A power failure image recognition method and system

Abstract

The application discloses a power failure image recognition method and system, compares the real-time power equipment image data with the historical power equipment image data, if the similarity in the comparison result of the power equipment image data is lower than a preset value, starts a real-time power equipment image data noise reduction strategy, then calls dynamic object feature data corresponding to the standard image background video data, collects power equipment operation data, performs image analysis on the noise reduction processing result, generates a power failure image recognition failure result in combination with the collected power equipment operation data, reduces noise of image data collected under a severe natural environment according to dynamic object features, analyzes the image data after noise reduction in combination with data collected by a sensor, and obtains a power equipment failure analysis result. The power equipment image analysis accuracy can be improved under a severe natural environment.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power system operation and maintenance, and particularly relates to a power fault image recognition method and system. BACKGROUND

[0002] With the wide application of wind power equipment in the power generation field, the number of wind power equipment also increases. Due to the particularity of the location of wind power equipment, which is mostly located in remote areas, remote monitoring technology is needed in wind power operation and maintenance. In actual application scenarios, power fault image recognition is widely used. The image of the power equipment is collected, and then the collected image data is analyzed to obtain the analysis result of whether the power equipment has a fault.

[0003] In actual application process, power fault image recognition has recognition error in some specific scenarios. Since the practical scenario of wind power equipment is in a region with strong wind, and there is no shelter around the wind power equipment, the region is mostly open land. In the region with strong wind, the air flow is fast. In stormy weather, sandstorm weather and snow weather environment, sundries (sand, snow, etc.) will appear in the video image in the process of video image collection under high wind speed, which causes the existing image recognition to collect video images with unclear situation. After the problem occurs in the video image data source, the accuracy of the power equipment in analyzing the fault through the video image data will also be affected. Therefore, the power fault image recognition method and system are developed. SUMMARY

[0004] The purpose of the present application is to provide a power fault image recognition method and system to solve the problem that the existing wind power equipment power fault image analysis in the natural environment of severe weather, the video image data collection is affected by the natural environment, which causes the power fault analysis to have error.

[0005] In a first aspect, the present application provides a power fault image recognition method, comprising:

[0006] Obtaining power equipment image data, the power equipment image collection equipment collects image data according to a specified period of time, marks the image data according to the collection order, obtains historical power equipment image data and real-time power equipment image data;

[0007] Comparing the real-time power equipment image data with the historical power equipment image data to obtain a power equipment image device comparison result. If the similarity in the power equipment image device comparison result is lower than a preset value, a real-time power equipment image data noise reduction strategy is started;

[0008] Collect standard image background video data, identify dynamic data in the standard image background data, obtain dynamic feature identification results, and extract features from the identification results to obtain dynamic object feature data;

[0009] Unify the real-time power equipment image data with the standard image background video data in time periods, and then retrieve the dynamic object feature data corresponding to the standard image background video data;

[0010] Collect power equipment operation data, perform image analysis on the noise reduction processing results, and generate power fault image recognition fault results in combination with the collected power equipment operation data.

[0011] Further, the power equipment image data is obtained by image equipment according to a specified periodic time, the image data is marked according to the collection sequence, and the historical power equipment image data and the real-time power equipment image data are obtained, including:

[0012] The power equipment image data includes first-view power equipment image data, second-view power equipment image data, and third-view power equipment image data;

[0013] The image collection cycle frequencies of the first-view power equipment, the second-view power equipment, and the third-view power equipment are adjusted to be consistent;

[0014] The image data collected by the first-view power equipment, the second-view power equipment, and the third-view power equipment within the same time period range is sequentially marked.

[0015] Further, the real-time power equipment image data is compared with the historical power equipment image data to obtain a power equipment image device comparison result, and if the similarity in the power equipment image device comparison result is lower than a preset value, a real-time power equipment image data noise reduction strategy is started, including:

[0016] The real-time power equipment image data is compared with the historical power equipment image data, and the comparison information includes image data collection time and image data similarity comparison;

[0017] The historical power equipment image data within 60 seconds of the collection time of the real-time power equipment image data is retrieved;

[0018] The historical power equipment image data is compared with the historical power equipment image data, and the extracted feature data is compared to obtain a power equipment image device comparison result.

