New energy power station operation and maintenance management system

Abstract

The invention discloses a new energy power station operation and maintenance management system, relates to the technical field of new energy operation and maintenance, and solves the technical field that collected data cannot be fully utilized to mine operation and maintenance related problems in the prior art. The system comprises a data acquisition subsystem which is used for acquiring operation data of a power generation assembly and storing the operation data in a database; the data analysis subsystem is used for processing the operation data to obtain a first-class feature value and a second-class feature value; the fault diagnosis subsystem is used for acquiring a first-class prediction result and a second-class prediction result; the strategy generation subsystem is used for determining a processing strategy according to the working state and issuing the working state and the corresponding processing strategy to the user side; the user side is used for receiving operation data and processing strategies of the power generation assembly in real time; according to the invention, operation and maintenance personnel are timely reminded of maintenance when the power generation assembly breaks down, and replacement is timely reminded when the data acquisition subsystem breaks down, so that the situation that the corresponding situation cannot be obtained when the power generation assembly breaks down due to untimely data acquisition is avoided.

Description

Technical Field

[0001] The present invention relates to the field of new energy operation and maintenance technology, and in particular to a new energy power station operation and maintenance management system. Background Art

[0002] In recent years, in order to reduce carbon emissions, new energy has been vigorously developed, including photovoltaics, wind power, hydropower, etc. In order to increase the proportion of new energy power supply, the research on new energy power stations has also received attention from all parties. The role of new energy power stations is mainly to convert the collected electric energy into electricity and transmit it to the power supply grid, but the key to realizing this function is to ensure the stability of new energy power generation.

[0003] The stability of renewable energy power generation is partly determined by the ability of power generation components to convert electrical energy. Generally, power generation components are located in uninhabited areas, including deserts, plateaus, etc., and large-scale power generation components generally occupy a large area. If manual patrols are required, it will require a lot of manpower and material resources. If a power generation component fails and is not repaired in time, it may cause failures of other normal components or cause serious consequences such as fire. Therefore, real-time operation and maintenance management of new energy power stations is a mainstream direction of current research.

[0004] In this regard, existing technologies have formed a complete intelligent + manual operation and maintenance model through real-time data collection combined with artificial intelligence and manual assisted processing, which has achieved faster and better technical effects on the maintenance of new energy power stations. However, most current research focuses on real-time monitoring of data from new energy power stations and early warning when problems are found. Although this helps operation and maintenance personnel to find problems in a timely manner, it lacks a comprehensive analysis of the problems and the designation of corresponding processing strategies based on the problems to guide operation and maintenance personnel to handle them in a timely manner, which means that it cannot help operation and maintenance personnel solve problems quickly and effectively. Summary of the invention

[0005] In order to solve the problems existing in the above-mentioned prior art, the present invention provides a new energy power station operation and maintenance management system to solve the problem that the existing new energy power station operation and maintenance technology cannot help operation and maintenance personnel solve problems quickly and effectively.

[0006] A new energy power station operation and maintenance management system, including a data acquisition subsystem, a data analysis subsystem, a fault diagnosis subsystem, a strategy generation subsystem and a user terminal,

[0007] The data acquisition subsystem is used to collect the operating data of the power generation component and save it in the database, the operating data includes a plurality of data items, each data item includes the photovoltaic array serial number and the corresponding operating information, and the data acquisition subsystem includes a plurality of data acquisition units;

[0008] The data analysis subsystem is used to process the operation data to obtain the first type of eigenvalues ​​and the second type of eigenvalues;

[0009] The fault diagnosis subsystem includes a first diagnosis module and a second diagnosis module, wherein the first diagnosis module determines the fault condition of the data acquisition subsystem according to the first type of characteristic value as a first type of prediction result, and the second diagnosis module determines the fault condition of the photovoltaic array according to the second type of characteristic value as a second type of prediction result;

[0010] The strategy generation subsystem is used to determine the processing strategy according to the working status and send the working status and the corresponding processing strategy to the user end;

[0011] The user end is used to receive the operation data and processing strategy of the power generation component in real time.

[0012] Furthermore, the data analysis subsystem includes a preprocessing module and a statistical analysis module. The preprocessing module divides the operating data into an abnormal data set and a normal data set. The statistical analysis module is used to extract a type of feature value from the abnormal data set and a type of feature value from the normal data set.

