Monitoring system and method for photovoltaic power station based on unmanned aerial vehicle

Abstract

The invention belongs to the technical field of photovoltaic power station inspection, and particularly relates to a photovoltaic power station monitoring system and method based on an unmanned aerial vehicle, and the system comprises a data collection module, a data processing and analysis module, an identification prompt module and a route planning module. Through uploading the processed image data to the data analysis module, the data analysis module is utilized to identify the fault types in the image data, the fault types including dust, shielding, surface stains, fragmentation, diode damage, string falling, assembly falling and the like can be automatically identified, in addition, according to the fault events after automatic identification, an analysis result can be correspondingly generated, and the analysis result can be used for analyzing the fault events. The inspection route of the unmanned aerial vehicle can be dynamically edited and modified through the route planning module, so that the system can timely discover the photovoltaic panel with frequent faults, thereby avoiding long fault duration of the fault photovoltaic panel caused by inspection according to a constant inspection route; and the overall abnormity and even damage of the photovoltaic module string are caused in severe cases.

Description

Technical Field

[0001] The present invention belongs to the technical field of photovoltaic power station inspection, and specifically relates to a monitoring system and method for a photovoltaic power station based on an unmanned aerial vehicle. Background Art

[0002] With the widespread application of photovoltaic power stations, their operation and maintenance management has become an important problem. The traditional monitoring method of photovoltaic power stations mainly relies on manual inspections, which is not only time-consuming and labor-intensive, but also has problems such as low detection efficiency and poor accuracy. Especially in large photovoltaic power stations, manual inspections are difficult to achieve full coverage and real-time monitoring, resulting in delayed fault detection, affecting photovoltaic power generation efficiency and system stability.

[0003] In recent years, the development of drone technology has provided a new solution for photovoltaic power station monitoring. Drones have the advantages of high efficiency, flexibility, and no terrain restrictions. They can quickly obtain high-definition image data of photovoltaic panels and realize comprehensive monitoring of photovoltaic power stations. However, there is a problem with the existing photovoltaic power station monitoring based on drones, that is, how to plan the inspection route of drones. In specific scenarios, the stable operation of photovoltaic panels is affected by problems such as hot spots, obstruction, and damage. Therefore, the existing technology follows a constant inspection route for inspection, which does not take into account that some photovoltaic panels may have frequent hot spots or obstruction problems. Therefore, according to the constant inspection route, it may not be possible to timely discover the abnormal phenomenon of photovoltaic panels, resulting in a long duration of faults, especially hot spot problems. If the duration is long, it may cause the overall abnormality of the photovoltaic module string or even damage.

[0004] To this end, the present invention provides a monitoring system and method for a photovoltaic power station based on an unmanned aerial vehicle. Summary of the invention

[0005] In order to make up for the deficiencies of the prior art, at least one technical problem raised in the background technology is solved.

[0006] The technical solution adopted by the present invention to solve the technical problem is: a monitoring system and method of a photovoltaic power station based on an unmanned aerial vehicle described in the present invention comprises:

[0007] A data acquisition module, used to acquire image data taken by the drone, including first image data and second image data;

[0008] A data processing and analysis module, used for preprocessing the image data, including preprocessing the first image data and the second image data, and analyzing the second image data, and outputting the analysis result;

[0009] Identification prompt module, used to output fault prompt according to the analysis result;

[0010] The route planning module is used to plan the inspection route of the drone and edit and modify the inspection route of the drone based on historical analysis results.

[0011] Preferably, the first image data is used to construct an orthophoto model of a photovoltaic power station, and the second image data is used to perform fault analysis of the photovoltaic power station.

[0012] Preferably, the method for planning the inspection route of the drone is:

[0013] Acquiring first image data;

[0014] constructing an orthophoto model of the photovoltaic power station based on the first image data;

[0015] According to the orthophoto model, an electronic map of the photovoltaic power station is obtained after processing;

[0016] Based on the electronic map, obtain the position coordinates of the center point of the photovoltaic component string;

[0017] The center point of the PV panel string is used as the waypoint, and all waypoints are connected to generate an inspection route.

