Multi-source data fusion photovoltaic power generation fault early warning method and system

Abstract

The invention relates to the technical field of photovoltaic power generation, and provides a photovoltaic power generation fault early warning method and system based on multi-source data fusion. The method comprises the following steps: obtaining an irradiation area and an inclination angle of a component through a photovoltaic array, and receiving an environment monitoring parameter by a meteorological monitoring port; fitting the irradiation area and the inclination angle for generating capacity prediction to generate unit generating capacity; receiving the factor state value and the distribution topology information, and performing multi-source data retrieval in combination with the unit generating capacity to obtain a retrieval state value; fusing the retrieval state values to generate an expected state value; performing deviation analysis on the time sequence information based on the expected state value to generate a deviation coefficient; and if the deviation coefficient is greater than or equal to a deviation coefficient threshold, generating an early warning signal. The technical problem that the early warning accuracy is not high due to the fact that analysis depends on a preset threshold value in traditional photovoltaic power generation fault analysis and the scene combinability is poor is solved, and the effect of improving the fault early warning accuracy and the scene adaptability through multi-source data fusion and deviation analysis is achieved.

Description

Technical Field

[0001] The present application relates to the field of new energy technology, specifically to the field of photovoltaic power generation technology, and in particular to a photovoltaic power generation fault early warning method and system based on multi-source data fusion. Background Art

[0002] With the continuous expansion of the photovoltaic industry, the core position of photovoltaic power generation systems in the production of electricity has become increasingly prominent. However, with the expansion of the scale of photovoltaic power stations and the complexity of process flows, fault warning technology faces unprecedented challenges. Traditional photovoltaic power generation fault warning methods mainly rely on preset threshold values ​​to determine whether the photovoltaic power generation system has a fault. However, due to the complexity of photovoltaic power generation systems and changeable environmental conditions, a single threshold value setting is often difficult to adapt to the fault warning needs in different scenarios, resulting in low warning accuracy and difficulty in meeting the needs of modern photovoltaic power stations for efficient and accurate warnings. Summary of the invention

[0003] The present application provides a photovoltaic power generation fault early warning method and system with multi-source data fusion, aiming to solve the technical problem that traditional photovoltaic power generation fault analysis relies on preset threshold values ​​for analysis, has poor scene integration, and leads to low early warning accuracy.

[0004] In view of the above problems, the present application provides a photovoltaic power generation fault early warning method and system with multi-source data fusion.

[0005] In a first aspect disclosed in the present application, a photovoltaic power generation fault warning method based on multi-source data fusion is provided, the method comprising: obtaining a plurality of irradiation areas and a plurality of irradiation inclination angles of a plurality of photovoltaic modules through a photovoltaic array, and receiving environmental monitoring parameters through a meteorological monitoring port; according to the environmental monitoring parameters, fitting the plurality of irradiation areas and the plurality of irradiation inclination angles one by one to predict power generation, and generating a plurality of unit time power generation; receiving genotype factor state values ​​and photovoltaic power station distribution topology information, performing multi-source data retrieval in combination with the plurality of unit time power generation, and obtaining a multi-source phenotype factor retrieval state value, wherein the genotype factor refers to a photovoltaic power station state that can be directly controlled, and the phenotype factor refers to a photovoltaic power station state that cannot be directly controlled but is indirectly affected by the genotype factor; fusing the multi-source phenotype factor retrieval state values ​​to generate a phenotype factor expected state value; based on the phenotype factor expected state value, performing deviation analysis on the phenotype factor state time series information to generate a phenotype factor deviation coefficient, wherein the phenotype factor state time series information and the environmental monitoring parameters have time series consistency; if the phenotype factor deviation coefficient is greater than or equal to the deviation coefficient threshold, generating a fault warning signal.

[0006] Another aspect disclosed in the present application provides a photovoltaic power generation fault warning system with multi-source data fusion, the system comprising: a data acquisition module, the data acquisition module is used to obtain a number of irradiation areas and a number of irradiation inclination angles of a number of photovoltaic components through a photovoltaic array, and receive environmental monitoring parameters through a meteorological monitoring port; a power generation prediction module, the power generation prediction module is used to fit the several irradiation areas and the several irradiation inclination angles one by one according to the environmental monitoring parameters to perform power generation prediction, and generate a number of unit time power generation; a multi-source data retrieval module, the multi-source data retrieval module is used to receive genotype factor state values ​​and photovoltaic power station distribution topology information, and perform multi-source data retrieval in combination with the several unit time power generation to obtain multi-source phenotype factor retrieval. A state value, wherein the genotype factor refers to the state of the photovoltaic power station that can be directly controlled, and the phenotype factor refers to the state of the photovoltaic power station that cannot be directly controlled but is indirectly affected by the genotype factor; a state value fusion module, the state value fusion module is used to fuse the multi-source phenotype factor retrieval state value to generate the phenotype factor expected state value; a deviation analysis module, the deviation analysis module is used to perform deviation analysis on the phenotype factor state time series information based on the phenotype factor expected state value to generate a phenotype factor deviation coefficient, wherein the phenotype factor state time series information and the environmental monitoring parameter have time series consistency; a coefficient judgment module, the coefficient judgment module is used to generate a fault warning signal if the phenotype factor deviation coefficient is greater than or equal to the deviation coefficient threshold.

[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages: The above-mentioned photovoltaic power generation fault warning method of multi-source data fusion collects the irradiation area and irradiation inclination of each photovoltaic module in the photovoltaic array, as well as the environmental parameters obtained by the meteorological monitoring equipment. Subsequently, these environmental parameters and the physical parameters of the photovoltaic modules are used to predict the power generation, that is, to predict the amount of power that the photovoltaic power station can generate in each time unit. After that, the real-time data on the status of the photovoltaic power station and the layout information of the power station are received. Then, multi-source data retrieval is performed by combining the power generation prediction data to obtain the state value of the phenotype factor. Then, the state values ​​of the multi-source phenotype factors are fused to generate an expected state value. Then, the actual state of the phenotype factor is compared and analyzed with the expected state to generate a deviation coefficient. This deviation coefficient reflects the difference between the current state of the power station and the ideal state. Finally, if the deviation coefficient exceeds the preset threshold, a fault warning signal is issued to notify the user to deal with possible problems in a timely manner. This method improves the accuracy and reliability of fault warning by comprehensively utilizing multiple data sources of the photovoltaic power station, which helps to ensure the safe and stable operation of the photovoltaic power station.