[0019] Further, standard image background video data is collected, dynamic data in the standard image background data is identified to obtain a dynamic feature identification result, feature extraction is performed on the identification result to obtain dynamic object feature data, including;

[0020] The standard image background video data collection terminal is consistent with the image data collection time of the power equipment image collection terminal;

[0021] The background image color of the standard image background video data collection terminal is the color of the outer wall of the wind power equipment, and dynamic object feature extraction is performed on the image data collected by the standard image background video data collection terminal based on the color of the outer wall of the wind power equipment;

[0022] The dynamic object feature extraction results of the standard image background video data collected at different time periods are subjected to cluster analysis to obtain dynamic object feature data.

[0023] Further, power equipment operation data is collected, image analysis is performed on the noise reduction processing result, and the power equipment operation data collected is combined to generate a power fault image recognition fault result, including;

[0024] The collected power equipment operation data includes sensor data collected by a sensor matrix of each operating component of the power equipment, and the sensor data collected by the sensor matrix is analyzed to obtain a monitoring equipment component operating state;

[0025] Image analysis is performed on the noise reduction processing result to obtain a noise reduction processing analysis result, the equipment state in the noise reduction processing analysis result is matched with the monitoring equipment component operating state, and if the equipment operating state is consistent, a power fault image recognition fault result is generated;

[0026] If the equipment operating state is inconsistent, re-video image data collection and analysis are performed on the inconsistent power equipment position.

[0027] In a second aspect, a power fault image recognition system includes a data collection terminal, a user operation terminal, and a server terminal. The server terminal is connected to the data collection terminal and the user operation terminal. The data collection terminal transmits collected data to the server terminal. The server terminal analyzes the data and transmits the analysis result to the user operation terminal.

[0028] The server end acquires power equipment image data, the power equipment image data is collected by an image device at a specified period of time, the image data is marked according to the collection sequence, the historical power equipment image data and the real-time power equipment image data are obtained, the real-time power equipment image data is compared with the historical power equipment image data, the power equipment image device comparison result is obtained, if the approximation in the power equipment image device comparison result is lower than a preset value, the real-time power equipment image data noise reduction strategy is started, the standard image background video data is collected, the dynamic data in the standard image background data is identified, the dynamic feature identification result is obtained, the feature extraction is performed on the identification result, the dynamic object feature data is obtained, the real-time power equipment image data is unified with the standard image background video data in a time period, then the dynamic object feature data corresponding to the standard image background video data is called, the power equipment operation data is collected, the noise reduction processing result is analyzed, and the power fault image recognition fault result is generated in combination with the collected power equipment operation data.

[0029] The power fault image recognition method and system provided by the application compare the real-time power equipment image data with the historical power equipment image data to obtain a power equipment image device comparison result, if the approximation in the power equipment image device comparison result is lower than a preset value, the real-time power equipment image data noise reduction strategy is started, the real-time power equipment image data is unified with the standard image background video data in a time period, then the dynamic object feature data corresponding to the standard image background video data is called, the power equipment operation data is collected, the noise reduction processing result is analyzed, and the power fault image recognition fault result is generated in combination with the collected power equipment operation data, the application extracts the dynamic object feature from the image data collected in the standard background state under the natural environment in a harsh state, the image data collected in the natural environment in a harsh state is reduced according to the dynamic object feature, the image data after reduction is analyzed and combined with the data collected by the sensor to obtain the power equipment fault analysis result, and the effect of improving the power equipment image analysis accuracy in the natural environment in a harsh state is realized. BRIEF DESCRIPTION OF DRAWINGS

[0030] In order to more clearly illustrate the technical scheme of the application, the drawings needed in the embodiments will be briefly introduced as follows, and obviously, other drawings can be obtained by those skilled in the art without creative labor.

[0031] Figure 1 A power fault image recognition method and system basic framework schematic diagram provided by the embodiment of the application. DETAILED DESCRIPTION

[0032] In order to make the objects, technical solutions and advantages of the present application clearer, the technical solutions of the present application will be described clearly and completely below in connection with specific embodiments of the present application and corresponding drawings. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application. The technical solutions provided by the embodiments of the present application will be described in detail below in connection with the drawings.