[0018] Preferably, the data processing and analysis module includes a first deep learning model, and the first deep learning model is used to process the electronic map; the method for processing the electronic map is:

[0019] Inputting the electronic map into the first deep learning model;

[0020] Segment each photovoltaic panel in the electronic map based on instance segmentation technology, and

[0021] Number each photovoltaic panel after segmentation;

[0022] The number is used to characterize the logical position of the photovoltaic panel in the electronic map; the number is created based on the logical hierarchy of new power station-new area-new sub-array-new bracket-new branch-new component.

[0023] Preferably, the method for generating an inspection route includes:

[0024] Obtaining historical analysis results, wherein the historical analysis results include historical faulty photovoltaic panels and corresponding fault types and fault probabilities;

[0025] Taking the photovoltaic module string as the basic group j, construct a set S to be inspected that includes all basic groups j;

[0026] Count the average failure rate s of each basic group j in the inspection set S j , according to the formula:

[0027]

[0028] Among them, T represents the number of inspections corresponding to one inspection cycle, t i，j represents the number of failures of any photovoltaic panel in basic group j within a patrol cycle, n represents the total number of photovoltaic panels in basic group j, and s j represents the average failure rate of photovoltaic panels in the basic group;

[0029] The average failure rate s j Sort all basic groups j and select the average failure rate s j The basic group j with a failure rate greater than the threshold value is obtained, and the priority inspection set S is obtained. z ;

[0030] Set the priority inspection set S z Delete from the waiting inspection set S to obtain the non-priority inspection set S w ;

[0031] Based on the priority inspection set S z , generate the first inspection route;

[0032] Based on non-priority inspection set S w , generate the second inspection route;

[0033] The first inspection route and the second inspection route are processed and merged to obtain an inspection route.

[0034] Preferably, the method for generating an inspection route further includes:

[0035] Get the waypoint corresponding to each basic group j in the set to be inspected S, denoted as P;

[0036] Where P = {P1, P2, ..., P N};

[0037] Set the priority inspection set S z The waypoint corresponding to the basic group j contained in is denoted as ΔP;

[0038] Where ΔP={P p1 , P p2 ,…,P pM}(M≤N);

[0039] Set the objective function, expressed as follows:

[0040]

[0041] Among them, C ij Indicates that from waypoint P i To waypoint P j The distance K ijis a decision variable. When there is a path from i to j, then K ij =1, otherwise 0;

[0042] Optimize the objective function, subject to the following constraints:

[0043]

[0044] Among them, u pi is the priority inspection set S z Waypoint P pi Inspection order; u i It is waypoint P i Inspection order; u j It is waypoint P j The inspection order;

[0045] A constraint function that requires each waypoint to be passed and left only once;

[0046] The inspection route is obtained by calculating the above objective function based on the simulated annealing algorithm.

[0047] Preferably, the data processing and analysis module further includes a second deep learning model, which is used to analyze the image data and output the analysis results; the method for processing and analyzing the image data is:

[0048] receiving second image data;

[0049] Inputting the second image data into a second deep learning model for analysis;

[0050] Based on the second deep learning model, identify whether the photovoltaic panel is faulty, and if there is a fault, determine the fault type and obtain the serial number of the faulty photovoltaic panel, and generate a fault event in combination with the corresponding fault image data; if there is no fault, do not perform any operation;

[0051] Based on all the acquired fault events, they are summarized according to the fault type and the analysis results are generated;

[0052] The fault image data is second image data that is identified as having a fault.

[0053] Preferably, the method of outputting a fault prompt according to the analysis result is:

[0054] Get analysis results;

[0055] Extract the serial number of the faulty photovoltaic panel from the analysis results;

[0056] According to the number of the faulty photovoltaic panel, the corresponding position is marked in the electronic map to provide a fault prompt.

[0057] Preferably, it also includes a data storage module for storing all image data and the analysis results and fault events output by the storage data analysis module;

[0058] The data storage module is also used for data indexing and retrieval, allowing users to query and filter by time and fault type.

[0059] A method for monitoring a photovoltaic power station based on an unmanned aerial vehicle comprises the following steps: S1: determining a photovoltaic power station to be inspected;

[0060] S2: Obtain an electronic map of the photovoltaic power station to be inspected;

[0061] S3: Planning the inspection route based on the electronic map of the photovoltaic power station to be inspected;

[0062] S4: Perform inspection tasks based on the inspection route and upload image data;

[0063] S5: Analyze the image data to obtain analysis results, and output a fault prompt according to the analysis results.

