Power data collection method, device, equipment and readable storage medium

Through dynamic adjustment of multimodal data and risk prediction model based on power lines, the problem of low resource utilization rate during power line inspection is solved, the risk prediction accuracy and acquisition efficiency are improved, and the adaptive collection of power data is realized Closed loop reduces inspection costs and time.

CN118604531BActive Publication Date: 2024-11-22CHINA SOUTHERN POWER GRID ARTIFICIAL INTELLIGENCE TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411081473.9
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-08
Publication Date
2024-11-22
Estimated Expiration
2044-08-08

AI Technical Summary

Technical Problem

During power line inspection, the long power line makes it difficult for fixed acquisition terminals to obtain electricity, drone inspections are time-consuming and labor-intensive, and data collection costs are high, which limits the inspection frequency, and defects and hidden dangers need to be discovered in time to avoid large-scale power outages.

Method used

Using multimodal data and preset risk prediction model based on power lines, determine risk prediction results, adjust data acquisition strategies, re-acquire target multimodal data, and update the risk prediction model based on actual faults and hidden dangers, dynamic Adjust data acquisition strategies to improve resource utilization.

Benefits of technology

It improves the utilization rate of power data acquisition resources, enhances the accuracy of risk prediction models, realizes dynamic adjustment of data acquisition strategies, forms a closed loop of adaptive power data acquisition, and reduces inspection costs and time.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118604531B_ABST
    Figure CN118604531B_ABST
Patent Text Reader

Abstract

The present application relates to a method, device, equipment and readable storage medium for collecting electric power data, and relates to the technical field of power line inspection. The method comprises: determining the risk prediction result based on the multimodal data of the power line and a preset risk prediction model; the multimodal data comprises image data, temperature data and electrical quantity data; re-collecting the target multimodal data according to the data collection strategy corresponding to the risk prediction result, updating the risk prediction model based on the actual fault, hidden danger elimination situation and the target multimodal data; obtaining the current multimodal data, performing risk prediction on the current multimodal data based on the updated risk prediction model, obtaining the risk detection result of the power line, and updating the data collection strategy according to the risk detection result, and the updated data collection strategy is used for multimodal data collection. The use of this method can improve the utilization rate of collection resources.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the technical field of power line inspection, and in particular to a power data collection method, device, equipment and readable storage medium. Background Art

[0002] With the development of the power industry, the scale of the power grid has expanded rapidly, and the demand for power systems in daily life has increased. Power inspection aims to detect the status of transmission equipment, timely discover equipment defects and safety hazards, and plays an important role in ensuring the normal operation of the power system. Safety inspection technology has been widely studied and applied at home and abroad. By combining technologies such as drones, mobile sensors, data analysis and predictive models, safety inspection technology can improve the safety and reliability of lines, reduce the risk of accidents and the possibility of failures, and is of great significance to the operation and maintenance of power systems.

[0003] However, the power lines are long, it is difficult for fixed data collection terminals to obtain electricity, drone inspections are time-consuming and labor-intensive, and the data collection cost is high, which limits the inspection frequency. Defects and hidden dangers must be discovered in time, otherwise they may lead to large-scale power outages and bring serious consequences, which puts forward requirements for the timeliness of inspections. Therefore, there is an urgent need for a power data collection method that can improve the utilization rate of collection resources. Summary of the invention

[0004] Based on this, it is necessary to provide a power data collection method, device, computer equipment, computer-readable storage medium and computer program product that can improve the utilization rate of collection resources in response to the above technical problems.

[0005] In a first aspect, the present application provides a method for collecting power data, comprising:

[0006] Determine the risk prediction result based on the multimodal data of the power line and the preset risk prediction model; the multimodal data includes image data, temperature data and electrical quantity data;

[0007] Re-collect the target multimodal data according to the data collection strategy corresponding to the risk prediction results, and update the risk prediction model based on the actual failure, hidden danger elimination situation and the target multimodal data;

[0008] The current multimodal data is obtained, and risk prediction is performed on the current multimodal data based on the updated risk prediction model to obtain risk detection results of the power lines, and the data acquisition strategy is updated according to the risk detection results. The updated data acquisition strategy is used for multimodal data acquisition.

[0009] In one embodiment, based on the multimodal data of the power line and a preset risk prediction model, determining the risk prediction result includes:

[0010] Using a preset identification and analysis system to identify hidden dangers on the multimodal data of the power line, and obtaining hidden danger identification results corresponding to at least one hidden danger of the power line;

[0011] The hidden danger identification result corresponding to at least one hidden danger is input into a preset risk prediction model to obtain a risk prediction result.