[0013] Furthermore, the preprocessing module first determines whether there are any data items with abnormal data. If so, the relevant data items are placed in the abnormal data set until there is no abnormal data and the remaining data items are used as the normal data set. The process of the preprocessing module determining abnormal data includes: first screening out duplicate data, missing data and dead data in the running information, and then using the 3-sigma method and the quartile method to screen out abnormal range data.

[0014] Furthermore, the operation information includes environmental data and power generation data, and the statistical analysis module processes the abnormal data set including:

[0015] If the abnormal data only exists in the environmental data, the environmental data in the data item of the normal photovoltaic array with the adjacent serial number is used to replace the abnormal data to realize the normalization of the abnormal data, and the normalized data item is put into the normal data set.

[0016] Further, the statistical analysis module processes the abnormal data set including: whether similar abnormal data appears in data items within adjacent acquisition time of the data items corresponding to the statistical abnormal data; if so, it is preliminarily determined that the abnormal data is related to the data acquisition subsystem and is caused abnormal data, and the data item is retained; if similar abnormal data does not appear in the data items within adjacent acquisition time, it is preliminarily determined that the abnormal data is not related to the data acquisition subsystem and is uncaused abnormal data, and the data item is eliminated; after the uncaused abnormal data is eliminated, a caused abnormal data set is obtained, and the characteristic value of each data item in the caused abnormal data set is extracted as a type of characteristic value, and the type of characteristic value includes the abnormal data type, the abnormal data position, and the photovoltaic array serial number corresponding to the data item.

[0017] Furthermore, the characteristic value of each data item in the normal data set is extracted and processed as a second type of characteristic value, and the second type of characteristic value includes: volt-ampere characteristic curve parameters and output voltage, output current, temperature, and light intensity. The volt-ampere characteristic curve includes short-circuit current, open-circuit voltage, and current and voltage corresponding to the maximum power point.

[0018] Furthermore, the one type of prediction results includes the location of the faulty data acquisition unit, the fault location in the data acquisition unit and the specific fault type; wherein, the faulty data acquisition unit is determined according to the photovoltaic array serial number in the data item, the faulty sensor is determined according to the abnormal data location, and then the corresponding sensor fault type is determined according to the abnormal data type. If it is missing data, the corresponding sensor is determined to be invalid; if it is abnormal range data, the corresponding sensor is determined to be faulty; if it is dead data, that is, the data continues to remain unchanged, it is determined that the corresponding information upload function is faulty.

[0019] Furthermore, for the data items determined as normal by the second diagnostic module, the power of the photovoltaic array under the standard state is calculated in combination with the temperature and the radiation intensity. The specific calculation formula is:

[0020] P = GτηA[1-0.045(T c -25)]

[0021] Where G is the radiation intensity, W / m 2 ; τ is the outer layer projection rate of the photovoltaic cell; η is the conversion efficiency of the photovoltaic cell; A is the receiving surface area of ​​the photovoltaic module; T c is the operating temperature of the photovoltaic cell;

[0022] If the difference between the calculated power value and the measured power value is too large, it is judged that the photovoltaic array corresponding to the data item may have too much dust accumulation, and the normal label corresponding to the data item is changed to a dust label; the measured power value is the average power value within a certain range of temperature and radiation intensity changes.

[0023] Furthermore, the strategy generation subsystem includes an expert experience base, a historical database and a strategy MAP table. The strategy MAP table is obtained through deep learning training based on the expert experience base and the historical database. The expert experience base and the historical database are updated in real time. The strategy MAP is regularly iterated and trained based on the updated data of the expert experience base and the historical database.

[0024] Furthermore, the data acquisition subsystem also includes a drone thermal imager, and the strategy generation subsystem also includes an identification module. The identification module determines the faulty photovoltaic array based on the second type of prediction results, controls the drone thermal imager to collect thermal infrared images of the designated photovoltaic array based on the location information of the faulty photovoltaic array, and verifies the fault condition of the photovoltaic array based on the collected thermal infrared images; if the accuracy of the second type of prediction results is verified to reach a minimum threshold for multiple times in a row, the second type of prediction results will no longer be verified by the drone thermal imager. At the same time, during the maintenance process, the operation and maintenance personnel will feed back the fault information of the second type of prediction results to the identification module. If it is found through the feedback information that the accuracy of the second type of prediction results is lower than the minimum threshold for multiple times in a row, the operation and maintenance personnel will be prompted to readjust the model and re-enable the drone thermal imager to verify the second type of prediction results.