[0064] The beneficial effects of the present invention are as follows:

[0065] The photovoltaic power station monitoring system and method based on drones described in the present invention uploads processed image data to a data analysis module, and uses the data analysis module to identify the fault type in the image data. The fault types that can be automatically identified include: dust, occlusion, surface stains, fragmentation, diode damage, string drop, component fall, etc. In addition, according to the fault events after automatic identification, analysis results can be generated accordingly and can be viewed by users. Through the route planning module, the inspection route of the drone can be dynamically edited and modified, so that the system can timely discover photovoltaic panels with frequent faults, thereby avoiding inspections along a constant inspection route, which leads to long fault duration of the faulty photovoltaic panel, and in severe cases, leads to abnormality or even damage of the entire photovoltaic component string; BRIEF DESCRIPTION OF THE DRAWINGS

[0066] The present invention will be further described below in conjunction with the accompanying drawings.

[0067] Figure 1 is a flow chart of the present invention;

[0068] Figure 2 is a schematic diagram of an orthophoto model in the system interface of the present invention;

[0069] Figure 3 It is a schematic diagram of an electronic map in the system interface of the present invention;

[0070] Figure 4 It is the first schematic diagram of the fault prompt in the system interface of the present invention;

[0071] Figure 5 It is a second schematic diagram of a fault prompt in the system interface of the present invention;

[0072] Figure 6 It is a schematic diagram of the analysis results in the system interface of the present invention;

[0073] Figure 7 It is a schematic diagram of inspection route planning in the system interface of the present invention. DETAILED DESCRIPTION

[0074] In order to make the technical means, creative features, objectives and effects achieved by the present invention easy to understand, the present invention is further explained below in conjunction with specific implementation methods.

[0075] like Figures 1 to 7 As shown, a monitoring system and method for a photovoltaic power station based on an unmanned aerial vehicle described in an embodiment of the present invention include a data acquisition module, a data processing and analysis module, an identification prompt module and a route planning module; the data acquisition module is used to collect image data taken by the unmanned aerial vehicle, including first image data and second image data; the data processing and analysis module is used to pre-process the image data, including pre-processing the first image data and the second image data, and analyzing the second image data, and outputting the analysis results; the identification prompt module is used to output fault prompts according to the analysis results; the route planning module is used to plan the inspection route of the unmanned aerial vehicle and edit and modify the inspection route of the unmanned aerial vehicle according to historical analysis results.

[0076] In one embodiment of the present invention, considering that the historical analysis results record detailed photovoltaic panels that have failed, the historical analysis results can be used to participate in editing and modifying the inspection route of the drone, so that when the drone performs the monitoring task, the photovoltaic panel areas where failure events frequently occur in the historical analysis results are prioritized for inspection, and abnormal phenomena in the photovoltaic panels are discovered in time, thereby reducing losses; specifically, the photovoltaic power station monitoring system based on drones includes a photovoltaic power station modeling stage and an actual monitoring stage; for the photovoltaic power station modeling stage, the drone only needs to perform drone mapping and two-dimensional reconstruction of the entire photovoltaic site to form a high-precision, proportional two-dimensional photovoltaic power station panoramic view, and at the same time accurately locate the geographic information of all components of the photovoltaic power station to form a power station model, which is integrated and superimposed on the satellite map to form an electronic map, and applied on the WEB platform. Secondly, in order to facilitate the accurate positioning of faults in the subsequent monitoring stage, the electronic map needs to be pre-processed; for the actual monitoring stage, first, according to the distribution of photovoltaic panel strings in the photovoltaic power station, the inspection route is customized and the inspection time is set. When the human-machine performs the inspection task, the inspection task is carried out according to the customized inspection route, which specifically includes taking the center point of the photovoltaic panel string in the inspection route as a waypoint, taking a real-life image of the photovoltaic panel string when passing the waypoint, and uploading the real-life image, that is, the image data, to the data processing module. The data processing module also needs to pre-process the real-life image collected by the drone, including denoising, binarization and other processing methods, and then upload the processed image data to the data analysis module, and use the data analysis module to identify the fault type in the image data. The fault types that can be automatically identified include: dust, obstruction, surface stains, fragmentation, diode damage, string drop, component fall, etc. In addition, according to the fault events after automatic identification, the corresponding analysis results can be generated and can be viewed by users. Through the route planning module, the inspection route of the drone can be dynamically edited and modified, so that the system can promptly detect photovoltaic panels with frequent faults, thereby avoiding inspections according to a constant inspection route, resulting in long-term faults of the faulty photovoltaic panels, and in severe cases, causing the overall abnormality or even damage of the photovoltaic component string.

