Automated monitoring method, device, equipment and storage medium for photovoltaic power station

Abstract

The present invention provides an automated monitoring method, device, equipment, and storage medium for a photovoltaic power station, relating to the field of photovoltaic power station monitoring, including: obtaining environmental feature data and inverter operating data, and calculating the average impact value of each environmental feature on the operating data of each inverter; establishing an environmental impact matrix for each inverter based on the average impact value, and calculating the correlation coefficient of the environmental impact matrices of adjacent inverters; performing regional segmentation and anomaly identification on the power station based on the physical spatial location and correlation coefficient of the inverter to obtain the inverter to be checked; updating the weights of the training samples and the diagnostic model based on the correlation coefficient between the environmental impact matrices of the inverters; and performing fault diagnosis on the inverter to be checked and generating an operation and maintenance plan. This application improves the accuracy of anomaly identification by performing correlation analysis on environmental features and inverter operating parameters, and at the same time optimizes the fault diagnosis model to improve the accuracy of fault diagnosis.

Description

Technical Field

[0001] The present invention relates to the field of photovoltaic power station monitoring, and in particular to an automated monitoring method, device, equipment and medium for a photovoltaic power station. Background Art

[0002] Fault diagnosis technology for photovoltaic power plants is crucial for automated monitoring. With the widespread adoption of photovoltaic power generation, fault diagnosis technology is constantly evolving. Common fault diagnosis techniques currently include imaging, electrical signature analysis, and machine learning. Machine learning-based diagnostic methods can identify a wide range of fault types and provide relatively accurate predictions of power plant output. However, PV power plants are complex and numerous, and the operating parameters of various devices are significantly affected by environmental fluctuations. Therefore, in some complex environments, the accuracy of machine learning-based diagnostic results is low. Summary of the Invention

[0003] The purpose of the present invention is to provide a method, device, equipment and medium for automated monitoring of photovoltaic power plants to improve the above-mentioned problems. To achieve the above-mentioned purpose, the technical solutions adopted by the present invention are as follows:

[0004] In a first aspect, the present application provides an automated monitoring method for a photovoltaic power station, comprising:

[0005] Acquiring environmental characteristic data and inverter operating data, wherein the inverter operating data includes string electrical characteristic data, and calculating the average impact value of each environmental characteristic on each inverter operating data;

[0006] Establishing an environmental impact matrix for each inverter based on the average impact value, and calculating a correlation coefficient between the environmental impact matrix of each inverter and its adjacent inverters;

[0007] Based on the physical spatial location of the inverters and the correlation coefficients between the inverters, the power station is segmented using a clustering algorithm to obtain multiple operating areas. Abnormal points in each operating area are then found, and the inverters to be checked are determined based on the abnormal points.

[0008] Calculate the correlation coefficients between the environmental impact matrices of all inverters and draw the correlation coefficient curves of each inverter and other inverters in time sequence;

[0009] A training sample is constructed based on the historical operating data of all inverters in the power plant. The weight of each training sample is updated based on the correlation coefficient curve between the inverter to be checked and other sample inverters. The pre-established diagnostic model is updated using the training sample with the updated weight.

[0010] Obtain the operating data of the inverter to be checked, input it into the updated diagnostic model, obtain the fault diagnosis result, search the historical maintenance strategy based on the fault diagnosis result, obtain the available resource list, and generate the operation and maintenance plan based on the historical maintenance strategy and the available resource list.

[0011] In a second aspect, the present application provides an automated monitoring device for a photovoltaic power station, characterized by comprising:

[0012] A first calculation module is configured to obtain environmental characteristic data and inverter operation data, wherein the inverter operation data includes string electrical characteristic data, and calculate an average impact value of each environmental characteristic on each inverter operation data;

[0013] a second calculation module, configured to establish an environmental impact matrix for each inverter based on the average impact value, and calculate a correlation coefficient between the environmental impact matrix of each inverter and the environmental impact matrix of its adjacent inverters;

[0014] The segmentation module is used to segment the power station into multiple operating areas based on the physical spatial location of the inverters and the correlation coefficients between the inverters using a clustering algorithm. The module then searches for abnormal points in each operating area and identifies the inverter to be checked based on the abnormal points.