[0025] The beneficial effects of the present invention include:

[0026] 1. The collected operation data is divided into abnormal data sets and normal data sets for full utilization, including using the data in the normal data set to analyze and predict the failure of the power generation components, and using the data in the abnormal data set to analyze and predict the failure of the data acquisition subsystem. This not only promptly reminds the operation and maintenance personnel to repair the power generation components when they fail, but also promptly reminds them to replace the data acquisition subsystem when it fails, avoiding the inability to obtain the corresponding situation when the power generation components fail due to untimely data collection.

[0027] 2. Combined with UAV thermal imagers, the accuracy of fault prediction of the system is ensured to avoid incorrect policy issuance and waste of manpower and material resources. Taking into account the use cost and depreciation cost of UAV thermal imagers, the use time of UAV thermal imagers is limited. It is only used in the early stage of system investment or when the system investment time is too long and the prediction accuracy cannot be guaranteed, so as to achieve a balance between accuracy and operating costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 The figure is a structural diagram of a new energy power station operation and maintenance management system. DETAILED DESCRIPTION

[0029] In order to make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings 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. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present application.

[0030] A new energy power station operation and maintenance management system, including a data acquisition subsystem, a data analysis subsystem, a fault diagnosis subsystem, a strategy generation subsystem and a user terminal,

[0031] The data acquisition subsystem is used to collect the operating data of the power generation component and save it in the database; the data acquisition subsystem includes several data acquisition modules and a UAV thermal imager, and the data acquisition module includes a temperature sensor, a light intensity sensor, a current sensor, a voltage sensor, and a power sensor.

[0032] The data analysis subsystem is used to process the operation data to obtain the first type of eigenvalues ​​and the second type of eigenvalues;

[0033] The fault diagnosis subsystem includes a first diagnosis module and a second diagnosis module. The first diagnosis module determines the fault condition of the data acquisition subsystem according to a first type of characteristic value as a first type of prediction result, and the second diagnosis module determines the fault condition of the photovoltaic array according to a second type of characteristic value as a second type of prediction result.

[0034] The strategy generation subsystem is used to determine the processing strategy according to the working status and send the working status and the corresponding processing strategy to the user end;

[0035] The user end is used to receive the working status and corresponding processing strategy in real time, and can call the interface to obtain the environmental information and power generation information of the power generation component.

[0036] The data analysis subsystem includes a preprocessing module and a statistical analysis module. The preprocessing module divides the operating data into an abnormal data set and a normal data set. The statistical analysis module is used to extract a type of feature value from the abnormal data set and a type of feature value from the normal data set; specifically.

[0037] The operation data includes several data items, each of which includes the PV array serial number and the corresponding operation information. The preprocessing module first determines whether there are data items with abnormal data. If there are, the relevant data items are put into the abnormal data set until there is no abnormal data and the remaining data items are used as normal data sets.

[0038] The process of determining abnormal data by the preprocessing module includes: first screening out duplicate data, missing data and dead data in the operation information, and then screening out abnormal range data using the 3-sigma method and the quartile method.

[0039] The 3-sigma method is also known as the empirical rule. This rule states that the values ​​of the measured physical quantity are mostly distributed within the range of plus or minus 3 standard deviations (±3s) of their mean value (i.e., they obey the normal distribution). The probability of data values ​​exceeding this range is extremely low and they are considered abnormal data. However, when the data does not follow the normal distribution, the effect of identifying abnormal data deteriorates significantly.

[0040] The quartile method, also known as the box plot method, is a commonly used method for identifying outliers in statistics. Compared with the 3-sigma method, the calculation of the quartile method relies on actual data and does not require the assumption that the data follows a specific distribution form.

[0041] Arrange the data in ascending (or descending) order, and the formula for calculating the interquartile range IQR is:

[0042] IQR=Q3-Q1

[0043] Among them, Q1 is the first quartile and Q3 is the third quartile;

[0044] Compared with the mean and standard deviation, the calculation of quartiles is less affected by the outliers themselves and can better detect outliers.

