics of equipment operation time and regional environment have a significant impact on the operation and maintenance costs of photovoltaic power stations. The equipment operation time is directly related to the degree of equipment aging, and the failure rate will increase with time, thus affecting the cost of maintenance and replacement of parts. The regional environment (such as climate conditions, geographical location, etc.) affects the operating conditions of photovoltaic power stations. For example, temperature and humidity may affect the cleaning frequency, failure rate and maintenance cost of components. Therefore, other operation and maintenance data and operation and maintenance costs can be evaluated through equipment operation time and regional environment.
[0060] Specifically, we can analyze different equipment operating times and regional environments to infer the per capita installed capacity, project technology type, and operation and maintenance model selection, and then evaluate the operation and maintenance costs.
[0061] In this embodiment, based on the analysis of equipment operation time and regional environment, other variables (such as per capita installed capacity, project technology type, operation and maintenance mode) and their impact on operation and maintenance costs can be inferred. By analyzing the relationship between these two key features and other variables, data support can be provided for building a more accurate prediction model.
[0062] In combination with the prediction model constructed by the random forest algorithm in this embodiment, the trained prediction model is obtained after training with historical operation and maintenance data. This trained prediction model is used to predict the operation and maintenance cost. For example, feature vector 1 (regional environment: warm and dry, plain, per capita installed capacity: 500MW, project technology type: fixed bracket, operation and maintenance mode: base operation and maintenance, equipment operation time: 5 years). Input feature vector 1 into the trained prediction model constructed by the random forest, and the basic prediction value of each MW of the project in the fifth year can be predicted. When the random forest model is input with feature vector 1, based on the training experience, the impact of regional environment, installed capacity, project technology type, operation and maintenance mode and equipment operation time on the operation and maintenance cost can be analyzed. Photovoltaic power stations in warm and dry areas may face fewer dust accumulation problems in equipment cleaning, and the environmental complexity of operation and maintenance is relatively low. The scale of a 500MW power station is large, and the unit labor cost will be lower than that of a 50M small power station. Fixed tilt brackets and string inverters are simple to maintain, have low failure rates, and relatively less maintenance workload. Base operation and maintenance can realize the sharing of technicians of multiple projects and reduce the labor cost of a single site. Five years of operation means that the equipment is less aged, has a lower failure rate, and has relatively low operation and maintenance costs.
[0063] For feature vector 2 (regional environment: cold and humid, mountainous area, per capita installed capacity: 100MW, project technology type: tracking system, centralized inverter, operation and maintenance mode: independent site, equipment operation time: 10 years), feature vector 2 is input into the trained prediction model constructed by the random forest algorithm, and the operation and maintenance cost per MW of the project in the 10th year can be predicted. Based on the experience of operation and maintenance data training, the random forest model can predict the basic prediction value of the photovoltaic power station in the region in the 10th year when feature vector 2 is used as the input value. Random forest model analysis process: A medium-sized power station of 100MW means more operation and maintenance personnel per MW, and the overall operation and maintenance cost is higher. Although the tracking system improves the power generation efficiency, it requires more maintenance and inspection. The centralized inverter is relatively complex and the maintenance workload is also large. Independent sites, without coordination, require more operation and maintenance personnel, and the operation and maintenance cost is high. 10 years of operation means that the equipment may be aged to a certain extent, the probability of failure is high, and the operation and maintenance cost may increase. Prediction results: The cold and humid environment, complex technical configuration, large-scale power stations and older equipment make the prediction results of this feature vector combination tend to be higher operation and maintenance costs.
[0064] The present application also provides a third embodiment, and the contents in the third embodiment that are the same or similar to the first embodiment or the second embodiment are not repeated here. In this embodiment, step S30 corrects the basic prediction value, and the step of obtaining the corrected prediction value includes steps S31 to S32:
[0065] Step S31, calculating the labor cost adjustment coefficient and the equipment cost adjustment coefficient;
[0066] Specifically, through the formula: , calculate the labor cost adjustment coefficient,
[0067] Among them, C GNI is the labor cost adjustment coefficient, and GNI is the real gross national income per capita.
[0068] By formula: Calculate the equipment cost adjustment factor, where C CAPEX is the equipment cost adjustment coefficient, CAPEX is the project construction cost, CAPEX of the project country is the construction cost of the estimated project, and China CAPEX is the average cost of China's photovoltaic projects in that year.
[0069] The calculated labor cost adjustment coefficient and equipment cost adjustment coefficient are stored in the adjustment coefficient library so that data can be directly obtained from the database during subsequent calculations.
[0070] In addition, it should be noted that the labor cost adjustment factor reflects the difference in labor costs between countries by comparing the per capita national income of the project country with that of China. For example, if the GNI of the project country is $10,000 and that of China is $5,000, then the CGNI is 2, which means that the labor cost is twice that of China. The equipment cost adjustment factor reflects the difference in equipment costs by comparing the average cost of photovoltaic projects in the project country with that of China. For example, if the CAPEX of the project country is $1,000,000 and that of China is $800,000, then the CCAPEX is 1.25, which means that the equipment cost is 25% higher than that of China.
[0071] Step S32, correcting the corrected prediction value in combination with the labor cost adjustment coefficient and the equipment cost adjustment coefficient.
[0072] Specifically, by combining the labor cost adjustment coefficient and the equipment cost adjustment coefficient, a weighted adjustment is made to the basic forecast value, so that the predicted cost is more in line with the actual regional cost level.