[0033] Referring to Figure 1 , in a first aspect, the present application provides a power failure image recognition method, comprising:

[0034] In step S101, power equipment image data is acquired, the power equipment image data is acquired by an image device at a specified periodic time, the image data is marked according to the acquisition order, and historical power equipment image data and real-time power equipment image data are obtained;

[0035] The power equipment image data is acquired by an image device at a specified periodic time, for example, picture data is acquired once every 2 seconds, or the image data of the power equipment is acquired in real time, the acquired data is stored in real time, and uploaded to a cloud server.

[0036] In step S102, the real-time power equipment image data is compared with the historical power equipment image data, a power equipment image device comparison result is obtained, if the approximation in the power equipment image device comparison result is lower than a preset value, a real-time power equipment image data noise reduction strategy is started;

[0037] The historical image data of the power equipment is stored and the corresponding data is called to compare with the real-time power equipment image data, the compared data is analyzed, whether the existing environment of the power equipment is an abnormal environment is obtained, if the abnormal natural environment is present, the real-time acquired picture will be greatly different from the historical image data, and the approximation is low, then the power equipment image data noise reduction strategy is started.

[0038] In step S103, standard image background video data is acquired, dynamic data in the standard image background data is identified, a dynamic feature recognition result is obtained, the feature recognition result is extracted, and dynamic object feature data is obtained;

[0039] The video equipment acquisition terminal of the application uses a camera to face the outer wall of the wind power equipment to collect image data of the position of the outer wall of the wind power equipment. The outer wall of the wind power equipment is of a uniform color, so that the uniform color of the outer wall facilitates feature extraction of dynamic objects in a case of a severe natural environment, and therefore standard image background video data is collected.

[0040] In step S104, the real-time power equipment image data is time-unified with the standard image background video data, and then the dynamic object feature data corresponding to the standard image background video data is called.

[0041] The extracted features need to be ensured to be features of the same period. Since the natural environment is dynamically changing, the features of dynamic objects in different periods are different, so the real-time image data and the standard image background data for comparison need to be time-unified.

[0042] In step S105, power equipment operation data is collected, the denoising result is subjected to image analysis, and the power fault image recognition fault result is generated in combination with the collected power equipment operation data.

[0043] The technical scheme of the application extracts dynamic object features from image data collected in a standard background state under a severe natural environment, performs denoising on the image data collected under the severe natural environment according to the dynamic object features, analyzes the denoised image data, and obtains a power equipment fault analysis result in combination with data collected by a sensor. The problem of errors in power fault image analysis of existing wind power equipment due to the influence of the natural environment on video image data collection in severe weather conditions is solved.

[0044] Specifically, power equipment image data is obtained, the power equipment image data is collected by an image device at a specified periodic time, the image data is marked in a collection order, and historical power equipment image data and real-time power equipment image data are obtained, including:

[0045] The power equipment image data includes first-view power equipment image data, second-view power equipment image data, and third-view power equipment image data.

[0046] The image collection periodic frequencies of the first-view power equipment, the second-view power equipment, and the third-view power equipment are adjusted to be consistent.

[0047] The image data collected by the first-view power equipment, the second-view power equipment, and the third-view power equipment within the same period range is sequentially marked.

[0048] Specifically, the real-time power equipment image data is compared with the historical power equipment image data to obtain a power equipment image comparison result. If the approximation degree in the power equipment image comparison result is lower than a preset value, a real-time power equipment image data noise reduction strategy is started, including:

[0049] The real-time power equipment image data is compared with the historical power equipment image data. The comparison information includes image data collection time and image data approximation comparison.

[0050] The historical power equipment image data within 60 seconds of the real-time power equipment image data collection time is retrieved.

[0051] The historical power equipment image data is compared with the historical power equipment image data for feature extraction, and the extracted feature data is compared to obtain a power equipment image comparison result.

[0052] Specifically, standard image background video data is collected. Dynamic data in the standard image background data is identified to obtain a dynamic feature identification result. The identification result is extracted for feature extraction to obtain dynamic object feature data, including:

[0053] The standard image background video data collection terminal is consistent with the image data collection time of the power equipment image collection terminal.