[0045] The operation information includes environmental data and power generation data. The statistical analysis module processes the abnormal data set including:

[0046] 1. If there are abnormal data in the environmental data, the temperature and light data in the data items of the normal photovoltaic array with adjacent serial numbers will replace the abnormal data in the data item to normalize the abnormal data, and the normalized data items will be placed in the normal data set; if there are abnormal data in the power generation data, no processing will be performed.

[0047] The environmental information includes radiation intensity and temperature; when the environmental information corresponding to a certain data is different from the environmental information of other data or the difference with the real-time weather forecast information is too large, it is determined that the environmental information collection instrument is faulty.

[0048] 2. Check whether similar abnormal data appears in the data items within the adjacent collection time of the data items corresponding to the statistical abnormal data. If so, it is preliminarily determined that the abnormal data is related to the data collection subsystem and is caused abnormal data, and the data item is retained. If similar abnormal data does not appear in the data items within the adjacent collection time, it is preliminarily determined that the abnormal data is not related to the data collection subsystem and is uncaused abnormal data, and the data item is eliminated. After the uncaused abnormal data is eliminated, the caused abnormal data set is obtained.

[0049] Extracting a characteristic value of each data item in the abnormal data set as a type of characteristic value, wherein the type of characteristic value includes the abnormal data type, the abnormal data location, and the photovoltaic array serial number corresponding to the data item;

[0050] The characteristic value of each data item in the normal data set is extracted and processed as the second type of characteristic value, and the second type of characteristic value includes: volt-ampere characteristic curve parameters and output voltage, output current, temperature, and light intensity. The volt-ampere characteristic curve includes short-circuit current, open-circuit voltage, and current and voltage corresponding to the maximum power point.

[0051] The first diagnosis module determines the fault condition of the data acquisition subsystem according to a type of characteristic value as a type of prediction result

[0052] The first type of prediction results includes the location of the data acquisition unit where the fault occurs, the location of the fault in the data acquisition unit, and the specific fault type; specifically, the data acquisition unit where the fault occurs is determined based on the PV array serial number in the data item, the faulty sensor is determined based on the location of the abnormal data, and then the corresponding sensor fault type is determined based on the type of abnormal data. If it is missing data, the corresponding sensor is determined to be invalid, if it is abnormal range data, the corresponding sensor is determined to be faulty, and if it is dead data, that is, the data continues to remain unchanged, the corresponding information upload function is determined to be faulty. During the specific implementation process, if the operation and maintenance personnel find that the sensor or data acquisition and transmission function is working normally when repairing according to the first type of prediction results, they can further detect whether the abnormal data is caused by a fault in the corresponding PV array.

[0053] The second diagnosis module predicts the fault condition of the photovoltaic array according to the second type of characteristic values ​​as a second type of prediction result;

[0054] During the specific implementation process, if the operation and maintenance personnel find that the PV array is working normally during maintenance based on the second-class prediction results, they can further detect whether the sensor or data acquisition and transmission function causes the deviation between the uploaded data and the actual data.

[0055] Specifically, the second diagnostic module includes a recognition model, and the recognition model is obtained by training and testing a neural network model using a deep learning method combined with a specific data set.

[0056] The data set preparation method includes: determining the digital model of the photovoltaic array according to its parameters, and building a corresponding simulation model in the MATLAB / Simulink environment, and then simulating the volt-ampere characteristic curves of the photovoltaic array under five working conditions of short circuit, open circuit, aging, shadow and normal under multiple temperatures and light intensities, and preparing the data set based on the volt-ampere characteristic curves and the corresponding working conditions.

[0057] Specifically, when a short circuit occurs in the photovoltaic array, there will be multiple local maximum power points in the volt-ampere characteristic curve. When an open circuit occurs in the photovoltaic array, the output current of the array is zero. When the photovoltaic array ages, the equivalent series resistance of the array increases, causing its volt-ampere characteristic curve to change. When a shadow occurs locally in the photovoltaic array, the global maximum power point will change.

[0058] The neural network model adopts an improved TCN-BiGRU model; first, by stacking multiple layers of TCN residual modules to obtain a wider range of input sequence receptive fields, while avoiding problems such as gradient explosion and gradient disappearance, each residual block has the same kernel size k, and its expansion factors D are 1, 2, and 4 respectively. Secondly, BiGRU obtains the data sequence processed by TCN, and it connects the GRU layers in the forward and reverse directions, one processes the input sequence from front to back, and the other processes from back to front. In this way, BiGRU can explore the dependency of features more completely and obtain contextual associations. Finally, a single-head attention mechanism is added with a key value of 3, and the high-dimensional features are mapped to the final prediction results through the fully connected layer, softmax layer, and classification layer. The specific optimization algorithm adopts the particle swarm algorithm.