[0077] In one embodiment, the first image data is used to construct an orthophoto model of a photovoltaic power station, and the second image data is used to perform fault analysis of the photovoltaic power station.

[0078] Based on the above, in one embodiment of the present invention, the image data required to be captured by the drone includes the first image data used in the photovoltaic power station modeling stage, such as Figure 2 As shown, the photovoltaic power station is mapped as a whole by using a drone, and an orthophoto model of the photovoltaic power station is constructed based on the first image data. Figures 4 to 5 As shown, the image data taken by the drone is also used in the actual monitoring stage, that is, the fault analysis of the photovoltaic power station is performed through the second image data taken by the drone along the inspection route.

[0079] In one embodiment, the method for planning the inspection route of the drone is:

[0080] Acquiring first image data;

[0081] constructing an orthophoto model of the photovoltaic power station based on the first image data;

[0082] According to the orthophoto model, an electronic map of the photovoltaic power station is obtained after processing;

[0083] Based on the electronic map, obtain the position coordinates of the center point of the photovoltaic component string;

[0084] The center point of the PV panel string is used as the waypoint, and all waypoints are connected to generate an inspection route.

[0085] Before the drone inspects the photovoltaic power station, it is necessary to customize the inspection route of the drone. The drone performs the inspection task according to the customized inspection route. Generally speaking, the inspection route is fixed, which is convenient for management and prediction of inspection time. In one embodiment of the present invention, Figures 2 to 3 As shown, before constructing the inspection route, the first image data taken by the drone is obtained first, and the orthophoto model of the photovoltaic power station is constructed based on the first image data. Then, based on the orthophoto model, it is fused and superimposed on the satellite map to form an electronic map, and then the electronic map is uploaded to the server deployed by the system for use by the drone. When the drone inspects according to the inspection route, it will read the electronic map on the server for navigation; in order to generate the inspection route, it is also necessary to identify the photovoltaic component string displayed on the electronic map based on the processed electronic map, and use the center point of the photovoltaic component string as the waypoint when the drone inspects, that is, the drone uses the waypoints as indicators when inspecting to determine whether the drone has missed inspections. After all the waypoints are obtained through processing, all the waypoints are connected in sequence to generate the inspection route;

[0086] In addition, if Figure 7 As shown, the inspection route planning also needs to pay attention to:

[0087] 1) Heading overlap rate: When the drone is taking pictures along a route, the overlap rate of the two photos is usually set to 70%-80% in actual operations.

[0088] 2) Lateral overlap rate: When a drone takes photos on two parallel routes, the overlap rate of two photos with similar positions is generally set to 60%-70% in actual operations.

[0089] In one embodiment, the data processing and analysis module includes a first deep learning model, and the first deep learning model is used to process the electronic map; the method for processing the electronic map is:

[0090] Inputting the electronic map into the first deep learning model;

[0091] Segment each photovoltaic panel in the electronic map based on instance segmentation technology, and

[0092] Number each photovoltaic panel after segmentation;

[0093] The number is used to characterize the logical position of the photovoltaic panel in the electronic map; the number is created based on the logical hierarchy of new power station-new area-new sub-array-new bracket-new branch-new component.