[0073] Optionally, step S32 of correcting the corrected prediction value in combination with the labor cost adjustment coefficient and the equipment cost adjustment coefficient includes:
[0074] According to the revised formula: M 区域修正值 =M*C GNI *C CAPEX (4), where M 基础预测值 is the basic prediction value obtained based on the prediction model, C GNI is the labor cost adjustment coefficient, C CAPEX is the equipment cost adjustment factor.
[0075] In this embodiment, the correction of the forecast value by combining the adjustment coefficients of labor cost and equipment cost can make the model forecast result closer to the actual market environment, avoiding the errors that may be caused by relying solely on model data. Through this correction process, the forecast results can fully consider the economic differences in different regions, thereby providing a more accurate reference for the operation and maintenance costs of photovoltaic power stations in different regions of the world, and improving the reliability and accuracy of the forecast.
[0076] In addition, the present application also provides a device, the device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the above-mentioned photovoltaic power station operation and maintenance cost prediction method. This device can achieve the beneficial effects of the above-mentioned embodiment.
[0077] In addition, the present application also provides a storage medium, which is a computer-readable storage medium, and stores a computer program on the storage medium, and when the computer program is executed by a processor, the steps of the photovoltaic power station operation and maintenance cost prediction method are implemented. The storage medium can achieve the beneficial effects of the above embodiments.
[0078] Finally, it should be noted that the steps in the method of the embodiment of the present application can be adjusted in order, combined and deleted according to actual needs.
[0079] In addition, a person skilled in the art may understand that all or part of the steps in the various methods of the above embodiments may be completed by instructing the hardware related to the device through a program, and the program may be stored in a computer-readable storage medium, and the storage medium may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.
[0080] In addition, those skilled in the art will appreciate that all or part of the steps in the various methods of the above embodiments may be completed by functional modules in the device, which will not be elaborated herein.
[0081] What is disclosed above is only a preferred embodiment of the present application, and it certainly cannot be used to limit the scope of rights of the present application. A person skilled in the art can understand that all or part of the processes of implementing the above embodiments and making equivalent changes according to the claims of the present application still fall within the scope of the invention.
Claims
1. A method for predicting operation and maintenance costs of a photovoltaic power station, characterized in that: The photovoltaic power station operation and maintenance cost prediction method comprises: Collecting historical operation and maintenance data of several photovoltaic power stations, extracting characteristic variables and corresponding characteristic variable data according to the historical operation and maintenance data, and establishing an operation and maintenance cost database; Building a prediction model based on a machine learning algorithm, and training the prediction model through the operation and maintenance cost database to obtain a trained prediction model; Based on the trained prediction model, a basic prediction value is obtained according to the predicted operation and maintenance data of the photovoltaic power station to be predicted, and the basic prediction value is corrected to obtain a corrected prediction value.
2. The photovoltaic power station operation and maintenance cost prediction method according to claim 1, characterized in that: The characteristic variables include regional environment, per capita installed capacity, project technology type, operation and maintenance mode, equipment operation time and historical operation and maintenance costs.
3. The photovoltaic power station operation and maintenance cost prediction method according to claim 2, characterized in that: The step of building a prediction model based on a machine learning algorithm also includes: Analyzing the correlation between the characteristic variables in the historical operation and maintenance data; The machine learning algorithm is determined based on the correlation analysis results.
4. The photovoltaic power station operation and maintenance cost prediction method according to claim 1, characterized in that: The machine learning algorithm may be a random forest algorithm.
5. The photovoltaic power station operation and maintenance cost prediction method according to claim 4, characterized in that: The step of building a prediction model based on a machine learning algorithm includes: According to the prediction formula of the random forest regression model: Construct the prediction model, where M is the basic prediction value, N is the number of decision trees, X represents the feature vector, which is composed of the feature variables, and f i (X) represents the prediction result of the i-th decision tree for the feature vector.
6. The photovoltaic power station operation and maintenance cost prediction method according to claim 5, characterized in that: After the step of building a prediction model based on a machine learning algorithm and training the prediction model through the operation and maintenance cost database to obtain a trained prediction model, the following steps are further included: The model parameters of the trained prediction model are optimized using a cross-validation method, wherein the model parameters include the number of decision trees, the maximum depth of the tree, the minimum number of sample classifications, the minimum number of sample leaves, and the maximum number of features.
7. The photovoltaic power station operation and maintenance cost prediction method according to claim 1, characterized in that: The step of correcting the basic predicted value to obtain a corrected predicted value comprises: Calculate labor cost adjustment factor and equipment cost adjustment factor; The revised forecast value is revised in combination with the labor cost adjustment coefficient and the equipment cost adjustment coefficient.
8. The photovoltaic power station operation and maintenance cost prediction method according to claim 7, characterized in that: The step of correcting the corrected forecast value by combining the labor cost adjustment coefficient and the equipment cost adjustment coefficient comprises: According to the revised formula: M 区域修正值 =M*C GNI *C CAPEX , where M 基础预测值 is the basic prediction value obtained based on the prediction model, C GNI is the labor cost adjustment coefficient, C CAPEX is the equipment cost adjustment coefficient.
9. A device, characterized in that: The device comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the photovoltaic power station operation and maintenance cost prediction method according to any one of claims 1 to 8.
10. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the photovoltaic power station operation and maintenance cost prediction method according to any one of claims 1 to 8 are implemented.