[0054] The background image color of the standard image background video data collection terminal is the color of the outer wall of the wind power equipment. The image data collected by the standard image background video data collection terminal is extracted for dynamic object feature extraction based on the color of the outer wall of the wind power equipment.

[0055] The dynamic object feature extraction results of the standard image background video data collected at different time periods are clustered and analyzed to obtain dynamic object feature data.

[0056] Specifically, power equipment operation data is collected. The noise reduction processing result is analyzed for image analysis, and the power equipment operation data is combined to generate a power fault image identification fault result, including:

[0057] The power equipment operation data includes sensor data collected by a sensor matrix of each operating component of the power equipment. The sensor data collected by the sensor matrix is analyzed to obtain a monitoring equipment component operating state.

[0058] The noise reduction processing result is analyzed for image analysis to obtain a noise reduction processing analysis result. The equipment state in the noise reduction processing analysis result is matched with the monitoring equipment component operating state. If the equipment operating state is consistent, a power fault image identification fault result is generated.

[0059] If the device operating state is inconsistent, the inconsistent power equipment position is re-video image data collection and analysis.

[0060] In a second aspect, a power failure image recognition system includes a data collection terminal, a user operation terminal, and a server terminal. The server terminal is connected to the data collection terminal and the user operation terminal. The data collection terminal transmits collected data to the server terminal. The server terminal analyzes data and transmits analysis results to the user operation terminal.

[0061] The server terminal acquires power equipment image data. The power equipment image data is collected by an image device at a specified periodic time. The image data is marked according to the collection order to obtain historical power equipment image data and real-time power equipment image data. The real-time power equipment image data is compared with the historical power equipment image data to obtain a power equipment image device comparison result. If the similarity in the power equipment image device comparison result is lower than a preset value, a real-time power equipment image data noise reduction strategy is started. Standard image background video data is collected. Dynamic data in the standard image background data is identified to obtain a dynamic feature identification result. The feature identification result is feature extracted to obtain dynamic object feature data. The real-time power equipment image data is time-unified with the standard image background video data. The dynamic object feature data corresponding to the standard image background video data is called. Power equipment operation data is collected. The noise reduction processing result is image analyzed. The power failure image recognition failure result is generated in combination with the collected power equipment operation data.

[0062] The application provides a power failure image recognition method and system, image data of power equipment is acquired, the image data of power equipment is collected by image equipment according to a specified period of time, the image data is marked according to the collection sequence, historical image data of power equipment and real-time image data of power equipment are obtained, the real-time image data of power equipment is compared with the historical image data of power equipment, a comparison result of image equipment of power equipment is obtained, if the approximation in the comparison result of image equipment of power equipment is lower than a preset value, a real-time image data noise reduction strategy of power equipment is started, standard image background video data is collected, dynamic data in the standard image background data is recognized, a dynamic feature recognition result is obtained, feature extraction is performed on the recognition result, dynamic object feature data is obtained, the real-time image data of power equipment is unified with the standard image background video data in a time period, then the dynamic object feature data corresponding to the standard image background video data is called, power equipment operation data is collected, the noise reduction processing result is subjected to image analysis, and the power failure image recognition failure result is generated by combining the collected power equipment operation data, the technical scheme of the application extracts dynamic object features in image data collected in a standard background state under a severe natural environment, the image data collected under the severe natural environment is subjected to noise reduction according to the dynamic object features, the image data after noise reduction is analyzed and combined with data collected by a sensor, and a power equipment failure analysis result is obtained. The existing wind power equipment power failure image analysis under severe natural environment is solved, video image data collection is affected by the natural environment, and the power failure analysis has an error.

[0063] Those skilled in the art can clearly understand that the technology in the embodiments of the application can be realized by means of software and necessary general hardware platforms. Based on such understanding, the technical solutions in the embodiments of the application can be embodied in the form of a software product, and the computer software product can be stored in a storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, etc., and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the method described in the embodiments of the application or some parts of the embodiments. The above-described embodiments of the application do not constitute a limitation on the protection scope of the application.