[0015] The third calculation module is used to calculate the correlation coefficient between the environmental impact matrices of all inverters and draw the correlation coefficient curve between each inverter and other inverters in time sequence;

[0016] An update module is used to construct training samples based on the historical operating data of all inverters in the power plant, and to update the weight of each training sample based on the correlation coefficient curve between the inverter to be checked and other sample inverters. The training samples with updated weights are used to update the pre-established diagnostic model;

[0017] The diagnostic module is used to obtain the operating data of the inverter to be checked, input it into the updated diagnostic model, obtain the fault diagnosis result, search the historical maintenance strategy based on the fault diagnosis result, obtain the list of available resources, and generate the operation and maintenance plan based on the historical maintenance strategy and the list of available resources.

[0018] In a third aspect, the present application provides an automated monitoring device for a photovoltaic power station, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the automated monitoring method for the photovoltaic power station as described above when executing the computer program.

[0019] In a fourth aspect, the present application provides a readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the above-mentioned method for automated monitoring of a photovoltaic power station are implemented.

[0020] The beneficial effects of the present invention are:

[0021] This application improves the accuracy of abnormality identification by performing correlation analysis between environmental characteristics and inverter operating parameters, rationally dividing the power station into areas with different operating states, and searching for abnormal inverters by area. At the same time, based on the similarity of the inverter operating states, the weights of training samples are adjusted to continuously optimize the fault diagnosis model, so that the fault diagnosis model can better adapt to changes in the environment and the working performance of each device, effectively improving the diagnosis accuracy.

[0022] Other features and advantages of the present invention will be set forth in the following description, and in part will be apparent from the description, or may be learned by practicing embodiments of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without paying any creative work.

[0024] Figure 1 This is a flow chart of the automated monitoring method for a photovoltaic power station according to an embodiment of the present application;

[0025] Figure 2 This is a schematic diagram of the structure of an automated monitoring device for a photovoltaic power station in an embodiment of the present application;

[0026] Figure 3 This is a schematic diagram of the structure of the automated monitoring equipment for a photovoltaic power station in an embodiment of the present application;

[0027] Markings in the figure: 100 - first calculation module; 110 - processing unit; 120 - generation unit; 130 - calculation unit; 140 - elimination unit; 200 - second calculation module; 300 - segmentation module; 400 - third calculation module; 500 - update module; 600 - diagnosis module; 800 - automated monitoring equipment for photovoltaic power station; 801 - processor; 802 - memory; 803 - multimedia component; 804 - I / O interface; 805 - communication component. DETAILED DESCRIPTION

[0028] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings herein can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0029] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, it does not require further definition or explanation in subsequent drawings. Embodiment 1:

[0030] See also Figure 1 , an automated monitoring method for a photovoltaic power station, comprising steps S100, S200, S300, S400, S500 and S600;

[0031] S100, acquiring environmental characteristic data and inverter operating data, wherein the inverter operating data includes string electrical characteristic data, and calculating the average impact value of each environmental characteristic on each inverter operating data;

[0032] In this step, the environmental characteristic data obtained include temperature (T), humidity (R), wind speed (W), and solar radiation intensity (I); the inverter operation data include DC power, AC power, operating temperature, and electrical characteristic data of each string connected to the inverter; among them, the electrical characteristic operation data obtained include output power (P), maximum power point current (I MPP ), maximum power point voltage (U MPP Each data is collected at a preset frequency, such as inverter operation data collected every 5 minutes, and environmental characteristic data collected every 1-2 minutes. The environmental characteristic data collected within 5 minutes are averaged, and the averaged value is used as the environmental characteristic data corresponding to the inverter operation data at that time point. After collecting data for a period of time (such as 3-4 hours), a data sequence arranged by time is obtained, and the data is processed as follows:

[0033] S110, cleaning the environmental characteristic data and the inverter operation data, and performing dynamic time warping on the cleaned data;

[0034] The environmental feature data can be processed for missing values, outliers, and smoothing, etc.; for the inverter operation data, since this application needs to identify the operation anomalies of the inverter, the inverter operation data cannot be processed for outliers, but the anomalies in the data must be retained; dynamic time warping (DTW) can match sequences by stretching and compressing the nonlinear time axis, and can effectively handle time offsets, scaling, and distortions in time series, so that the time series data can be aligned to the greatest extent; in this step, considering that the impact of environmental changes on the operation of the inverter and string may be delayed, dynamic time warping is selected to match the two time series data to improve the credibility of the subsequent calculation of the correlation coefficient.