[0059] For the data items that are determined to be normal by the identification model, the power of the photovoltaic array under standard conditions is calculated by combining the temperature and radiation intensity. The specific calculation formula is:

[0060] P = GτηA[1-0.045(T c -25)]

[0061] Where G is the radiation intensity, W / m 2 ; τ is the outer layer projection rate of the photovoltaic cell; η is the conversion efficiency of the photovoltaic cell; A is the receiving surface area of ​​the photovoltaic module; T c is the operating temperature of the photovoltaic cell.

[0062] If the difference between the calculated power value and the measured power value is too large, it is judged that the photovoltaic array corresponding to the data item may have too much dust accumulation, and the normal label corresponding to the data item is changed to a dust label. Among them, the measured power value is the average power value within a certain range of temperature and radiation intensity changes.

[0063] The strategy generation subsystem includes an expert experience base, a historical database and a strategy MAP table. The strategy MAP table is obtained through deep learning training based on the expert experience base and the historical database. The expert experience base and the historical database are updated in real time. The strategy MAP is regularly iterated and trained based on the updated data of the expert experience base and the historical database.

[0064] The strategy generation subsystem also includes an identification module, which determines the faulty photovoltaic array based on the second type of prediction results output by the second diagnosis module, controls the drone thermal imager to collect thermal infrared images of the specified photovoltaic array based on the location information of the faulty photovoltaic array, and verifies the fault condition of the photovoltaic array based on the collected thermal infrared images, thereby ensuring the accuracy of subsequent strategy issuance.

[0065] If the accuracy of the second-category prediction results is verified to reach the minimum threshold for multiple times in a row, the second-category prediction results will no longer be verified through the drone thermal imager. At the same time, the operation and maintenance personnel will feed back the fault information of the second-category prediction results to the identification module during the maintenance process. If the feedback information finds that the accuracy of the second-category prediction results is lower than the minimum threshold for multiple times in a row, the operation and maintenance personnel will be prompted to readjust the model and re-activate the drone thermal imager to verify the second-category prediction results.

[0066] The above-mentioned embodiments only express the specific implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the protection scope of the present application. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the technical solution concept of the present application, and these all belong to the protection scope of the present application.

Claims

1. A new energy power station operation and maintenance management system, characterized in that: Including data acquisition subsystem, data analysis subsystem, fault diagnosis subsystem, strategy generation subsystem and user end, The data acquisition subsystem is used to collect the operating data of the power generation component and save it in the database, the operating data includes a plurality of data items, each data item includes the photovoltaic array serial number and the corresponding operating information, and the data acquisition subsystem includes a plurality of data acquisition units; The data analysis subsystem is used to process the operation data to obtain the first type of eigenvalues ​​and the second type of eigenvalues; The fault diagnosis subsystem includes a first diagnosis module and a second diagnosis module, wherein the first diagnosis module determines the fault condition of the data acquisition subsystem according to the first type of characteristic value as a first type of prediction result, and the second diagnosis module determines the fault condition of the photovoltaic array according to the second type of characteristic value as a second type of prediction result; The strategy generation subsystem is used to determine the processing strategy according to the working status and send the working status and the corresponding processing strategy to the user end; The user end is used to receive the operation data and processing strategy of the power generation component in real time.

2. A new energy power station operation and maintenance management system according to claim 1, characterized in that: The data analysis subsystem includes a preprocessing module and a statistical analysis module. The preprocessing module divides the operation data into an abnormal data set and a normal data set. The statistical analysis module is used to extract a type of feature value from the abnormal data set and a type of feature value from the normal data set.

3. A new energy power station operation and maintenance management system according to claim 2, characterized in that: The preprocessing module first determines whether there are any data items with abnormal data. If so, the relevant data items are placed in the abnormal data set until there is no abnormal data and the remaining data items are used as the normal data set. The process of the preprocessing module determining abnormal data includes: first screening out duplicate data, missing data and dead data in the running information, and then using the 3-sigma method and the quartile method to screen out abnormal range data.