[0008] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0010] Figure 1 A schematic diagram of a process of a photovoltaic power generation fault early warning method using multi-source data fusion in one embodiment; Figure 2 The present invention is an architecture diagram of a photovoltaic power generation fault warning system with multi-source data fusion in one embodiment.

[0011] Explanation of the accompanying drawings: data acquisition module 1, power generation prediction module 2, multi-source data retrieval module 3, state value fusion module 4, deviation analysis module 5, coefficient judgment module 6. DETAILED DESCRIPTION

[0012] The embodiments of the present application provide a photovoltaic power generation fault early warning method and system with multi-source data fusion to solve the technical problem that traditional photovoltaic power generation fault analysis relies on preset threshold values ​​for analysis, has poor scene integration, and leads to low early warning accuracy.

[0013] 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, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0014] It should be noted that the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules that are not explicitly listed or inherent to these processes, methods, products or devices.

[0015] Embodiment 1 like Figure 1 As shown, the present application provides a photovoltaic power generation fault early warning method based on multi-source data fusion, the method comprising: Obtaining a number of irradiation areas and a number of irradiation inclination angles of a number of photovoltaic modules through a photovoltaic array, and receiving environmental monitoring parameters through a meteorological monitoring port; Traditional photovoltaic power generation fault warning methods rely on preset threshold values. This method is often difficult to provide accurate warnings when facing complex and changing environmental conditions and photovoltaic power station operation status. The multi-source data fusion method can comprehensively consider multiple data sources such as the irradiation area, irradiation inclination, meteorological monitoring data, and power station operation data of the photovoltaic array, and generate more accurate fault warning information through comprehensive analysis and processing.

[0016] In the embodiment of the present application, in order to ensure the efficient operation of the power station and prevent potential failures, the system terminal collects and analyzes a variety of data. First, the photovoltaic array is used to obtain the specific irradiation area and irradiation inclination of each photovoltaic module. This information is crucial to understanding the amount of solar radiation received by the photovoltaic module, because the amount of solar radiation directly affects the power generation efficiency of the photovoltaic module. At the same time, the stored environmental monitoring parameters are extracted through the meteorological monitoring port, and the current real-time environmental monitoring parameters are obtained by using the meteorological monitoring port. The system terminal analyzes the stored environmental monitoring parameters and the current real-time environmental monitoring parameters to determine the change of the current environmental parameters. If the current environmental parameters change greatly, the system terminal executes the update of the stored environmental monitoring parameters. Updating the environmental monitoring parameters can help the system terminal more accurately predict the power generation and performance of the photovoltaic power station and identify adverse conditions that may affect the operation of the power station. In summary, by collecting the irradiation area and irradiation inclination of the photovoltaic module, as well as the environmental monitoring parameters, an important data basis is provided for photovoltaic power generation fault warning. These data will be used for subsequent power generation prediction and fault warning analysis to ensure the stable operation and optimized performance of the photovoltaic power station.

[0017] Furthermore, the present application provides methods for obtaining a plurality of irradiation areas and a plurality of irradiation inclination angles of a plurality of photovoltaic modules through a photovoltaic array, and receiving environmental monitoring parameters through a meteorological monitoring port, including: Obtaining a first environmental monitoring parameter, wherein the first environmental monitoring parameter is an original stored environmental monitoring parameter; Obtaining a second environmental monitoring parameter, wherein the second environmental monitoring parameter is a real-time storage environmental monitoring parameter; Preferably, in the operation and maintenance process of the photovoltaic power generation system, the acquisition of environmental monitoring parameters is crucial. Specifically, the system terminal acquires two environmental monitoring parameters, namely the first environmental monitoring parameter and the second environmental monitoring parameter. The first environmental monitoring parameter refers to the original storage environmental monitoring parameter, which is long-term stored data. They represent the environmental conditions of the photovoltaic power station area in the past period of time, such as temperature, humidity, solar radiation intensity, etc. These original monitoring parameters provide the system terminal with basic information about the power station operation environment, which helps to understand the historical background and trend of the power station operation. The second environmental monitoring parameter refers to the real-time storage environmental monitoring parameter, which is obtained through real-time meteorological monitoring equipment and reflects the real-time state of the current environment of the photovoltaic power station. Real-time environmental monitoring parameters have a direct impact on the current operation status analysis and fault warning. For example, real-time temperature, humidity and solar radiation intensity data can help the system terminal accurately predict the current power generation and promptly discover potential operation problems. In summary, by obtaining the original storage environmental monitoring parameters and the real-time storage environmental monitoring parameters, the operating environment of the photovoltaic power station can be fully understood and evaluated. These data provide valuable reference information for the system terminal, which helps to achieve more accurate fault warning.

[0018] Analyzing the environmental fluctuation distance of the second environmental monitoring parameter compared with the first environmental monitoring parameter; When the environmental fluctuation distance is greater than or equal to an environmental fluctuation distance threshold, the first environmental monitoring parameter is updated by the second environmental monitoring parameter.

[0019] Preferably, when evaluating the operating environment of the photovoltaic power station, the system terminal performs a difference analysis between the first environmental monitoring parameter and the second environmental monitoring parameter after obtaining the first environmental monitoring parameter and the second environmental monitoring parameter. Specifically, the system terminal performs deviation calculation on each environmental monitoring parameter attribute in the first environmental monitoring parameter and the second environmental monitoring parameter to obtain the environmental monitoring parameter deviation modulus. When each deviation value in the environmental monitoring parameter deviation modulus is less than or equal to the corresponding threshold, the system terminal performs normalization and other operations on the environmental monitoring parameter deviation modulus to generate an environmental fluctuation distance. On the contrary, when there is an environmental monitoring parameter attribute greater than the corresponding threshold in the environmental monitoring parameter deviation modulus, the system terminal directly uses the current environmental fluctuation distance threshold as the environmental fluctuation distance. After obtaining the environmental fluctuation distance, the system terminal performs a joint analysis on the environmental fluctuation distance and the environmental fluctuation distance threshold. If the environmental fluctuation distance between the first environmental monitoring parameter and the second environmental monitoring parameter exceeds the preset environmental fluctuation distance threshold, it means that the current environmental conditions have changed significantly. In this case, the system terminal uses the second environmental monitoring parameter to update the first environmental monitoring parameter to ensure that monitoring and early warning are always based on the most practical and accurate environmental information, thereby improving the accuracy of early warning.