[0094] Based on the above, in order to obtain each waypoint in the electronic map, that is, corresponding to each photovoltaic component string displayed in the electronic map, the electronic map needs to be processed. In one embodiment of the present invention, by training a first deep learning model, the input electronic map can be processed based on the model, specifically, the photovoltaic panels displayed in the electronic map are segmented, and each photovoltaic panel generated after the segmentation is numbered. The purpose of numbering is to facilitate the data analysis module to analyze the second image data taken by the drone. If there is a fault, the number of the photovoltaic panel with the fault can be queried for prompting. In addition, the segmented photovoltaic panels can be summarized into photovoltaic component strings according to the series connection conditions, using the photovoltaic component string as a unit, and based on the center point of the photovoltaic component string as a waypoint, it is convenient for the planning of the drone inspection route; in addition, if Figures 4 to 6 As shown, the numbering creation logic is created according to the logical hierarchy of new power station-new area-new sub-array-new bracket-new branch-new component. For example, 01-01HL-1011-11-11_12 represents 01 area, 01HL sub-array, 1011 bracket, 11 branch, 11_12 component of a power station. Based on the determination of the above numbering, in the subsequent monitoring of the photovoltaic power station by the drone, the location of the faulty photovoltaic panel can be more accurately found, providing guidance and direction for the maintenance personnel.

[0095] In one embodiment, the method for generating an inspection route includes:

[0096] Obtaining historical analysis results, wherein the historical analysis results include historical faulty photovoltaic panels and corresponding fault types and fault probabilities;

[0097] Taking the photovoltaic module string as the basic group j, construct a set S to be inspected that includes all basic groups j;

[0098] Count the average failure rate s of each basic group j in the inspection set S j , according to the formula:

[0099]

[0100] Among them, T represents the number of inspections corresponding to one inspection cycle, t i，j represents the number of failures of any photovoltaic panel in basic group j within a patrol cycle, n represents the total number of photovoltaic panels in basic group j, and s j represents the average failure rate of PV panels within the base group j;

[0101] The average failure rate s j Sort all basic groups j and select the average failure rate s j The basic group j with a failure rate greater than the threshold value is obtained, and the priority inspection set S is obtained. z ;

[0102] Set the priority inspection set S z Delete from the waiting inspection set S to obtain the non-priority inspection set S w ;

[0103] Based on the priority inspection set S z , generate the first inspection route;

[0104] Based on non-priority inspection set S w , generate the second inspection route;

[0105] The first inspection route and the second inspection route are processed and merged to obtain an inspection route.

[0106] Based on the above, since the inspection route of the drone in the prior art connects all waypoints in sequence and is generally fixed or constant, if there are frequently faulty photovoltaic panels in the photovoltaic power station, and the faulty photovoltaic panel is located in the middle or end of the inspection route, it will take a long time for the drone to find the faulty photovoltaic panel when performing the inspection task, which will cause the photovoltaic panel failure to last for a long time, which may cause the entire photovoltaic component string to be abnormal or damaged, causing greater losses; based on this, in one embodiment of the present invention, considering the inspection priority of frequently faulty photovoltaic panels, by obtaining historical analysis results, the faulty photovoltaic panel number, fault type, and fault frequency in the historical analysis results can be understood, and based on the information obtained, the average failure rate s is calculated for the basic group j based on the photovoltaic component string. j The calculation of the average failure rate s corresponding to all basic groups j is calculated j , all the basic groups j in the inspection set S can be sorted. It can be understood that the average failure rate s jThe larger the value is, the more abnormal the photovoltaic component string corresponding to the basic group j is in the historical analysis results, such as hot spots, obstruction, damage and other problems. Therefore, before the upcoming inspection task begins, it is also necessary to give priority to the inspection of the photovoltaic component string to avoid the drone taking a long time to arrive at the photovoltaic component string, resulting in a longer duration of the fault and greater losses. Secondly, the purpose of taking the photovoltaic component string as the unit instead of the photovoltaic panel is that the drone inspection takes the center point of the photovoltaic component string as the waypoint. Therefore, when considering the inspection priority, the photovoltaic component string is used as the unit. When obtaining the average failure rate s of all basic groups j in the set to be inspected S, j After that, the average failure rate s can be filtered by sorting j The basic group j with a failure rate greater than the threshold value is obtained, and the priority inspection set S is obtained. z ; Priority inspection set S z It can be understood as the set of photovoltaic component strings that need to be inspected first, and the remaining set of photovoltaic component strings does not need to be inspected first, which is recorded as the non-priority inspection set S w , according to the obtained priority inspection set S z , non-priority inspection set S w The first inspection route and the second inspection route are customized respectively, and the first inspection route and the second inspection route are merged through data processing to obtain the inspection route of the drone;

[0107] It is worth noting that the first inspection route takes precedence over the second inspection route, that is, when the drone performs an inspection task, its starting point starts from the first inspection route. The above merging of the first inspection route with the second inspection route only processes the connection between the final waypoint of the first inspection route and the initial waypoint of the second inspection route.