[0035] S120. Based on the processed environmental feature data, scikit-learn is used to generate multiple variable interaction terms. Scikit-learn is a machine learning library in Python. The PolynomialFeatures class of Scikit-Learn can generate polynomial features and interaction terms between variables. These interaction terms can be the product of temperature and humidity T×R, or the second-order interaction R 2 ×I, etc., and then the feature selection technology in scikit-learn can be used to preliminarily screen important variable interactions.

[0036] S130, using the processed environmental characteristic data and variable interaction terms as input, and the processed inverter operation data as output, to calculate an average impact value to obtain an average impact value of each variable interaction term;

[0037] The mean influence value (MIV) is a method used to quantify the impact of features on model output. MIV quantifies the average impact of each feature on the output by perturbing the input features and observing the changes in the model output.

[0038] The environmental characteristic data and variable interaction terms are used as input features, and the corresponding inverter operation data is used as output to construct the feature matrix X (size is N×m, where N is the number of samples and m is the number of features). It should be noted that the output of each sample in this step is only one inverter operation data, that is, the inverter DC power, AC power, operating temperature, etc. are used as output respectively, so as to calculate the impact value of the input features on different inverter operation data.

[0039] The neural network model is trained using N samples, and then each input feature is perturbed in turn;

[0040] Denote the input feature as j, j=1,2,3...m, where m is the number of features; set the perturbation value β;

[0041] Add disturbance to feature j: increase the value of feature j by β to obtain a new feature matrix X j+ ;

[0042] X j+ =X+βM j；

[0043] Among them, M j is an N×m matrix with the jth column being 1 and the rest being 0;

[0044] Reduce the disturbance of feature j: reduce the value of feature j by β to obtain a new feature matrix X j- ;

[0045] X j- =X-βM j；

[0046] The disturbed sample is input into the trained neural network model, and the output changes before and after the disturbance are calculated; the output change of each sample is averaged to obtain the average influence value MIV of feature j j ;

[0047] S140. Eliminate variable interaction items according to the average impact value to obtain retained interaction items, and record the average impact value of each environmental characteristic data and the retained interaction items.

[0048] If the average influence value is very small, it proves that the variable interaction term has a very low impact on the prediction result. These terms are eliminated and only the features that have a greater impact on the inverter operating status are retained.

[0049] S200, establishing an environmental impact matrix for each inverter based on the average impact value, and calculating a correlation coefficient between the environmental impact matrix of each inverter and its adjacent inverters;

[0050] The environmental impact matrix has environmental characteristics and retained interaction terms as rows and different inverter operation data as columns, and the elements thereof are correlation coefficients;

[0051] In a photovoltaic power station, inverters are set up according to the location of the photovoltaic panels and are generally distributed in an array. Adjacent inverters refer to the four inverters on the top, bottom, left, and right. Inverters at the edge of the array only calculate the correlation coefficient with two or three adjacent inverters. The correlation coefficient can be calculated using the Pearson correlation coefficient or the Kendall correlation coefficient.

[0052] S300: Based on the physical spatial locations of the inverters and the correlation coefficients between the inverters, the power station is segmented using a clustering algorithm to obtain multiple operating areas. Anomalies are found in each operating area, and the inverter to be checked is determined based on the anomalies. Specifically:

[0053] S310: Randomly select an inverter as an encoding origin, encode the coordinate data of each inverter according to the physical spatial position of the inverter, and calculate the initial distance between adjacent inverters based on the coordinate data; the coordinate data is a two-dimensional coordinate;

[0054] In this step, if the inverters are arranged neatly and orderly, and the spacing between rows and columns is not much different, the inverters can be considered to be equally spaced during encoding. If the spacing between rows and columns is significantly increased, the distance between the inverter and the adjacent inverters is increased accordingly when encoding the coordinates. At the same time, in power stations with special terrain, the height difference between inverters needs to be considered. If the horizontal distance between inverters is not large, but the height difference is significant, the distance between inverters should also be increased. After encoding, the initial data distribution of the inverters can be obtained.

[0055] S320, starting from the inverter at the encoding origin, correct the initial distance between the inverters according to the correlation coefficient between adjacent inverters to obtain a corrected distribution data point;

[0056] When the correlation coefficient value between adjacent inverters is large, it means that the operating status of the two is basically the same, and the change trends of the operating status of the two are highly similar. Therefore, the distance between the two data points needs to be closer. If the correlation coefficient value between adjacent inverters is small, it means that the operating status of the two is different. The distance between the two data points needs to be farther, so that the data points can be separated in subsequent clustering or identified as outliers.