[0012] In one embodiment, re-collecting target multimodal data according to a data collection strategy corresponding to a risk prediction result includes:

[0013] Adjust the collection cycle of the collection equipment according to the data collection strategy corresponding to the risk prediction results;

[0014] The acquisition device is controlled to re-acquire the target multimodal data according to the adjusted acquisition cycle.

[0015] In one embodiment, updating the risk prediction model based on actual failures, hidden danger elimination conditions and target multimodal data includes:

[0016] Update risk prediction results based on target multimodal data and risk prediction models;

[0017] Check the correctness of the updated risk prediction results based on the actual failures and hidden danger elimination situations;

[0018] The correctness check results, actual faults and hidden danger elimination situations are input into the preset reinforcement learning model to obtain the parameter changes, and the risk prediction model is updated based on the parameter changes.

[0019] In one embodiment, the correctness check results, actual faults and hidden danger elimination situations are input into a preset reinforcement learning model to obtain parameter changes, including:

[0020] The reward and punishment function in the preset reinforcement learning model is used to pre-process the correctness check results, actual faults and hidden danger elimination situations to obtain pre-processing results;

[0021] The preprocessing results are converted into parameter changes based on the value function in the preset reinforcement learning model.

[0022] In one embodiment, the power data collection method further includes:

[0023] The multimodal data to be processed are collected according to the updated data collection strategy;

[0024] The updated risk prediction model is updated based on the multimodal data to be processed.

[0025] In a second aspect, the present application also provides a power data collection device, comprising:

[0026] An acquisition module is used to determine the risk prediction result based on the multimodal data of the power line and a preset risk prediction model; the multimodal data includes image data, temperature data and electrical quantity data;

[0027] A first updating module is used to re-collect target multimodal data according to the data collection strategy corresponding to the risk prediction result, and update the risk prediction model based on the actual fault, hidden danger elimination situation and the target multimodal data;

[0028] The second updating module is used to obtain the current multimodal data, perform risk prediction on the current multimodal data based on the updated risk prediction model, obtain the risk detection result of the power line, and update the data collection strategy according to the risk detection result. The updated data collection strategy is used for multimodal data collection.

[0029] In a third aspect, the present application further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0030] Determine the risk prediction result based on the multimodal data of the power line and the preset risk prediction model; the multimodal data includes image data, temperature data and electrical quantity data;

[0031] Re-collect the target multimodal data according to the data collection strategy corresponding to the risk prediction results, and update the risk prediction model based on the actual failure, hidden danger elimination situation and the target multimodal data;

[0032] The current multimodal data is obtained, and risk prediction is performed on the current multimodal data based on the updated risk prediction model to obtain risk detection results of the power lines, and the data acquisition strategy is updated according to the risk detection results. The updated data acquisition strategy is used for multimodal data acquisition.

[0033] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the following steps are implemented:

[0034] Determine the risk prediction result based on the multimodal data of the power line and the preset risk prediction model; the multimodal data includes image data, temperature data and electrical quantity data;

[0035] Re-collect the target multimodal data according to the data collection strategy corresponding to the risk prediction results, and update the risk prediction model based on the actual failure, hidden danger elimination situation and the target multimodal data;

[0036] The current multimodal data is obtained, and risk prediction is performed on the current multimodal data based on the updated risk prediction model to obtain risk detection results of the power lines, and the data acquisition strategy is updated according to the risk detection results. The updated data acquisition strategy is used for multimodal data acquisition.

[0037] In a fifth aspect, the present application further provides a computer program product, including a computer program, which implements the following steps when executed by a processor:

[0038] Determine the risk prediction result based on the multimodal data of the power line and the preset risk prediction model; the multimodal data includes image data, temperature data and electrical quantity data;

[0039] Re-collect the target multimodal data according to the data collection strategy corresponding to the risk prediction results, and update the risk prediction model based on the actual failure, hidden danger elimination situation and the target multimodal data;

[0040] The current multimodal data is obtained, and risk prediction is performed on the current multimodal data based on the updated risk prediction model to obtain risk detection results of the power lines, and the data acquisition strategy is updated according to the risk detection results. The updated data acquisition strategy is used for multimodal data acquisition.