4. A new energy power station operation and maintenance management system according to claim 3, characterized in that: The operation information includes environmental data and power generation data. The statistical analysis module processes the abnormal data set including: If the abnormal data only exists in the environmental data, the environmental data in the data item of the normal photovoltaic array with the adjacent serial number is used to replace the abnormal data to realize the normalization of the abnormal data, and the normalized data item is put into the normal data set.

5. A new energy power station operation and maintenance management system according to claim 3, characterized in that: The statistical analysis module processes the abnormal data set including: counting whether the data items within the adjacent acquisition time of the data items corresponding to the abnormal data have the same type of abnormal data, if so, preliminarily determining that the abnormal data is related to the data acquisition subsystem, and is caused abnormal data, and retaining the data item; if the data items within the adjacent acquisition time do not have the same type of abnormal data, preliminarily determining that the abnormal data is not related to the data acquisition subsystem, and is uncaused abnormal data, and eliminating the data item; after eliminating the uncaused abnormal data, obtaining the caused abnormal data set, extracting the characteristic value of each data item in the caused abnormal data set as a type of characteristic value, the type of characteristic value includes the abnormal data type, the abnormal data position, and the photovoltaic array serial number corresponding to the data item.

6. A new energy power station operation and maintenance management system according to claim 3, characterized in that: The characteristic value of each data item in the normal data set is extracted and processed as the second type of characteristic value, and the second type of characteristic value includes: volt-ampere characteristic curve parameters and output voltage, output current, temperature, and light intensity. The volt-ampere characteristic curve includes short-circuit current, open-circuit voltage, and current and voltage corresponding to the maximum power point.

7. A new energy power station operation and maintenance management system according to claim 1, characterized in that: The first type of prediction results includes the location of the faulty data acquisition unit, the fault location in the data acquisition unit and the specific fault type; wherein, the faulty data acquisition unit is determined according to the photovoltaic array serial number in the data item, the faulty sensor is determined according to the abnormal data location, and then the corresponding sensor fault type is determined according to the abnormal data type. If it is missing data, the corresponding sensor is determined to be invalid; if it is abnormal range data, the corresponding sensor is determined to be faulty; if it is dead data, that is, the data continues to remain unchanged, it is determined that the corresponding information upload function is faulty.

8. The new energy power station operation and maintenance management system according to claim 1, characterized in that: For the data items determined as normal by the second diagnostic module, the power of the photovoltaic array under the standard state is calculated in combination with the temperature and radiation intensity. The specific calculation formula is: P=GτηA[1-0.045(T c -25)] Where G is the radiation intensity, W / m 2 ; τ is the outer layer projection rate of the photovoltaic cell; η is the conversion efficiency of the photovoltaic cell; A is the receiving surface area of ​​the photovoltaic module; T c is the operating temperature of the photovoltaic cell; If the difference between the calculated power value and the measured power value is too large, it is judged that the photovoltaic array corresponding to the data item may have too much dust accumulation, and the normal label corresponding to the data item is changed to a dust label; the measured power value is the average power value within a certain range of temperature and radiation intensity changes.

9. A new energy power station operation and maintenance management system according to claim 1, characterized in that: The strategy generation subsystem includes an expert experience base, a historical database and a strategy MAP table. The strategy MAP table is obtained through deep learning training based on the expert experience base and the historical database. The expert experience base and the historical database are updated in real time. The strategy MAP is regularly iterated and trained based on the updated data of the expert experience base and the historical database.

10. A new energy power station operation and maintenance management system according to claim 1, characterized in that: The data acquisition subsystem also includes a UAV thermal imager, and the strategy generation subsystem also includes an identification module. The identification module determines the faulty photovoltaic array based on the second type of prediction results, controls the UAV thermal imager based on the location information of the faulty photovoltaic array to collect thermal infrared images of the designated photovoltaic array, and verifies the fault condition of the photovoltaic array based on the collected thermal infrared images; if the accuracy of the second type of prediction results is verified to reach a minimum threshold for multiple times in a row, the second type of prediction results will no longer be verified by the UAV thermal imager. At the same time, during the maintenance process, the operation and maintenance personnel will feed back the fault information of the second type of prediction results to the identification module. If it is found through the feedback information that the accuracy of the second type of prediction results is lower than the minimum threshold for multiple times in a row, the operation and maintenance personnel will be prompted to readjust the model and re-enable the UAV thermal imager to verify the second type of prediction results.

Images