[0057] S330: clustering the corrected distribution data points using a density clustering algorithm to obtain clustering results, performing image segmentation based on the clustering results to obtain multiple operating areas.

[0058] This step clusters the corrected distribution data points using the DBSCAN clustering algorithm, and maps the clustering results back to the actual spatial location of each inverter, thereby segmenting the inverters in the power station.

[0059] Specifically, different clusters after clustering are displayed in different colors to obtain the actual inverter distribution map of the entire power station. The color label of each inverter is mapped to the inverter distribution map. An image segmentation algorithm is used to extract the contours and segment the inverters into different operating areas. The operating status of the inverters in each operating area is similar, while the operating status of the inverters in different operating areas is different.

[0060] Outliers identified by the DBSCAN clustering algorithm are also mapped to the inverter distribution map for visualization. Outliers in this step include noise points and potential outliers. Noise points are points that are not assigned to any cluster. Potential outliers include data points that are incorrectly assigned to a cluster, as well as when a small number of data points are assigned to the same cluster but the cluster is completely surrounded by another larger cluster, indicating that all data points in the cluster may be outliers. These points can be marked as potential outliers. In this step, potential outliers can be identified using the ODIC-DBSCAN (Outlier detection of inner-cluster based on DBSCAN) method, methods based on local outlier factors, or methods based on isolation forests.

[0061] S400, calculating the correlation coefficients between the environmental impact matrices of all inverters, and drawing correlation coefficient curves between each inverter and other inverters in time sequence;

[0062] The correlation coefficient curve is drawn and saved according to a preset period. For example, a correlation coefficient curve between inverters within one day is drawn and saved in a database. By analyzing historical data (within the past month), correlation coefficient curves with continuously low correlation coefficients can be found, and the correlation coefficient curves of the two inverters within this month can be deleted to avoid occupying computing resources.

[0063] S500: Construct training samples based on the historical operating data of all inverters in the power plant, and update the weight of each training sample based on the correlation coefficient curve between the inverter to be checked and other sample inverters. Use the training samples with updated weights to update the pre-established diagnostic model. Specifically:

[0064] S510. When it is necessary to inspect a particular inverter, first obtain a correlation coefficient curve between the inverter and other inverters from a database, obtain a preset correlation coefficient threshold, and segment each correlation coefficient curve using the correlation coefficient threshold to obtain correlation regions greater than the correlation coefficient threshold, and obtain a corresponding correlation time domain. If the Pearson correlation coefficient is used, the range of the correlation coefficient curve is [-1, 1]. In this step, the correlation coefficient threshold is set to 0.78, and the region of the curve above 0.78 is obtained. At the same time, the time period corresponding to the region is obtained, i.e., the correlation time domain, indicating that the operating states of the two inverters during this period are highly similar.

[0065] It is understandable that if there are many correlation coefficient curves stored in the database, in order to improve the calculation efficiency, the correlation coefficient curves can be searched by season, or some correlation coefficient curves can be randomly selected for calculation;

[0066] S520: Search for historical environmental characteristic data within a relevant time domain, compare the data with the environmental characteristic data of the inverter to be checked, and obtain environmental similarity;

[0067] The acquisition time of the data can be determined based on the relevant time domain, so as to find the environmental feature data of the time period, compare it with the environmental feature data of the inverter to be checked, calculate the Euclidean distance between the corresponding environmental feature data, and then add the similarity of each environmental feature to obtain the environmental similarity;

[0068] S530 : Acquire historical operating data of the inverter in each relevant time domain, and generate a first weight for the historical operating data according to the environmental similarity and the area of ​​the relevant region.

[0069] First, number each relevant area, and the number is the first weight A of the k relevant areas. k for:

[0070] A k =uS k +(1-u)H k

[0071] Among them, u is the first weight coefficient, S k is the area of ​​the relevant region k, H k is the environmental similarity of the relevant region k;

[0072] In this step, in addition to obtaining the historical operating data within the relevant time domain, it is also necessary to search for the historical operating data before and after the relevant time domain according to the preset length, so as to ensure that the inverter failure near the relevant time domain is not missed; if there is another relevant time domain before and after the preset length of the relevant time domain, that is, the historical operating data is repeated, the repeated historical operating data are merged; these historical operating data are used to construct the weighted data set E k ;

[0073] Each weighted dataset E k corresponds to a relevant region k;

[0074] Assign the first weight of the relevant region k to the corresponding weighted dataset E k All historical operating data in the system; historical operating data includes inverter DC power, AC power, operating temperature and string operating data, etc.