[0108] In one embodiment, the method for generating an inspection route further includes:

[0109] Get the waypoint corresponding to each basic group j in the set to be inspected S, denoted as P;

[0110] Where P = {P1, P2, ..., P N};

[0111] Set the priority inspection set S z The waypoint corresponding to the basic group j contained in is denoted as ΔP;

[0112] Where ΔP={P p1 , P p2 ,…,P pM}(M≤N);

[0113] Set the objective function, expressed as follows:

[0114]

[0115] Among them, C ij Indicates that from waypoint P i To waypoint P j Distance; K ij is a decision variable. When there is a path from i to j, then K ij =1, otherwise 0;

[0116] Optimize the objective function, subject to the following constraints:

[0117]

[0118] Among them, u pi is the priority inspection set S z Waypoint P pi Inspection order; u i It is waypoint P i Inspection order; u j It is waypoint P j The inspection order;

[0119] A constraint function that requires each waypoint to be passed and left only once;

[0120] The inspection route is obtained by calculating the above objective function based on the simulated annealing algorithm.

[0121] It is worth noting that the customized first inspection route and second inspection route do not specify how the first inspection route and the second inspection route connect the scattered waypoints, which can be understood as the priority inspection set S z The connection rules of the multiple waypoints included in the method are not described in detail. According to the implementation method in the prior art, they are generally connected in sequence. However, due to the priority inspection set S z The waypoints in are not necessarily adjacent, so connecting them in order may result in a large mileage problem for the first inspection route. Similarly, the mileage of the second inspection route also has a large problem. Based on the above, in one embodiment of the present invention, in order to obtain the shortest mileage of the first inspection route and the second inspection route, it is necessary to construct a priority inspection set S based on the priority inspection set S. z , non-priority inspection set S w The objective function is used to solve the shortest mileage of the first inspection route and the second inspection route. z , non-priority inspection set S w The objective function can be simplified to solve the mileage objective function of traveling all grid points in a grid; for example:

[0122] There is a 5x5 grid where the priority points are (0, 0), (1, 1), (2, 2); we need to find a path so that all points are visited in sequence and the total path is the shortest. Define the cost matrix C, where C ij From point P i To point P j The cost matrix is ​​as follows:

[0123]

[0124] Calculate the shortest path through the points (0, 0), (1, 1), and (2, 2):

[0125] The path is determined to be (0, 0)-(1, 1)-(2, 2), which corresponds to the priority order of the three points;

[0126] According to the given cost matrix C, the above path is divided into two segments:

[0127] The cost C from (0,0) to (1,1) 01 ,as well as

[0128] The cost C from (1, 1) to (2, 2) 12 ;

[0129] Find the corresponding value in the cost matrix C;

[0130] According to the above, the cost C from (0,0) to (1,1) is 01 :

[0131] C 01 =C[0][1]=1

[0132] The cost C from (1, 1) to (2, 2) 12 :

[0133] C 12 =C[1][2]=1

[0134] Calculate the total cost, that is, the cost C 01 Cost C 12 Add and sum:

[0135] Total cost = C 01 +C 12 =1+1=2

[0136] Based on the above, construct a priority inspection set S z , non-priority inspection set S w The objective function is used to solve the shortest mileage of the first inspection route and the second inspection route. Based on the solution of the objective function, the priority inspection set S can be obtained. z, non-priority inspection set S w The shortest mileage, that is, the shortest inspection route, can be considered simultaneously with how to shorten the mileage in addition to considering the PV module strings that need to be inspected first, avoiding problems such as complex drone heading planning and unnecessary mileage during drone inspections.

[0137] In one embodiment, the data processing and analysis module further includes a second deep learning model, and the second deep learning model is used to analyze the image data and output the analysis results; the method for processing and analyzing the image data is:

[0138] receiving second image data;

[0139] Inputting the second image data into a second deep learning model for analysis;

[0140] Based on the second deep learning model, identify whether the photovoltaic panel is faulty, and if there is a fault, determine the fault type and obtain the serial number of the faulty photovoltaic panel, and generate a fault event in combination with the corresponding fault image data; if there is no fault, do not perform any operation;

[0141] Based on all the acquired fault events, they are summarized according to the fault type and the analysis results are generated;

[0142] The fault image data is second image data that is identified as having a fault.