[0041] The above-mentioned power data collection method, device, computer equipment, computer-readable storage medium and computer program product obtain risk prediction results through multimodal data of power lines and a preset risk prediction model, and then re-collect target multimodal data according to the data collection strategy corresponding to the risk prediction results. The target multimodal data is re-collected in combination with the risk prediction results of the power circuit. Therefore, it is beneficial to improve the utilization rate of collection resources; the risk prediction model can be continuously updated according to actual faults, hidden danger elimination conditions and target multimodal data, which is beneficial to improve the accuracy of risk detection of power lines by the risk prediction model; the data collection strategy is updated using the risk detection results detected by the updated risk prediction model, thereby realizing dynamic adjustment of the data collection strategy. The updated data collection strategy is used for multimodal data collection to form an adaptive collection closed loop of power data, thereby improving the utilization rate of collection resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings required for use in the embodiments of the present application or related technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0043] FIG1 is a diagram showing an application environment of a power data collection method according to an embodiment;

[0044] FIG2 is a schematic diagram of a flow chart of a method for collecting power data in one embodiment;

[0045] FIG3 is a schematic diagram of a flow chart of a method for collecting power data in another embodiment;

[0046] FIG4 is a schematic diagram of a flow chart of a power data collection method in yet another embodiment;

[0047] FIG5 is a structural block diagram of a power data acquisition device in one embodiment;

[0048] FIG. 6 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0049] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0050] The power data collection method provided in the embodiment of the present application can be applied to the application environment shown in Figure 1. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be integrated on the server 104, or it can be placed on the cloud or other network servers. This embodiment is illustrated by applying the method to the terminal 102. It can be understood that the method can also be applied to the server 104, and can also be applied to a system including the terminal 102 and the server 104, and is implemented through the interaction between the terminal 102 and the server 104. The terminal 102 determines the risk prediction result based on the multimodal data of the power line and the preset risk prediction model; the multimodal data includes image data, temperature data and electrical quantity data; according to the data collection strategy corresponding to the risk prediction result, the target multimodal data is re-collected, and the risk prediction model is updated based on the actual fault, hidden danger elimination situation and the target multimodal data; the current multimodal data is obtained, and the risk prediction of the current multimodal data is predicted based on the updated risk prediction model to obtain the risk detection result of the power line, and the data collection strategy is updated according to the risk detection result. The updated data collection strategy is used for multimodal data collection. The terminal 102 may be, but is not limited to, various personal computers, laptops, smart phones, tablet computers, IoT devices, and portable wearable devices. The IoT devices may be smart speakers, smart TVs, smart air conditioners, smart car devices, projection devices, etc. Portable wearable devices may be smart watches, smart bracelets, head-mounted devices, etc. Head-mounted devices may be virtual reality (VR) devices, augmented reality (AR) devices, smart glasses, etc. The server 104 may be an independent physical server, or a server cluster or distributed system consisting of multiple physical servers, or a cloud server that provides cloud computing services.

[0051] In an exemplary embodiment, as shown in FIG2 , a method for collecting power data is provided, which is described by taking the method applied to the terminal 102 in FIG1 as an example, and includes the following steps 202 to 206. Among them:

[0052] Step 202, determining a risk prediction result based on multimodal data of the power line and a preset risk prediction model; the multimodal data includes image data, temperature data, and electrical quantity data.

[0053] Among them, multimodal data refers to a variety of power-related data in the power line. For example, multimodal data includes image data, temperature data, and electrical quantity data. Among them, the image data can be an image obtained by patrolling and photographing the power line using a patrol drone or a monitoring camera. The types of image data can include visible light images, infrared images, ultraviolet images, etc. When the image acquisition device is a three-dimensional camera, three-dimensional point cloud data can be collected. Temperature data can be the temperature collected by the environmental sensor deployed in the power line. Electrical quantity data includes current data, voltage data, etc., and the electrical quantity data can be data collected by the electrical quantity sensor deployed in the power line.

[0054] The risk prediction model refers to the machine learning model preset in the terminal, which is used to predict the risks of power lines and obtain risk prediction results.

[0055] The risk prediction results are used to indicate the risk type of power lines. Since the risk prediction results are obtained based on multimodal data and risk prediction models, the accuracy of the risk prediction results depends on the real-time and accuracy of the multimodal data and the prediction ability of the risk prediction model.

[0056] Step 204, re-collect the target multimodal data according to the data collection strategy corresponding to the risk prediction result, and update the risk prediction model based on the actual fault, hidden danger elimination situation and the target multimodal data.