Claims

1. A power failure image recognition method characterized by, Comprise; Acquire power equipment image data, and the power equipment image acquisition equipment carries out image data collection according to specified period time, and image data is marked according to collection order, obtains historical power equipment image data and real-time power equipment image data; The real-time power equipment image data is compared with the historical power equipment image data, and the power equipment image equipment comparison result is obtained, if the approximation in the power equipment image equipment comparison result is lower than the preset value, then the real-time power equipment image data noise reduction strategy is started; Collect standard image background video data, identify dynamic data in the standard image background video data, obtain dynamic feature identification result, and the feature identification result is extracted, and dynamic object feature data is obtained; The real-time power equipment image data is unified with the standard image background video data in time period, and then the dynamic object feature data corresponding to the standard image background video data is called; Collect power equipment operation data, image analysis is carried out to the noise reduction processing result, and the power fault image recognition fault result is generated in combination with the collected power equipment operation data; The real-time power equipment image data is compared with the historical power equipment image data, and the power equipment image equipment comparison result is obtained, if the approximation in the power equipment image equipment comparison result is lower than the preset value, then the real-time power equipment image data noise reduction strategy is started, comprising; The real-time power equipment image data is compared with the historical power equipment image data, and the comparison information includes image data collection time, image data approximation comparison; The historical power equipment image data within 60 seconds of the real-time power equipment image data collection time is called; The historical power equipment image data is extracted with the historical power equipment image data, and the extracted feature data is compared, and the power equipment image equipment comparison result is obtained; Collect standard image background video data, identify dynamic data in the standard image background video data, obtain dynamic feature identification result, and the feature identification result is extracted, and dynamic object feature data is obtained, comprising; The standard image background video data acquisition terminal is consistent with the image data collection time of the power equipment image acquisition terminal; The background image color of the standard image background video data acquisition terminal is the color of the outer wall of the wind power equipment, and the image data collected by the standard image background video data acquisition terminal is extracted based on the color of the outer wall of the wind power equipment; The dynamic object feature extraction result of the standard image background video data collected in different time periods is clustered and analyzed, and the dynamic object feature data is obtained; Collect power equipment operation data, image analysis is carried out to the noise reduction processing result, and the power fault image recognition fault result is generated in combination with the collected power equipment operation data, comprising; The power equipment operation data includes sensor data collected by a sensor matrix of each operating component of the power equipment. The sensor data collected by the sensor matrix is analyzed to obtain the operating state of the monitored equipment components. The noise reduction processing result is analyzed to obtain a noise reduction processing analysis result. The equipment state in the noise reduction processing analysis result is matched with the operating state of the monitored equipment components. If the equipment operating states are consistent, a power fault image recognition fault result is generated. If the equipment operating states are inconsistent, inconsistent power equipment positions are re-analyzed by video image data collection.

2. The power failure image recognition method of claim 1, wherein, The power equipment image data is obtained. The power equipment image collection equipment collects image data at a specified periodic time. The image data is marked in the collection order to obtain historical power equipment image data and real-time power equipment image data, including: The power equipment image data includes first-view power equipment image data, second-view power equipment image data, and third-view power equipment image data. The image collection periods of the first-view power equipment, the second-view power equipment, and the third-view power equipment are adjusted to be consistent. The image data collected by the first-view power equipment, the second-view power equipment, and the third-view power equipment within the same time range is sequentially marked.

3. A power failure image recognition system for use in the power failure image recognition method according to any one of claims 1 to 2, characterized by The data collection terminal, the user operation terminal, and the server terminal are connected. The data collection terminal transmits the collected data to the server terminal. The server terminal analyzes the data and transmits the analysis result to the user operation terminal. The server terminal obtains power equipment image data. The power equipment image data is collected by an image device at a specified periodic time. The image data is marked in the collection order to obtain historical power equipment image data and real-time power equipment image data. The real-time power equipment image data is compared with the historical power equipment image data to obtain a power equipment image device comparison result. If the similarity in the power equipment image device comparison result is lower than a preset value, a real-time power equipment image data noise reduction strategy is started. Standard image background video data is collected. Dynamic data in the standard image background video data is identified to obtain a dynamic feature identification result. The feature identification result is feature-extracted to obtain dynamic object feature data. The real-time power equipment image data is time-unified with the standard image background video data. The dynamic object feature data corresponding to the standard image background video data is retrieved. Power equipment operation data is collected. The noise reduction processing result is analyzed by combining the collected power equipment operation data to generate a power fault image recognition fault result.

Images