[0075] This step updates the weights for some historical operating data to optimize the fault diagnosis model and improve diagnostic accuracy. However, in some special climates and conditions, the spatial distribution of inverters may significantly affect the accuracy of model predictions (the spatial distribution of inverters can be equivalent to the spatial distribution of strings). Therefore, based on the above method, the spatial distribution characteristics of inverters can be further considered to optimize the model. The steps are as follows:

[0076] S540: Acquire coordinate data of each inverter, and calculate a spatial distance between the inverters based on the coordinate data; obtain a second weight of the inverter according to the spatial distance between the inverters;

[0077] The coordinate data used in this step can be the same as the coordinate data encoded in step S310, and the second weight can directly use the spatial distance d;

[0078] S550: Perform linear weighted fusion on the first weight and the second weight to obtain a comprehensive weight for each training sample;

[0079] The calculation formula of the comprehensive weight C is as follows:

[0080] C=α·A k + (1-α) e -βd ;

[0081] Among them, α is the second weight coefficient; β is the parameter that controls the attenuation rate of the distance effect. The longer the distance, the faster the influence on the weight decays. The second weight coefficient can be dynamically adjusted according to the environment or time. By deep learning historical data, the optimal α corresponding to a certain type of environmental conditions (cloudy, overcast or different seasons) can be determined.

[0082] In step S530, the first weight is generated for only part of the historical operation data. For other data for which the first weight is not generated, the first weight value A is set to k Set to 0.

[0083] The present application preferably uses the XGBoost algorithm to construct a fault diagnosis test model, and updates the comprehensive weight of each sample calculated above to the XGBoost fault diagnosis test model, so that the model pays more attention to certain samples, and obtains an updated diagnosis model.

[0084] S600. Obtain operating data of the inverter to be checked, input it into an updated diagnostic model, obtain a fault diagnosis result, search for a historical maintenance strategy based on the fault diagnosis result, obtain a list of available resources, and generate an operation and maintenance plan based on the historical maintenance strategy and the list of available resources.

[0085] The fault diagnosis results can reflect the type of fault, such as component damage, inverter failure, wiring problems, photovoltaic panel shadows, etc.; according to the fault type, the corresponding maintenance strategy is searched in the historical maintenance records and the urgency is analyzed, and keywords are extracted from the historical maintenance strategy, including spare parts keywords, such as inverter, cable and iV curve tester, etc.; personnel keywords, that is, the list of necessary or available personnel to handle this type of fault; obtain a list of available resources, which includes current inventory spare parts and personnel duty rosters; filter out the resources needed to solve the current fault from the list of available resources, and then automatically generate one or more operation and maintenance plans based on the urgency of the fault. The operation and maintenance plans are pushed to the staff while issuing fault warnings, providing the staff with accurate fault information and recommended maintenance strategies.

[0086] The above method focuses on fault diagnosis of a single inverter or inverters in a small area, but it is difficult to reflect abnormalities in a large area. Therefore, in order to monitor abnormalities in a large area at the same time, the method of the present application further includes step S700:

[0087] updating the correlation coefficient according to a preset frequency to obtain operation area information arranged in time; the operation area information includes an operation area map;

[0088] Obtaining operation area change information based on the operation area graphs at different times, the operation area change information including movement information of the operation area center point and movement information of each operation area edge, specifically including movement distance and movement direction of the operation area center point and movement distance and movement direction of the operation area edge;

[0089] Acquiring cloud image information and environmental information, wherein the environmental information includes solar radiation angle, solar radiation intensity, temperature, wind speed and wind direction;

[0090] Using cloud map information, environmental information, and operating area change information as training sets, a convolutional neural network model is trained to obtain an operating area prediction model;

[0091] Based on the operating area prediction model, the changes in the operating area are predicted, and the actual change information of the operating area is compared with the predicted results to diagnose regional anomalies. The operating area prediction model can predict the change trend of the operating area according to the changes in the environment, and judge whether the actual change trend is reasonable based on the predicted change trend. If the actual value and the predicted value are significantly different, it may indicate that there is an abnormality in a certain area of ​​the power station, and an alarm will be issued to prompt the staff to conduct further inspection. Example 2:

[0092] See also Figure 2 The present application also provides an automated monitoring device for a photovoltaic power station, comprising:

[0093] A first calculation module 100 is configured to obtain environmental characteristic data and inverter operation data, wherein the inverter operation data includes string electrical characteristic data, and calculate an average impact value of each environmental characteristic on each inverter operation data;

[0094] A second calculation module 200 is configured to establish an environmental impact matrix for each inverter based on the average impact value, and calculate a correlation coefficient between the environmental impact matrix of each inverter and its adjacent inverters;

[0095] The segmentation module 300 is configured to segment the power station into regions based on the physical spatial locations of the inverters and the correlation coefficients between the inverters using a clustering algorithm to obtain multiple operating regions, find abnormal points in each operating region, and determine the inverter to be checked based on the abnormal points;

[0096] The third calculation module 400 is used to calculate the correlation coefficients between the environmental impact matrices of all inverters and draw the correlation coefficient curves between each inverter and other inverters in time sequence;

[0097] An updating module 500 is configured to construct training samples based on the historical operating data of all inverters in the power plant, update the weight of each training sample based on a correlation coefficient curve between the inverter to be checked and other sample inverters, and update the pre-established diagnostic model using the training samples with the updated weights;

[0098] The diagnostic module 600 is used to obtain the operating data of the inverter to be checked, input it into the updated diagnostic model, obtain the diagnostic results, search the historical maintenance strategy based on the fault diagnosis results, and obtain the available resource list, and generate an operation and maintenance plan based on the historical maintenance strategy and the available resource list.

[0099] As an optional implementation, the first calculation module 100 includes:

[0100] The processing unit 110 is configured to clean the environmental characteristic data and the inverter operation data, and perform dynamic time warping on the cleaned data;

[0101] A generating unit 120 is configured to generate a plurality of variable interaction terms using scikit-learn based on the processed environmental feature data;

[0102] A calculation unit 130 is configured to take the processed environmental characteristic data and the variable interaction term as input and the processed inverter operation data as output, perform average impact value calculation, and obtain an average impact value of each variable interaction term;

[0103] The elimination unit 140 is used to eliminate the variable interaction items according to the average impact value to obtain the retained interaction items, and record each environmental feature data and the average impact value of the retained interaction items. Example 3:

[0104] Corresponding to the above method embodiment, this embodiment further provides an automated monitoring system for a photovoltaic power station. The automated monitoring system for a photovoltaic power station described below and the automated monitoring method for a photovoltaic power station described above can refer to each other.

[0105] Figure 3 FIG. 8 is a block diagram of an automatic monitoring panel 800 for a photovoltaic power station according to an exemplary embodiment. Figure 3 As shown, the photovoltaic power plant automated monitoring panel 800 includes a processor 801 and a memory 802. The photovoltaic power plant automated monitoring panel 800 may also include one or more of a multimedia component 803, an input / output (I / O) interface 804, and a communication component 805. The processor 801 is used to control the overall operation of the photovoltaic power plant automated monitoring panel 800 to complete all or part of the steps in the photovoltaic power plant automated monitoring panel method described above. The memory 802 is used to store various types of data to support the operation of the photovoltaic power plant automated monitoring panel 800. This data may include, for example, commands for any application or method operating on the photovoltaic power plant automated monitoring panel 800, as well as application-related data, such as contact information, sent and received messages, pictures, audio, video, and the like. The memory 802 can be implemented by any type of volatile or non-volatile memory device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0106] The multimedia component 803 may include a screen and an audio component. The screen may be, for example, a touch screen, and the audio component is used to output and / or input audio signals. For example, the audio component may include a microphone for receiving external audio signals.

[0107] The received audio signal can be further stored in the memory 802 or sent through the communication component 805. The audio component also includes at least one speaker for outputting audio signals. The I / O interface 804 provides an interface between the processor 801 and other interface modules. The above-mentioned other interface modules can be keyboards, mice, buttons, etc. These buttons can be virtual buttons or physical buttons. The communication component 805 is used for wired or wireless communication between the automated monitoring panel 800 of the photovoltaic power station and other devices. Wireless communication, such as Wi-Fi, Bluetooth, near field communication (NFC), 2G, 3G or 4G, or a combination of one or more of them, so the corresponding communication component 805 can include: Wi-Fi module, Bluetooth module, NFC module.

[0108] In an exemplary embodiment, the device 800 for mutual signing and verification of digital files can be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors or other electronic components to execute the above-mentioned automated monitoring method for photovoltaic power stations.