[0020] Further, the present application provides a method for analyzing the environmental fluctuation distance of the second environmental monitoring parameter compared to the first environmental monitoring parameter, the method also includes: Comparing the first environmental monitoring parameter with the second environmental monitoring parameter one by one to obtain a deviation modulus of the environmental monitoring parameter; When the environmental monitoring parameter deviation modulus values ​​are all less than or equal to the deviation modulus value threshold associated with the environmental monitoring parameter attribute, the environmental monitoring parameter deviation modulus values ​​are normalized, and a square root operation is performed on the square sum of the normalized eigenvalues ​​to generate the environmental fluctuation distance; When any one of the environmental monitoring parameter deviation modulus values ​​is greater than the deviation modulus value threshold associated with the environmental monitoring parameter attribute, the environmental fluctuation distance is set equal to the environmental fluctuation distance threshold; The environmental monitoring parameter attributes include at least solar radiation intensity, ambient temperature, ambient humidity, ambient wind speed and ambient cloud cover.

[0021] Optionally, the system terminal compares each of the environmental monitoring parameter attributes in the first environmental monitoring parameter and the second environmental monitoring parameter one by one, and the environmental monitoring parameter attributes include solar radiation intensity, ambient temperature, ambient humidity, ambient wind speed, and ambient cloud cover, etc. Subsequently, the difference between the environmental monitoring parameter attributes in the second environmental monitoring parameter and the corresponding environmental monitoring parameter attributes in the first environmental monitoring parameter is calculated to obtain the environmental monitoring parameter deviation modulus. This environmental monitoring parameter deviation modulus represents the degree of difference between the corresponding attributes of the first environmental monitoring parameter and the second environmental monitoring parameter. It quantitatively reflects the changes or differences in the attributes of the two environmental monitoring data such as solar radiation intensity, ambient temperature, ambient humidity, ambient wind speed, and ambient cloud cover, which helps the system terminal to more accurately judge the changes in the environment. If all the environmental monitoring parameter deviation modulus values ​​are less than or equal to the deviation modulus value threshold associated with each attribute, the system terminal divides each deviation modulus value by the deviation modulus value threshold value corresponding to the attribute, so that all deviation modulus values ​​are mapped to the same numerical range, that is, mapped to between 0 and 1. Afterwards, all normalized environmental monitoring parameter deviation modulus values ​​are squared. These square values ​​are then added together to obtain the sum of the squares of the normalized eigenvalues. Then, the square root of the sum of the squares of the normalized eigenvalues ​​is taken to obtain a single value, namely the environmental fluctuation distance. This environmental fluctuation distance comprehensively reflects the degree of difference between the attributes of all environmental monitoring parameters in the first environmental monitoring parameter and the second environmental monitoring parameter, and is a comprehensive measurement indicator. However, if the deviation modulus of any environmental monitoring parameter exceeds the deviation modulus threshold associated with its attribute, this represents a significant change in the parameter. In this case, the system terminal directly sets the environmental fluctuation distance threshold to the environmental fluctuation distance, indicating that the environment has undergone significant fluctuations. In this way, the system terminal can accurately evaluate the changes in the environment in which the photovoltaic power station is located, and provide strong data support for the operation and maintenance of the power station.

[0022] According to the environmental monitoring parameters, the plurality of irradiation areas and the plurality of irradiation inclination angles are fitted one by one to perform power generation prediction, and generate a plurality of power generation per unit time; In one embodiment, in order to accurately predict the power generation, the system terminal comprehensively considers multiple factors, including environmental monitoring parameters and the irradiation area and irradiation inclination of the photovoltaic module. Specifically, before the power generation prediction is performed, the system terminal fits the relationship between these factors and the power generation through the Stacking integrated model based on the collected training data and the photovoltaic module model. This fitting process essentially uses the existing data to establish a predictor that can describe and predict the photovoltaic power generation, that is, the photovoltaic module power generation predictor. The system terminal constructs a photovoltaic module power generation predictor for each photovoltaic module model by performing the above steps multiple times. After obtaining several photovoltaic module power generation predictors, the system terminal inputs the current environmental monitoring parameters and the irradiation area and irradiation inclination of the photovoltaic module into the corresponding photovoltaic module power generation predictor for prediction. Each photovoltaic module power generation predictor will calculate the expected power generation in a given time period based on the input data and generate several unit time power generation. This unit time refers to the time interval used to predict the power generation, which is determined according to actual needs, for example, the expected power generation per hour, the expected power generation per day, etc. In this way, the system terminal can understand the power generation potential of the photovoltaic power station under different environmental conditions in advance, providing an important reference for the operation scheduling and maintenance plan of the power station. At the same time, this also helps to timely discover potential problems or failures, take measures to intervene in advance, and ensure the safe and efficient operation of the photovoltaic power station.

[0023] Furthermore, the present application provides a method for predicting power generation by fitting the plurality of irradiation areas and the plurality of irradiation inclination angles one by one according to the environmental monitoring parameters, and generating a plurality of power generation per unit time, and the method further includes: Among them, the steps of constructing any photovoltaic module power generation predictor include: Taking the photovoltaic module model as a constraint, collecting a first training data set, a second training data set and a third training data set; According to the first training data set, the second training data set and the third training data set, three parallel base predictors are trained respectively; Preferably, taking any photovoltaic module power generation predictor as an example, the system terminal uses the model of the photovoltaic module as a constraint condition and collects multiple training data sets of the model, namely the first training data set, the second training data set and the third training data set. These data sets may contain photovoltaic power generation data at different times and under different environmental conditions, as well as corresponding environmental monitoring parameters, such as solar radiation intensity, ambient temperature, etc., and information such as the irradiation area and irradiation inclination of the photovoltaic module. Subsequently, the system terminal initializes three base predictors, which are initialized based on decision trees, random forests, and neural networks. For the first base predictor, the system terminal sets the parameters of the decision tree according to the principle and actual needs of the algorithm, including the maximum depth of the tree, the minimum number of sample splits, etc., and uses the first training data set for training. After the training is completed, the model can predict the power generation according to the input environmental monitoring parameters, irradiation area and irradiation inclination. For the second base predictor, the system terminal sets the parameters of the random forest, including the number of trees, the depth of the tree, etc., and uses the second training data set for training. The model also has the ability to predict power generation. For the third base predictor, the system terminal sets the number of neural network layers, the number of neurons in each layer, the activation function, the optimizer and other parameters, and uses the third training data set for training. The neural network model can capture the complex nonlinear relationships in the data for accurate prediction. Through the above steps, the system terminal successfully obtains three parallel base predictors. Each base predictor uses the same features (environmental monitoring parameters, irradiation area and irradiation inclination) and labels (power generation), but has different algorithms to achieve power generation prediction from multiple angles.