[0057] The data collection strategy refers to the strategy of each collection device for multimodal data collection. In some embodiments, a mapping relationship between risk type and data collection strategy is preset in the terminal, and the data collection strategy corresponding to the risk prediction result can be obtained by searching the mapping relationship.

[0058] The target multimodal data is the multimodal data re-collected according to the data collection strategy. Since the power line is long, the risk prediction result of the power line can be obtained through the risk prediction of the risk prediction model. For different risk prediction results, different data collection strategies are adopted, which is conducive to focusing on the multimodal data of the risk area and reducing the monitoring of the multimodal data of the non-risk area, thereby improving the utilization rate of the collection resources.

[0059] Actual faults refer to real faults of power lines confirmed by manual inspections. Hidden danger elimination refers to the elimination of faults and hidden dangers that have been found.

[0060] With the collection of multimodal data and the prediction of risks, the number of samples used for fault detection increases. In order to improve the prediction accuracy of the risk prediction model, the embodiment of the present application proposes to update the risk prediction model based on actual faults, hidden danger elimination conditions and target multimodal data to ensure that the risk prediction model can be dynamically updated according to the multimodal data collected in real time and the actual fault conditions. In the field of power line fault detection, due to the limitation of the amount of fault sample data, this method of dynamically updating the risk prediction model is conducive to improving the prediction ability of the risk prediction model.

[0061] Step 206, obtain the current multimodal data, perform risk prediction on the current multimodal data based on the updated risk prediction model, obtain the risk detection result of the power line, and update the data collection strategy according to the risk detection result. The updated data collection strategy is used for multimodal data collection.

[0062] Among them, after the risk prediction model is updated, the terminal re-acquires the current multimodal data collected at the current moment, and performs risk prediction on the current multimodal data to obtain the risk detection result of the power line.

[0063] Since both the multimodal data and the risk prediction model have been updated, the risk detection results have been updated. The terminal can obtain the data collection strategy corresponding to the risk detection results, realize the dynamic update of the data collection strategy, and collect multimodal data according to the updated data collection strategy. This can guide the collection equipment to pay attention to risk points, thereby ensuring the efficient use of collection resources.

[0064] In the above-mentioned power data collection method, the risk prediction result is obtained through the multimodal data of the power line and the preset risk prediction model, and then the target multimodal data is re-collected according to the data collection strategy corresponding to the risk prediction result. The target multimodal data is re-collected in combination with the risk prediction result of the power circuit. Therefore, it is beneficial to improve the utilization rate of the collection resources; the risk prediction model can be continuously updated according to the actual fault, hidden danger elimination situation and the target multimodal data, which is beneficial to improve the risk detection accuracy of the risk prediction model for the power line; the data collection strategy is updated by the risk detection result detected by the updated risk prediction model, and the dynamic adjustment of the data collection strategy is realized. The updated data collection strategy is used for multimodal data collection to form an adaptive collection closed loop of power data, thereby improving the utilization rate of collection resources.

[0065] In an exemplary embodiment, a risk prediction result is determined based on multimodal data of a power line and a preset risk prediction model, including: using a preset identification and analysis system to identify hidden dangers in the multimodal data of the power line to obtain a hidden danger identification result corresponding to at least one hidden danger of the power line; inputting the hidden danger identification result corresponding to at least one hidden danger into the preset risk prediction model to obtain a risk prediction result.

[0066] Among them, the recognition and analysis system preset in the terminal can be a large multimodal model or multiple small models. The large multimodal model is used to uniformly process the multimodal data, and the multiple small models are used to process each modal data separately to identify possible hidden dangers in the power line and predict the hidden danger score of the power line.

[0067] For example, the terminal inputs the multimodal data into the identification and analysis system to obtain a hidden danger identification result corresponding to at least one hidden danger, and the hidden danger identification result corresponding to each hidden danger may include a hidden danger score and a confidence level. The confidence level is a measure of the credibility of the hidden danger.

[0068] The terminal inputs the hidden danger identification result corresponding to at least one hidden danger into a preset risk prediction model to obtain a risk prediction result output by the risk prediction model. Exemplarily, the risk prediction result can be a risk prediction matrix, and each element in the matrix is ​​the probability or intensity of the corresponding hidden danger.

[0069] In some embodiments, the terminal further abstracts the risk prediction matrix into a collection strategy matrix, and then automatically generates a data collection strategy based on the collection strategy matrix. The collection strategy matrix and the risk prediction matrix have a preset corresponding relationship, and the data collection strategy and the collection strategy matrix have a preset corresponding relationship.