[0143] For the second image data acquired during the actual inspection phase of the drone, in one embodiment of the present invention, the second image data is acquired by a data acquisition module, and then the second image data is preprocessed by a data processing module, including denoising and other operations, and then the preprocessed second image data is uploaded to the data analysis module, and the second image data is analyzed by a second deep learning model in the data analysis module. The second deep learning model is constructed based on a deep learning algorithm, and the second deep learning model is trained and tested using historical analysis results. When the second image data acquired in real time is input into the second deep learning model, the model can be used to analyze the second image data to determine whether there is a fault in the photovoltaic panel. It is worth noting that, if Figures 4 to 6As shown, when fault defects such as hot spots, damage and occlusion are identified in the captured image, various fault types can be subdivided through image processing. After the fault type is determined, the number of the faulty photovoltaic panel must be obtained. Since the second image data represents not only one photovoltaic panel, but a photovoltaic component string composed of multiple photovoltaic panels, the waypoint corresponding to the second image data can be used to obtain the photovoltaic component string in which the faulty photovoltaic panel exists. Then, based on the position relationship and the heading of the drone, the number of the faulty photovoltaic panel can be determined, and the number and fault type of the faulty photovoltaic panel are generated together to generate a fault event. After all the second image data are analyzed, the generated fault events are summarized to obtain the analysis report corresponding to this inspection task, and the analysis report is displayed on the deployer page for easy viewing by operators.

[0144] It is worth noting that the analysis report should also carry a fault image associated with the fault event, and the fault image is included in the second image data.

[0145] In one embodiment, if Figure 6 As shown, the method for outputting a fault prompt according to the analysis result is:

[0146] Get analysis results;

[0147] Extract the serial number of the faulty photovoltaic panel from the analysis results;

[0148] According to the number of the faulty photovoltaic panel, the corresponding position is marked in the electronic map to provide a fault prompt.

[0149] In one embodiment, it further includes a data storage module for storing all image data and analysis results and fault events output by the storage data analysis module;

[0150] The data storage module is also used for data indexing and retrieval, allowing users to query and filter by time and fault type.

[0151] A method for monitoring a photovoltaic power station based on an unmanned aerial vehicle comprises the following steps: S1: determining a photovoltaic power station to be inspected;

[0152] S2: Obtain an electronic map of the photovoltaic power station to be inspected;

[0153] S3: Planning the inspection route based on the electronic map of the photovoltaic power station to be inspected;

[0154] S4: Perform inspection tasks based on the inspection route and upload image data;

[0155] S5: Analyze the image data to obtain analysis results, and output a fault prompt according to the analysis results.

[0156] The above shows and describes the basic principles, main features and advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A photovoltaic power station monitoring system based on drones, characterized in that: include: A data acquisition module, used to acquire image data taken by the drone, including first image data and second image data; A data processing and analysis module, used for preprocessing the image data, including preprocessing the first image data and the second image data, and analyzing the second image data, and outputting the analysis result; Identification prompt module, used to output fault prompt according to the analysis result; The route planning module is used to plan the inspection route of the drone and edit and modify the inspection route of the drone based on historical analysis results.

2. The photovoltaic power station monitoring system based on drone according to claim 1, characterized in that: The first image data is used to construct an orthophoto model of a photovoltaic power station, and the second image data is used to perform fault analysis of the photovoltaic power station.

3. The photovoltaic power station monitoring system based on drone according to claim 1, characterized in that: The method for planning the inspection route of the drone is: Acquiring first image data; constructing an orthophoto model of the photovoltaic power station based on the first image data; According to the orthophoto model, an electronic map of the photovoltaic power station is obtained after processing; Based on the electronic map, obtain the position coordinates of the center point of the photovoltaic component string; The center point of the PV panel string is used as the waypoint, and all waypoints are connected to generate an inspection route.