[0024] constructing a fourth training data set by combining the output data of the three parallel base predictors with the supervision data of the first training data set, the second training data set and the third training data set; Training a fusion predictor based on the fourth training data set; Preferably, the system terminal uses the output data of the three base predictors in the training process as new features, and combines them with the supervisory data of the first training data set, the second training data set and the third training data set, that is, the real power generation data, to construct a new data set, that is, the fourth training data set. This data set not only contains the original feature information, but also contains the predicted power generation data generated by the base predictor, which provides richer information for the subsequent fusion predictor. Subsequently, the system terminal initializes the fusion predictor with a neural network, and determines the architecture of the neural network according to actual needs, including the number of layers, the number of neurons in each layer, the activation function, the loss function and the optimizer. After that, the neural network is trained using the prediction results of the base predictors in the fourth training data set and the real power generation data. During the training process, the neural network learns how to best combine the outputs of the three base predictors to produce the most accurate prediction results. The weights and bias terms of the neural network are then updated by the back propagation algorithm and gradient descent to minimize the prediction error. When the training reaches the preset maximum number of iterations, the system terminal uses the untrained data to evaluate the performance of the fusion predictor, and evaluates the prediction accuracy of the neural network by calculating performance indicators such as mean square error, mean absolute error, etc. Then, according to the evaluation results, the architecture, activation function, etc. of the fusion predictor are adjusted and optimized. Once the fusion predictor training is completed and reaches satisfactory performance, the system terminal outputs the current fusion predictor. Finally, the trained fusion predictor can combine the outputs of multiple base predictors to generate more accurate power generation prediction results. This integrated learning method can make full use of the advantages of different base predictors, more accurately predict the power generation of different types of photovoltaic modules under different environmental conditions, and provide strong support for subsequent fault warnings.

[0025] The output layers of the three parallel base predictors and the input layer of the fusion predictor are connected to generate the photovoltaic module power generation predictor.

[0026] Preferably, in the process of constructing the photovoltaic module power generation predictor, the system terminal directly connects the output layer of the trained base predictor to the input layer of the fusion predictor to form a complete integrated model, namely the photovoltaic module power generation predictor. The photovoltaic module power generation predictor can integrate the prediction results of multiple base predictors and provide a more accurate and comprehensive prediction result. This integrated learning method improves the performance and reliability of the prediction and provides strong support for the fault warning of photovoltaic modules.

[0027] Receiving the genotype factor state value and the photovoltaic power station distribution topology information, performing multi-source data retrieval in combination with the power generation per unit time, and obtaining the multi-source phenotype factor retrieval state value, wherein the genotype factor refers to the photovoltaic power station state that can be directly controlled, and the phenotype factor refers to the photovoltaic power station state that cannot be directly controlled but is indirectly affected by the genotype factor; In one embodiment, in the fault warning of photovoltaic power generation, the system terminal focuses on two types of key factors. One type is genotype factors, which are the states of photovoltaic power stations that can be directly controlled and adjusted, such as the switch state of the power station, the setting parameters of the inverter, etc.; the other type is phenotype factors, which cannot be directly controlled, but will be indirectly affected by genotype factors, such as the power generation and operating efficiency of the power station. In order to deeply understand and manage these factors, the system terminal performs multi-source data retrieval in combination with several unit time power generation. Specifically, the system terminal first defines a blank multi-source phenotype factor retrieval state value object. Subsequently, the retrieval responder is determined according to the genotype factor state value and the photovoltaic power station distribution topology information, which is used for the subsequent selection of information sources with higher confidence. After that, the first phenotype factor retrieval state value of the retrieval responder that meets the confidence threshold requirement is obtained and added to the defined blank multi-source phenotype factor retrieval state value. These state values ​​reflect the actual state of the phenotype factors of the photovoltaic power station and the possible fault risks under the current genotype factor settings. In this way, the operating state of the photovoltaic power station can be more comprehensively understood, and more reasonable fault warnings can be made based on this information.

[0028] Furthermore, the present application provides receiving genotype factor status values ​​and photovoltaic power station distribution topology information, performing multi-source data retrieval in combination with the power generation per unit time, and obtaining multi-source phenotype factor retrieval status values, the method also includes: Using the genotype factor state value and the photovoltaic power station distribution topology information as search restrictions, using the phenotype factor as a search target, publishing a multi-source search task online, and obtaining a search responder; Preferably, the system terminal first defines the search constraints, i.e., the genotype factor state value and the photovoltaic power station distribution topology information. These constraints will be used to accurately screen and locate the required data sources. Then determine the target of the search, i.e., the phenotype factor. The phenotype factor is the ultimate goal of the search and reflects the actual operation status of the power station. Subsequently, a clear search request is constructed based on the determined search constraints and targets. This request describes in detail the required data type, time range, geographic location and other specific information. This search request also includes details such as data format and transmission method to ensure the accuracy and efficient transmission of data. After that, the constructed search request is published to the networked data source and service provider platform. These platforms include public databases, power station monitoring systems, meteorological data providers, etc. After publishing the search task, wait for the response of the target data source or service provider platform. These responses include information such as data sets and data links that meet the search conditions. After receiving the response, the system terminal performs preliminary screening and verification on these responses. Check whether the response meets the constraints and target requirements in the search request, and then verify the accuracy and completeness of the data. If a response that does not meet the requirements or data errors are found, a request is sent to the responder based on the data for these issues, requiring more accurate or complete data. If the data provided again still does not meet the restrictions and target requirements or the data is still wrong, these data will be eliminated and the specific information of the wrong data will be stored in the exception database. After screening and verification, the system terminal will integrate the response data that meets the requirements and generate a retrieval responder. This includes format conversion, data cleaning, data fusion and other operations for data from different sources for subsequent analysis and application. The above method makes full use of the advantages of the Internet and big data technology, improves the efficiency and accuracy of information retrieval, and provides support for fault warning of power stations.