[0070] In this embodiment, hidden dangers are identified in multimodal data through a preset identification and analysis system, and then a preset risk prediction model is used to process the hidden danger identification results corresponding to at least one identified hidden danger to obtain a risk prediction result. This method of processing multimodal data through a preset system and model is conducive to obtaining accurate risk prediction results.

[0071] In an exemplary embodiment, the target multimodal data is re-collected according to the data collection strategy corresponding to the risk prediction result, including: adjusting the collection period of the collection device according to the data collection strategy corresponding to the risk prediction result; and controlling the collection device to re-collect according to the adjusted collection period to obtain the target multimodal data.

[0072] Among them, the collection cycle refers to the interval time each time the collection device collects data. Different data collection strategies correspond to different collection cycles. For example, if the data collection strategy is to increase the collection cycle, the collection cycle of the inspection drones, monitoring cameras, electrical quantity sensors, environmental sensors and other collection devices in the power line will be controlled and adjusted to increase the preset time. This method of adjusting the collection cycle according to the data collection strategy realizes the monitoring strategy of the power line based on the dynamic control of risk prediction, and controls the collection equipment to re-collect the target multimodal data according to the adjusted collection cycle, which is conducive to substantial energy saving when the impact of defects and hidden dangers is small.

[0073] In this embodiment, the data collection strategy obtained by dynamic adjustment corresponds to the adjustment of the collection period of the collection device, thereby realizing dynamic adjustment of the monitoring strategy of the power line, which is beneficial to substantial energy saving and improving the utilization rate of collection resources.

[0074] In an exemplary embodiment, a risk prediction model is updated based on actual failures, hidden danger elimination conditions and target multimodal data, including: updating risk prediction results based on target multimodal data and the risk prediction model; checking the correctness of the updated risk prediction results according to actual failures and hidden danger elimination conditions; inputting the correctness check results, actual failures and hidden danger elimination conditions into a preset reinforcement learning model to obtain parameter changes, and updating the risk prediction model based on the parameter changes.

[0075] Among them, since the collection cycle of the collection equipment has been adjusted, the collected target multimodal data has changed. Therefore, the risk prediction model performs risk prediction on the target multimodal data and can update the risk prediction result.

[0076] The updated risk prediction results may differ from the actual fault conditions, so the updated risk prediction results can be checked for correctness based on the actual faults and hidden danger elimination conditions. Since the actual faults and hidden danger elimination conditions are obtained through manual inspection, for example, if the actual fault, risk prediction results, and updated risk prediction results do not identify the fault, the correctness check result is determined to be incorrect, and it is necessary to correct the risk prediction model to improve the accuracy of the risk prediction results.

[0077] In some embodiments, the preset reinforcement learning model can be a preset corresponding relationship, and different input data corresponds to different parameter changes. The terminal inputs the correctness check results, actual faults and hidden danger elimination conditions into the preset reinforcement learning model to obtain the parameter change. The parameter change is the change in each model parameter in the risk prediction model. The terminal adjusts the model parameters of the risk prediction model according to the parameter change to obtain an updated risk prediction model.

[0078] In this embodiment, the risk prediction model is used to perform risk prediction on the re-collected target multimodal data to update the risk prediction results, and the correctness of the updated risk prediction results is checked according to the actual fault situation. Since the updated risk prediction results are different from the actual fault situation, the reinforcement learning model can be used to determine the parameter change amount, so as to update the risk prediction model according to the parameter change amount, thereby ensuring that the risk prediction model is more in line with the actual fault situation, which is conducive to improving the prediction ability of the risk prediction model.

[0079] In an exemplary embodiment, the correctness check results, actual faults and hidden danger elimination situations are input into a preset reinforcement learning model to obtain parameter changes, including: using the reward and punishment function in the preset reinforcement learning model to preprocess the correctness check results, actual faults and hidden danger elimination situations to obtain preprocessing results; and converting the preprocessing results into parameter changes based on the value function in the preset reinforcement learning model.

[0080] Among them, the reinforcement learning model includes a reward and punishment function and a value function. The reward and punishment function is used to preprocess the correctness check results, actual faults and hidden danger elimination conditions input into the reinforcement learning model to obtain preprocessing results. Preprocessing mainly involves thresholding each data to eliminate abnormal data and obtain the result after the abnormal data is eliminated as the preprocessing result.

[0081] The value function is a preset functional relationship, through which the preprocessing results can be converted into parameter changes.