[0109] In another exemplary embodiment, a computer-readable storage medium including program commands is further provided. When executed by a processor, the program commands implement the steps of the aforementioned method for automated monitoring of a photovoltaic power station. For example, the computer-readable storage medium may be the aforementioned memory 802 including the program commands. The program commands may be executed by the processor 801 of the automated monitoring system 800 for a photovoltaic power station to implement the aforementioned method for automated monitoring of a photovoltaic power station. Example 4

[0110] Corresponding to the above embodiment of the method for automated monitoring of a photovoltaic power station, this embodiment further provides a readable storage medium. The readable storage medium described below and the above-described method for automated monitoring of a photovoltaic power station can refer to each other.

[0111] A readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the embodiment of the above-mentioned method for automated monitoring of a photovoltaic power station.

[0112] The readable storage medium may specifically be any readable storage medium that can store program code, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0113] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or device comprising the element.

[0114] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A method for automatically monitoring a photovoltaic power station, characterized in that: include: Acquiring environmental characteristic data and inverter operating data, wherein the inverter operating data includes string electrical characteristic data, and calculating the average impact value of each environmental characteristic on each inverter operating data; Establishing an environmental impact matrix for each inverter based on the average impact value, and calculating a correlation coefficient between the environmental impact matrix of each inverter and its adjacent inverters; Based on the physical spatial location of the inverters and the correlation coefficients between the inverters, the power station is segmented using a clustering algorithm to obtain multiple operating areas. Abnormal points in each operating area are then found, and the inverters to be checked are determined based on the abnormal points. Based on the physical spatial location of the inverters and the correlation coefficients between the inverters, the power station is segmented using a clustering algorithm to obtain multiple operating areas, including: Randomly select an inverter as the encoding origin, encode the coordinate data of each inverter according to the physical spatial position of the inverter, and calculate the initial distance between adjacent inverters based on the coordinate data; Starting from the inverter at the encoding origin, the initial distance between inverters is corrected according to the correlation coefficient between adjacent inverters to obtain the corrected distribution data points; The density clustering algorithm is used to cluster the corrected distribution data points to obtain clustering results. The image is segmented based on the clustering results to obtain multiple operating areas. Calculate the correlation coefficients between the environmental impact matrices of all inverters and draw the correlation coefficient curves of each inverter and other inverters in time sequence; A training sample is constructed based on the historical operating data of all inverters in the power plant. The weight of each training sample is updated based on the correlation coefficient curve between the inverter to be checked and other sample inverters. The pre-established diagnostic model is updated using the training sample with the updated weight. Obtain the operating data of the inverter to be checked, input it into the updated diagnostic model, obtain the fault diagnosis result, search the historical maintenance strategy based on the fault diagnosis result, obtain the available resource list, and generate the operation and maintenance plan based on the historical maintenance strategy and the available resource list.

2. The method for automatic monitoring of a photovoltaic power station according to claim 1, characterized in that: The acquiring of environmental characteristic data and inverter operation data, wherein the inverter operation data includes string electrical characteristic data, and respectively calculating an average impact value of each environmental characteristic on each inverter operation data, includes: Cleaning the environmental characteristic data and the inverter operation data, and performing dynamic time warping on the cleaned data; Based on the processed environmental characteristic data, scikit-learn was used to generate multiple variable interaction terms; The processed environmental characteristic data and variable interaction terms are used as input, and the processed inverter operation data is used as output to calculate the average impact value and obtain the average impact value of each variable interaction term; The variable interaction terms were eliminated according to the average effect value to obtain the retained interaction terms, and the average effect value of each environmental characteristic data and the retained interaction terms was recorded.

3. The method for automatic monitoring of a photovoltaic power station according to claim 1, characterized in that: The method constructs training samples based on the historical operating data of all inverters in the power station, updates the weight of each training sample based on the correlation coefficient curve between the inverter to be checked and other sample inverters, and uses the training samples with updated weights to update the pre-established diagnostic model, including: Obtain a preset correlation coefficient threshold, segment each correlation coefficient curve using the correlation coefficient threshold, obtain a correlation region greater than the correlation coefficient threshold, and obtain a corresponding correlation time domain; Find the historical environmental feature data within the relevant time domain, compare it with the environmental feature data of the current inverter to be checked, and obtain the environmental similarity; According to the historical operation data of the inverter acquired in each relevant time domain, a first weight is generated for the historical operation data according to the environmental similarity and the area of ​​the relevant region.