[0029] Evaluating the retrieval responder confidence of the retrieval responder, wherein the retrieval responder confidence refers to the accuracy of historical uploaded data; When the confidence of the retrieval responder is greater than or equal to the confidence threshold, the interactive retrieval responder is used to obtain the first phenotype factor retrieval status value and add it to the multi-source phenotype factor retrieval status value.

[0030] Preferably, the system terminal first sets a confidence threshold, which represents the minimum requirement for the accuracy of the data of the retrieval responder. This threshold is set according to actual needs and historical performance of data quality. Subsequently, the system terminal evaluates the confidence of each retrieval responder, which is obtained by analyzing the ratio of the number of erroneous data belonging to the retrieval responder in the abnormal database to the total amount of data provided by the retrieval responder, and then subtracting this ratio from 1. The difference is the data accuracy of the retrieval responder. If the data accuracy of a responder is high, then its confidence is high. After that, the system terminal compares the confidence of each responder with the set threshold. If the confidence of a responder is greater than or equal to the threshold, the system terminal determines that the data provided by the responder is reliable, and interacts with it to obtain the phenotypic factor retrieval status value provided by it. Then, the first phenotypic factor retrieval status value obtained from the responder that meets the confidence requirement is added to the pre-defined multi-source phenotypic factor retrieval status value. This process ensures that the data sources of the final integration are verified and reliable, thereby improving the overall quality and accuracy of the data. Through such an evaluation and selection process, high-quality data sources can be effectively screened out, and accurate retrieval status values ​​of phenotype factors can be obtained, providing more reliable and accurate data support for fault warning of photovoltaic power stations.

[0031] Fusion of the multi-source phenotypic factor retrieval state values ​​to generate phenotypic factor expected state values; In one embodiment, after obtaining the phenotypic factor retrieval state values ​​from multiple sources, the system terminal performs data fusion on the multi-source phenotypic factor retrieval state values ​​to generate a comprehensive and more reliable phenotypic factor expected state value. The main purpose of this process is to integrate the information from different data sources and convert multiple data points that may be different or complementary into a unified result with higher confidence. Specifically, the system terminal uses two methods, box plot analysis and local density clustering, to perform de-extreme analysis on the multi-source phenotypic factor retrieval state values ​​to obtain two phenotypic factor expected state values. Subsequently, the two phenotypic factor expected state values ​​obtained are averaged to generate the final phenotypic factor expected state value. This expected state value is a more accurate and reliable evaluation result, which can better reflect the actual operating status and future trends of the photovoltaic power station. Based on this expected state value, a more scientific fault warning can be made to improve the power generation efficiency and reliability of the photovoltaic power station. Therefore, fusing the multi-source phenotypic factor retrieval state value is a key step, which can integrate the information from multiple data sources into a unified result, improve the reliability and application value of the data, and provide more powerful support for the fault warning of the photovoltaic power station.

[0032] Furthermore, the present application provides a method for fusing the multi-source phenotypic factor retrieval status values ​​to generate the phenotypic factor expected status values, the method further comprising: Performing a box plot analysis on the multi-source phenotypic factor retrieval state values ​​to generate an expected state value of the first phenotypic factor; Performing local density clustering on the multi-source phenotypic factor retrieval state values ​​to generate an expected state value of a second phenotypic factor; The average of the first phenotypic factor expected state value and the second phenotypic factor expected state value is calculated and set as the phenotypic factor expected state value.

[0033] Preferably, the system terminal calculates the statistics of each phenotypic factor retrieval state value, including the minimum value, the first quartile, the median, the third quartile and the maximum value. These statistics are then used to draw a virtual box plot. The main part of the box plot is composed of the first quartile to the third quartile, representing the median and interquartile range of the data. A line is drawn above and below the box, representing the maximum value and the minimum value, respectively. Subsequently, the system terminal analyzes the drawn box plot to find possible outliers. An outlier is a value that is less than the product of the first quartile minus 1.5 and the interquartile range or greater than the product of the third quartile plus 1.5 and the interquartile range. If an outlier exists, the system terminal deletes the outlier. In the absence of an outlier or when the outlier has been deleted, the system terminal selects the median as the expected state value of the first phenotypic factor.

[0034] After obtaining the expected state value of the first phenotypic factor, the system terminal performs local density clustering on the multi-source phenotypic factor retrieval state values. Local density is the core concept in the local density clustering algorithm, which measures the density around the data point. The local density is determined by calculating the distance between each data point and its surrounding data points. In this process, the system terminal determines a distance threshold based on actual needs so that the local density of the data point can be accurately reflected. Subsequently, the system terminal uses the local density clustering algorithm to cluster the multi-source phenotypic factor retrieval state values. The local density clustering algorithm divides the data points into different clusters based on their local density. Each cluster represents a group of data points with similar local density. This local density is determined based on a determined distance threshold, that is, the number of neighboring points within the distance threshold around the point. Afterwards, the system terminal selects the largest cluster as the representative cluster, and determines the expected state value of the second phenotypic factor based on the distribution of this cluster. For example, if a skewed distribution is presented, the system terminal selects the median as the expected state value of the cluster, or if a normal distribution is presented, the mean value is selected as the expected state value of the cluster.

[0035] After obtaining the expected state value of the first phenotype factor and the expected state value of the second phenotype factor, the system terminal averages the expected state value of the first phenotype factor and the expected state value of the second phenotype factor to obtain the expected state value of the phenotype factor. This comprehensive value combines the expected states obtained by two different methods, more comprehensively reflects the overall situation of the phenotype factor, and is used for the subsequent generation of the deviation coefficient.