4. The photovoltaic power station monitoring system based on drone according to claim 1, characterized in that: The data processing and analysis module includes a first deep learning model, and the first deep learning model is used to process the electronic map; the method for processing the electronic map is: Inputting the electronic map into the first deep learning model; Segment each photovoltaic panel in the electronic map based on instance segmentation technology, and Number each photovoltaic panel after segmentation; The number is used to characterize the logical position of the photovoltaic panel in the electronic map; the number is created based on the logical hierarchy of new power station-new area-new sub-array-new bracket-new branch-new component.

5. The photovoltaic power station monitoring system based on drone according to claim 3 is characterized by: The method for generating an inspection route comprises: Obtaining historical analysis results, wherein the historical analysis results include historical faulty photovoltaic panels and corresponding fault types and fault probabilities; Taking the photovoltaic module string as the basic group j, construct a set S to be inspected that includes all basic groups j; Count the average failure rate s of each basic group j in the inspection set S j , according to the formula: Among them, T represents the number of inspections corresponding to one inspection cycle, t i，j represents the number of failures of any photovoltaic panel in basic group j within a patrol cycle, n represents the total number of photovoltaic panels in basic group j, and s j represents the average failure rate of PV panels within the base group j; The average failure rate s j Sort all basic groups j and select the average failure rate s j The basic group j with a failure rate greater than the threshold value is obtained, and the priority inspection set S is obtained. z ; Set the priority inspection set S z Delete from the waiting inspection set S to obtain the non-priority inspection set S w ; Based on the priority inspection set S z , generate the first inspection route; Based on non-priority inspection set S w , generate the second inspection route; The first inspection route and the second inspection route are processed and merged to obtain an inspection route.

6. The photovoltaic power station monitoring system based on drone according to claim 3 is characterized by: The method for generating an inspection route also includes: Get the waypoint corresponding to each basic group j in the set to be inspected S, denoted as P; Where P = {P1, P2, ..., P N }; Set the priority inspection set S z The waypoint corresponding to the basic group j contained in is denoted as ΔP; Where ΔP={P p1 , P p2 , ..., P pM }(M≤N); Set the objective function, expressed as follows: Among them, C ij Indicates that from waypoint P i To waypoint P j The distance K ij is a decision variable. When there is a path from i to j, then K ij =1, otherwise 0; Optimize the objective function, subject to the following constraints: Among them, u pi is the priority inspection set S z Waypoint P pi Inspection order of u i is waypoint P i Inspection order of u j is waypoint P j The inspection order; A constraint function that requires each waypoint to be passed and left only once; The inspection route is obtained by calculating the above objective function based on the simulated annealing algorithm.

7. The photovoltaic power station monitoring system based on drone according to claim 1, characterized in that: The data processing and analysis module further includes a second deep learning model, which is used to analyze the image data and output the analysis results; the method for processing and analyzing the image data is: receiving second image data; Inputting the second image data into a second deep learning model for analysis; Identify whether the photovoltaic panel is faulty based on the second deep learning model, and if a fault exists, determine the fault type and obtain the serial number of the faulty photovoltaic panel, and generate a fault event in combination with the corresponding fault image data; If there is no fault, no operation is performed; Based on all the acquired fault events, they are summarized according to the fault type and the analysis results are generated; The fault image data is second image data that is identified as having a fault.

8. The photovoltaic power station monitoring system based on drone according to claim 1, characterized in that: The method for outputting a fault prompt according to the analysis result is: Get analysis results; Extract the serial number of the faulty photovoltaic panel from the analysis results; According to the number of the faulty photovoltaic panel, the corresponding position is marked in the electronic map to provide a fault prompt.

9. The photovoltaic power station monitoring system based on drone according to claim 1, characterized in that: It also includes a data storage module for storing all image data and analysis results and fault events output by the storage data analysis module; The data storage module is also used for data indexing and retrieval, allowing users to query and filter by time and fault type.

10. A method for monitoring a photovoltaic power station based on an unmanned aerial vehicle, characterized in that: The following steps are involved: S1: Determine the photovoltaic power station to be inspected; S2: Obtain an electronic map of the photovoltaic power station to be inspected; S3: Planning the inspection route based on the electronic map of the photovoltaic power station to be inspected; S4: Perform inspection tasks based on the inspection route and upload image data; S5: Analyze the image data to obtain analysis results, and output a fault prompt according to the analysis results.

Images