4. The method for automatic monitoring of a photovoltaic power station according to claim 3, characterized in that: The method constructs training samples based on the historical operating data of all inverters in the power station, updates the weight of each training sample based on the correlation coefficient curve between the inverter to be checked and other sample inverters, and updates the pre-established diagnostic model using the updated training samples, including: Obtaining coordinate data of each inverter, and calculating a spatial distance between the inverters based on the coordinate data; obtaining a second weight of the inverter according to the spatial distance between the inverters; Performing linear weighted fusion on the first weight and the second weight to obtain a comprehensive weight for each training sample.

5. The method for automatic monitoring of a photovoltaic power station according to claim 1, characterized in that: The method further comprises: updating the correlation coefficient according to a preset frequency to obtain operation area information arranged in time; the operation area information includes an operation area map; Obtaining operation area change information according to the operation area graphs at different times, wherein the operation area change information includes movement information of the operation area center point and movement information of each operation area edge; Acquiring cloud image information and environmental information, wherein the environmental information includes solar radiation angle, solar radiation intensity, temperature, wind speed and wind direction; Using cloud map information, environmental information, and operating area change information as training sets, a convolutional neural network model is trained to obtain an operating area prediction model; Based on the operating area prediction model, the changes in the operating area are predicted, and the actual change information of the operating area is compared with the prediction results to perform regional anomaly diagnosis.

6. An automatic monitoring device for a photovoltaic power station, characterized in that: include: A first calculation module is configured to obtain environmental characteristic data and inverter operation data, wherein the inverter operation data includes string electrical characteristic data, and calculate an average impact value of each environmental characteristic on each inverter operation data; a second calculation module, configured to establish an environmental impact matrix for each inverter based on the average impact value, and calculate a correlation coefficient between the environmental impact matrix of each inverter and the environmental impact matrix of its adjacent inverters; The segmentation module is used to segment the power station into multiple operating areas based on the physical spatial location of the inverters and the correlation coefficients between the inverters using a clustering algorithm. The module then searches for abnormal points in each operating area and identifies the inverter to be checked based on the abnormal points. Based on the physical spatial location of the inverters and the correlation coefficients between the inverters, the power station is segmented using a clustering algorithm to obtain multiple operating areas, including: Randomly select an inverter as the encoding origin, encode the coordinate data of each inverter according to the physical spatial position of the inverter, and calculate the initial distance between adjacent inverters based on the coordinate data; Starting from the inverter at the encoding origin, the initial distance between inverters is corrected according to the correlation coefficient between adjacent inverters to obtain the corrected distribution data points; The density clustering algorithm is used to cluster the corrected distribution data points to obtain clustering results. The image is segmented based on the clustering results to obtain multiple operating areas. The third calculation module is used to calculate the correlation coefficient between the environmental impact matrices of all inverters and draw the correlation coefficient curve between each inverter and other inverters in time sequence; An update module is used to construct training samples based on the historical operating data of all inverters in the power plant, and to update the weight of each training sample based on the correlation coefficient curve between the inverter to be checked and other sample inverters. The training samples with updated weights are used to update the pre-established diagnostic model; The diagnostic module is used to obtain the operating data of the inverter to be checked, input it into the updated diagnostic model, obtain the fault diagnosis result, search the historical maintenance strategy based on the fault diagnosis result, obtain the list of available resources, and generate the operation and maintenance plan based on the historical maintenance strategy and the list of available resources.

7. The automatic monitoring device for a photovoltaic power station according to claim 6, characterized in that: The first calculation module includes: a processing unit, configured to clean the environmental characteristic data and the inverter operation data, and perform dynamic time warping on the cleaned data; A generation unit, used to generate multiple variable interaction terms using scikit-learn based on the processed environmental feature data; a calculation unit, configured to take the processed environmental characteristic data and the variable interaction term as input, take the processed inverter operation data as output, perform average impact value calculation, and obtain the average impact value of each variable interaction term; The elimination unit is used to eliminate the variable interaction items according to the average impact value to obtain the retained interaction items, and record the average impact value of each environmental characteristic data and the retained interaction items.

8. An automated monitoring device for a photovoltaic power station, characterized in that: The method comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the method implements the steps of the automated monitoring method for a photovoltaic power station as claimed in any one of claims 1 to 5.

9. A readable storage medium, characterized in that: The readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the automatic monitoring method for a photovoltaic power station according to any one of claims 1 to 5 are implemented.

Images