[0036] Based on the expected state value of the phenotypic factor, a deviation analysis is performed on the phenotypic factor state time series information to generate a phenotypic factor deviation coefficient, wherein the phenotypic factor state time series information and the environmental monitoring parameter have time series consistency; In one embodiment, the system terminal performs deviation analysis on the phenotypic factor state time series information based on the expected state value of the phenotypic factor, and the process of generating the deviation coefficient is a comprehensive data processing and evaluation process. In this process, the system terminal first creates a blank phenotypic factor deviation coefficient object, and then calculates the time proportion when the deviation distance between the phenotypic factor state time series information and the phenotypic factor expected state value is greater than or equal to the deviation distance threshold. This phenotypic factor state time series information has time series consistency with the environmental monitoring parameter, that is, whenever the environmental parameter is updated, the system terminal collects the phenotypic factor state information at the corresponding time point after the update. Subsequently, the maximum deviation distance between the phenotypic factor state time series information and the phenotypic factor expected state value is calculated. After that, the mean deviation distance between the phenotypic factor state time series information and the phenotypic factor expected state value is calculated. Then, the calculated moment proportion, the maximum deviation distance, and the mean deviation distance are added to the created blank phenotypic factor deviation coefficient. These deviation coefficients reflect the difference between the phenotypic factor state and the expected state, and provide the system terminal with an important reference for evaluating the degree of influence of environmental parameters on the phenotypic factor. In this way, we can better understand the impact mechanism of environmental parameters on the state of phenotypic factors and provide a scientific basis for photovoltaic power generation failure warning.

[0037] Furthermore, the present application provides a method for performing deviation analysis on the phenotypic factor state time series information based on the expected state value of the phenotypic factor to generate a phenotypic factor deviation coefficient, and the method further includes: Calculate the proportion of moments when the deviation distance between the phenotypic factor state time series information and the phenotypic factor expected state value is greater than or equal to the deviation distance threshold, and set it as a first deviation coefficient; Calculate the maximum value of the deviation distance between the phenotypic factor state time series information and the phenotypic factor expected state value, and set it as a second deviation coefficient; Calculate the mean of the deviation distance between the phenotypic factor state time series information and the phenotypic factor expected state value, and set it as the third deviation coefficient; The first deviation coefficient, the first deviation coefficient and the third deviation coefficient are added to the phenotypic factor deviation coefficient.

[0038] Preferably, the system terminal traverses the values ​​of each pair of corresponding moments in the phenotype factor state time series information and the corresponding phenotype factor expected state value. For each pair of values, the absolute value difference between them is calculated as the deviation distance. Subsequently, the calculated deviation distance list is traversed, and the number of deviation distances greater than or equal to the threshold is counted. The number greater than the threshold is divided by the total number of time points, i.e., the length of the phenotype factor state time series information, to obtain the first deviation coefficient. Afterwards, the calculated deviation distance list is arranged in descending order, and the first deviation distance is extracted as the second deviation coefficient. Then, the data in the calculated deviation distance list is traversed and accumulated, and then divided by the total number of data in the deviation distance list, and the deviation distance mean is calculated as the third deviation coefficient. Finally, the system terminal adds these three deviation coefficients into the pre-created phenotype factor deviation coefficient to comprehensively analyze and understand the deviation between the phenotype factor state and the expected state. This phenotype factor deviation coefficient can provide more abundant information, helping the system terminal to more accurately judge the nature, degree and influence of the deviation, so as to determine whether there is an abnormality in photovoltaic power generation.

[0039] If the phenotypic factor deviation coefficient is greater than or equal to the deviation coefficient threshold, a fault warning signal is generated.

[0040] In one embodiment, if any of the calculated deviation coefficients of the phenotype factors is greater than or equal to a preset deviation coefficient threshold, it means that the state of the phenotype factor has significantly deviated from the normal or expected range, then the system terminal considers that there is a potential problem or an impending failure in the photovoltaic power generation, and generates a fault warning signal to remind relevant personnel to pay attention and take corresponding measures. This early warning mechanism helps to detect potential problems in a timely manner, avoid further expansion of the failure, and ensure the stable operation of photovoltaic power generation.

[0041] In summary, the embodiments of the present application have at least the following technical effects: The embodiment of the present application obtains the irradiation area and irradiation inclination of multiple photovoltaic modules through the photovoltaic array, and receives environmental monitoring parameters through the meteorological monitoring port, including environmental monitoring parameters stored in the original and real-time storage. Then analyze the environmental fluctuation distance. If the distance is greater than the set threshold, the environmental monitoring parameters are updated to ensure accuracy. Subsequently, the irradiation area and irradiation inclination of the photovoltaic module are fitted according to the environmental monitoring parameters, and the power generation is predicted to generate the power generation per unit time. This process takes into account the model of the photovoltaic module, and constructs a fusion predictor through the training data set and the base predictor to achieve accurate power generation prediction. After that, the genotype factor state value and the photovoltaic power station distribution topology information are received, and multi-source data retrieval is performed in combination with the power generation per unit time to obtain the multi-source phenotype factor retrieval state value. This step involves publishing a multi-source retrieval task online, and evaluating the confidence of the retrieval responder to ensure data accuracy. Then, the multi-source phenotype factor retrieval state value is fused to generate the expected state value of the phenotype factor. This process calculates the expected state value of the phenotype factor through box plot analysis and local density clustering to reflect the overall state of the photovoltaic power station. Finally, based on the expected state value of the phenotype factor, the deviation analysis of the phenotype factor state time series information is performed to generate the phenotype factor deviation coefficient. The degree of deviation of the phenotype factor state is then evaluated. If the deviation coefficient is greater than the set threshold, a fault warning signal is generated. These technical effects jointly solve the technical problem that the traditional photovoltaic power generation fault analysis relies on preset threshold values ​​for analysis, has poor scene integration, and leads to low warning accuracy. It achieves the effect of improving the accuracy of fault warning and scene adaptability through multi-source data fusion and deviation analysis.

[0042] Embodiment 2 Based on the same inventive concept as the photovoltaic power generation fault early warning method of multi-source data fusion in the above-mentioned embodiment, Figure 2 As shown, the present application provides a photovoltaic power generation fault early warning system with multi-source data fusion, the system comprising: Data acquisition module 1: the data acquisition module 1 is used to obtain a number of irradiation areas and a number of irradiation inclinations of a number of photovoltaic modules through the photovoltaic array, and to receive environmental monitoring parameters through the meteorological monitoring port; Power generation prediction module 2: the power generation prediction module 2 is used to perform power generation prediction by fitting the plurality of irradiation areas and the plurality of irradiation inclination angles one by one according to the environmental monitoring parameters, and generate a plurality of power generation per unit time; Multi-source data retrieval module 3: The multi-source data retrieval module 3 is used to receive the genotype factor state value and the photovoltaic power station distribution topology information, and perform multi-source data retrieval in combination with the power generation per unit time to obtain the multi-source phenotype factor retrieval state value, wherein the genotype factor refers to the photovoltaic power station state that can be directly controlled, and the phenotype factor refers to the photovoltaic power station state that cannot be directly controlled but is indirectly affected by the genotype factor; State value fusion module 4: the state value fusion module 4 is used to fuse the multi-source phenotype factor retrieval state values ​​to generate the phenotype factor expected state value; Deviation analysis module 5: the deviation analysis module 5 is used to perform deviation analysis on the phenotypic factor state time series information based on the expected state value of the phenotypic factor, and generate a phenotypic factor deviation coefficient, wherein the phenotypic factor state time series information and the environmental monitoring parameter have time series consistency; Coefficient judgment module 6: The coefficient judgment module 6 is used to generate a fault warning signal if the deviation coefficient of the phenotypic factor is greater than or equal to the deviation coefficient threshold.

[0043] Furthermore, the data acquisition module 1 is also used to execute the following method: Obtaining a first environmental monitoring parameter, wherein the first environmental monitoring parameter is an original stored environmental monitoring parameter; Obtaining a second environmental monitoring parameter, wherein the second environmental monitoring parameter is a real-time storage environmental monitoring parameter; Analyzing the environmental fluctuation distance of the second environmental monitoring parameter compared with the first environmental monitoring parameter; When the environmental fluctuation distance is greater than or equal to an environmental fluctuation distance threshold, the first environmental monitoring parameter is updated by the second environmental monitoring parameter.

[0044] Furthermore, the data acquisition module 1 is also used to execute the following method: Comparing the first environmental monitoring parameter with the second environmental monitoring parameter one by one to obtain a deviation modulus of the environmental monitoring parameter; When the environmental monitoring parameter deviation modulus values ​​are all less than or equal to the deviation modulus value threshold associated with the environmental monitoring parameter attribute, the environmental monitoring parameter deviation modulus values ​​are normalized, and a square root operation is performed on the square sum of the normalized eigenvalues ​​to generate the environmental fluctuation distance; When any one of the environmental monitoring parameter deviation modulus values ​​is greater than the deviation modulus value threshold associated with the environmental monitoring parameter attribute, the environmental fluctuation distance is set equal to the environmental fluctuation distance threshold; The environmental monitoring parameter attributes include at least solar radiation intensity, ambient temperature, ambient humidity, ambient wind speed and ambient cloud cover.

[0045] Furthermore, the power generation prediction module 2 is also used to execute the following method: According to the models of several photovoltaic modules, several photovoltaic module power generation predictors are activated, and the several irradiation areas and the several irradiation inclination angles are fitted one by one according to the environmental monitoring parameters to generate the several unit time power generation; Among them, the steps of constructing any photovoltaic module power generation predictor include: Taking the photovoltaic module model as a constraint, collecting a first training data set, a second training data set and a third training data set; According to the first training data set, the second training data set and the third training data set, three parallel base predictors are trained respectively; constructing a fourth training data set by combining the output data of the three parallel base predictors with the supervision data of the first training data set, the second training data set and the third training data set; Training a fusion predictor based on the fourth training data set; The output layers of the three parallel base predictors and the input layer of the fusion predictor are connected to generate the photovoltaic module power generation predictor.

[0046] Furthermore, the multi-source data retrieval module 3 is also used to execute the following method: Using the genotype factor state value and the photovoltaic power station distribution topology information as search restrictions, using the phenotype factor as a search target, publishing a multi-source search task online, and obtaining a search responder; Evaluating the retrieval responder confidence of the retrieval responder, wherein the retrieval responder confidence refers to the accuracy of historical uploaded data; When the confidence of the retrieval responder is greater than or equal to the confidence threshold, the interactive retrieval responder is used to obtain the first phenotype factor retrieval status value and add it to the multi-source phenotype factor retrieval status value.

[0047] Furthermore, the state value fusion module 4 is also used to execute the following method: Performing a box plot analysis on the multi-source phenotypic factor retrieval state values ​​to generate an expected state value of the first phenotypic factor; Performing local density clustering on the multi-source phenotypic factor retrieval state values ​​to generate an expected state value of a second phenotypic factor; The average of the first phenotypic factor expected state value and the second phenotypic factor expected state value is calculated and set as the phenotypic factor expected state value.

[0048] Furthermore, the deviation analysis module 5 is also used to perform the following method: Calculate the proportion of moments when the deviation distance between the phenotypic factor state time series information and the phenotypic factor expected state value is greater than or equal to the deviation distance threshold, and set it as a first deviation coefficient; Calculate the maximum value of the deviation distance between the phenotypic factor state time series information and the phenotypic factor expected state value, and set it as a second deviation coefficient; Calculate the mean of the deviation distance between the phenotypic factor state time series information and the phenotypic factor expected state value, and set it as the third deviation coefficient; The first deviation coefficient, the first deviation coefficient and the third deviation coefficient are added to the phenotypic factor deviation coefficient.

[0049] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description and does not represent the advantages and disadvantages of the embodiments. And the above-mentioned specific embodiments of this specification are described. The processes depicted in the accompanying drawings do not necessarily require the specific order and continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0050] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

[0051] This specification and drawings are merely exemplary illustrations of the present application and are deemed to cover any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, a person skilled in the art may make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application intends to include these modifications and variations.

Claims

1. A photovoltaic power generation fault early warning method based on multi-source data fusion, characterized in that: include: Obtaining a number of irradiation areas and a number of irradiation inclination angles of a number of photovoltaic modules through a photovoltaic array, and receiving environmental monitoring parameters through a meteorological monitoring port; According to the environmental monitoring parameters, the plurality of irradiation areas and the plurality of irradiation inclination angles are fitted one by one to perform power generation prediction, and generate a plurality of power generation per unit time; Receiving the genotype factor state value and the photovoltaic power station distribution topology information, performing multi-source data retrieval in combination with the power generation per unit time, and obtaining the multi-source phenotype factor retrieval state value, wherein the genotype factor refers to the photovoltaic power station state that can be directly controlled, and the phenotype factor refers to the photovoltaic power station state that cannot be directly controlled but is indirectly affected by the genotype factor; Fusion of the multi-source phenotypic factor retrieval state values ​​to generate phenotypic factor expected state values; Based on the expected state value of the phenotypic factor, a deviation analysis is performed on the phenotypic factor state time series information to generate a phenotypic factor deviation coefficient, wherein the phenotypic factor state time series information and the environmental monitoring parameter have time series consistency; If the phenotypic factor deviation coefficient is greater than or equal to the deviation coefficient threshold, a fault warning signal is generated.

2. The method according to claim 1, characterized in that The irradiation areas and irradiation angles of several PV modules are obtained through the PV array, and the environmental monitoring parameters are received through the meteorological monitoring port, including: Obtaining a first environmental monitoring parameter, wherein the first environmental monitoring parameter is an original stored environmental monitoring parameter; Obtaining a second environmental monitoring parameter, wherein the second environmental monitoring parameter is a real-time storage environmental monitoring parameter; Analyzing the environmental fluctuation distance of the second environmental monitoring parameter compared with the first environmental monitoring parameter; When the environmental fluctuation distance is greater than or equal to an environmental fluctuation distance threshold, the first environmental monitoring parameter is updated by the second environmental monitoring parameter.

3. The method according to claim 2, characterized in that Analyzing the environmental fluctuation distance of the second environmental monitoring parameter compared with the first environmental monitoring parameter includes: Comparing the first environmental monitoring parameter with the second environmental monitoring parameter one by one to obtain a deviation modulus of the environmental monitoring parameter; When the environmental monitoring parameter deviation modulus values ​​are all less than or equal to the deviation modulus value threshold associated with the environmental monitoring parameter attribute, the environmental monitoring parameter deviation modulus values ​​are normalized, and a square root operation is performed on the square sum of the normalized eigenvalues ​​to generate the environmental fluctuation distance; When any one of the environmental monitoring parameter deviation modulus values ​​is greater than the deviation modulus value threshold associated with the environmental monitoring parameter attribute, the environmental fluctuation distance is set equal to the environmental fluctuation distance threshold; The environmental monitoring parameter attributes include at least solar radiation intensity, ambient temperature, ambient humidity, ambient wind speed and ambient cloud cover.

4. The method according to claim 1, characterized in that According to the environmental monitoring parameters, the plurality of irradiation areas and the plurality of irradiation inclination angles are fitted one by one to perform power generation prediction, and generate a plurality of unit time power generation, including: According to the models of several photovoltaic modules, several photovoltaic module power generation predictors are activated, and the several irradiation areas and the several irradiation inclination angles are fitted one by one according to the environmental monitoring parameters to generate the several unit time power generation; Among them, the steps of constructing any photovoltaic module power generation predictor include: Taking the photovoltaic module model as a constraint, collecting a first training data set, a second training data set and a third training data set; According to the first training data set, the second training data set and the third training data set, three parallel base predictors are trained respectively; constructing a fourth training data set by combining the output data of the three parallel base predictors with the supervision data of the first training data set, the second training data set and the third training data set; Training a fusion predictor based on the fourth training data set; The output layers of the three parallel base predictors and the input layer of the fusion predictor are connected to generate the photovoltaic module power generation predictor.

5. The method according to claim 1, characterized in that Receiving the genotype factor state value and the photovoltaic power station distribution topology information, performing multi-source data retrieval in combination with the power generation per unit time, and obtaining the multi-source phenotype factor retrieval state value, including: Using the genotype factor state value and the photovoltaic power station distribution topology information as search restrictions, using the phenotype factor as a search target, publishing a multi-source search task online, and obtaining a search responder; Evaluating the retrieval responder confidence of the retrieval responder, wherein the retrieval responder confidence refers to the accuracy of historical uploaded data; When the confidence of the retrieval responder is greater than or equal to the confidence threshold, the interactive retrieval responder is used to obtain the first phenotype factor retrieval status value and add it to the multi-source phenotype factor retrieval status value.

6. The method according to claim 1, characterized in that Fusion of the multi-source phenotypic factor retrieval status values ​​to generate phenotypic factor expected status values ​​includes: Performing box plot analysis on the multi-source phenotypic factor retrieval state values ​​to generate an expected state value of the first phenotypic factor; Performing local density clustering on the multi-source phenotypic factor retrieval state values ​​to generate an expected state value of a second phenotypic factor; The average of the first phenotypic factor expected state value and the second phenotypic factor expected state value is calculated and set as the phenotypic factor expected state value.

7. The method according to claim 1, characterized in that Based on the expected state value of the phenotypic factor, a deviation analysis is performed on the phenotypic factor state time series information to generate a phenotypic factor deviation coefficient, including: Calculate the proportion of moments when the deviation distance between the phenotypic factor state time series information and the phenotypic factor expected state value is greater than or equal to the deviation distance threshold, and set it as a first deviation coefficient; Calculate the maximum value of the deviation distance between the phenotypic factor state time series information and the phenotypic factor expected state value, and set it as a second deviation coefficient; Calculate the mean of the deviation distance between the phenotypic factor state time series information and the phenotypic factor expected state value, and set it as the third deviation coefficient; The first deviation coefficient, the first deviation coefficient and the third deviation coefficient are added to the phenotypic factor deviation coefficient.

8. A photovoltaic power generation fault early warning system with multi-source data fusion, characterized in that: The steps for implementing the photovoltaic power generation fault early warning method of multi-source data fusion as described in any one of claims 1 to 7 include: Data acquisition module: obtains several irradiation areas and several irradiation inclination angles of several photovoltaic modules through the photovoltaic array, and receives environmental monitoring parameters through the meteorological monitoring port; Power generation prediction module: according to the environmental monitoring parameters, the plurality of irradiation areas and the plurality of irradiation inclination angles are fitted one by one to perform power generation prediction, and generate a plurality of unit time power generation; Multi-source data retrieval module: receiving genotype factor state values ​​and photovoltaic power station distribution topology information, performing multi-source data retrieval in combination with the power generation per unit time, and obtaining multi-source phenotype factor retrieval state values, wherein the genotype factor refers to the photovoltaic power station state that can be directly controlled, and the phenotype factor refers to the photovoltaic power station state that cannot be directly controlled but is indirectly affected by the genotype factor; State value fusion module: fuses the multi-source phenotypic factor retrieval state values ​​to generate the expected state value of the phenotypic factor; Deviation analysis module: based on the expected state value of the phenotypic factor, the phenotypic factor state time series information is subjected to deviation analysis to generate a phenotypic factor deviation coefficient, wherein the phenotypic factor state time series information and the environmental monitoring parameter have time series consistency; Coefficient judgment module: if the deviation coefficient of the phenotypic factor is greater than or equal to the deviation coefficient threshold, a fault warning signal is generated.

Images