[0082] In this embodiment, the input data are processed by the reward and punishment function and the value function in the preset reinforcement learning model to obtain the parameter change amount, which is beneficial to determine the parameter change amount of the risk prediction model based on the risk prediction results of the risk prediction model and the actual fault situation, and is used to adjust the model parameters of the risk prediction model to improve the risk prediction ability of the risk prediction model.

[0083] In an exemplary embodiment, the power data collection method further includes: collecting multimodal data to be processed according to an updated data collection strategy; and updating the updated risk prediction model based on the multimodal data to be processed.

[0084] Among them, after the risk prediction model is updated, the data collection strategy is updated accordingly, and the terminal controls the collection device to collect data according to the updated data collection strategy to obtain multimodal data to be processed.

[0085] As the multimodal data has changed, the model samples of the risk prediction model have increased. In the field of risk detection of power lines, the re-collected multimodal data to be processed can be used as model samples to further update the updated risk prediction model.

[0086] In this embodiment, after the risk prediction model is updated, the updated data collection strategy is used to re-collect the unprocessed multimodal data, and the unprocessed multimodal data is used as a model sample to continue to update the updated risk prediction model, thereby realizing the dynamic update of the risk prediction model, ensuring that the risk prediction model learns more power line risk fault conditions, and is conducive to improving the risk prediction capability of the risk prediction model.

[0087] To explain the power data collection method and effect in this solution in detail, a most detailed embodiment is described below:

[0088] The power line inspection scenario is used as an example for explanation. As shown in FIG3, a flow chart of the power data collection method in another embodiment is shown. First, the multimodal data of the power line is collected, and the hidden dangers are identified on the multimodal data to obtain the hidden danger identification results to predict the current risk, and the data collection strategy is dynamically updated to form an adaptive collection closed loop of environmental data to achieve efficient use of collection resources. At the same time, a feedback reward and punishment mechanism is established based on actual faults, hidden danger elimination conditions and risk prediction results, and the risk prediction model is updated using an online reinforcement learning method to form an adaptive update closed loop of the risk prediction model, guiding the collection system to pay attention to high-risk and scarce defect hidden dangers. Points are prone to occur.

[0089] Figure 4 is a schematic flow chart of a power data collection method in another embodiment. The power data collection method includes two major processes: an environmental data adaptive collection closed loop and a risk model adaptive update closed loop.

[0090] 1. The closed-loop process of adaptive environmental data collection is as follows:

[0091] 1.1. Obtain a series of multimodal data such as visible light images, infrared images, 3D point clouds, current values, voltage values, and temperature of power lines through inspection drones, surveillance cameras, electrical quantity sensors, environmental sensors and other equipment.

[0092] 1.2. The multimodal data is transmitted to the identification and analysis system. The identification and analysis system processes the multimodal data uniformly or separately through a large multimodal model or multiple small models to identify the hidden danger identification results corresponding to at least one hidden danger of the power line, including possible abnormal conditions in the power line and the risk score of the power line.

[0093] 1.3. Input the hidden danger identification results into the risk prediction model to obtain the risk prediction matrix, that is, the risk prediction results, and further abstract it into the collection strategy matrix.

[0094] 1.4. Automatically generate data collection strategies for power lines based on the collection strategy matrix and upload them to the power grid system. Control and adjust the collection cycles of inspection drones, surveillance cameras, electrical quantity sensors, environmental sensors and other equipment in the power lines to dynamically control power line monitoring strategies based on risk prediction, thereby achieving significant energy savings with little impact on the discovery of defects and hidden dangers.

[0095] 2. The closed-loop process of adaptive updating of risk models is as follows:

[0096] 2.1. According to the data collection strategy corresponding to the risk prediction results, a series of target multimodal data such as visible light images, infrared images, three-dimensional point clouds, current values, voltage values, and temperatures of power lines are obtained through inspection drones, monitoring cameras, electrical quantity sensors, environmental sensors and other equipment.

[0097] 2.2. The target multimodal data is transmitted to the identification and analysis system. The identification and analysis system processes each modal data uniformly or separately through a large multimodal model or multiple small models, identifies the hidden danger identification results corresponding to at least one hidden danger of the power line, and updates the risk prediction results based on the hidden danger identification results and the risk prediction model.

[0098] 2.3. Determine the accuracy of risk prediction results and updated risk prediction results based on actual failures and hidden danger elimination situations.

[0099] 2.4. The online reinforcement learning model includes reward and punishment functions and value functions. The correctness check results, actual faults and hidden danger elimination situations are input into the online reinforcement learning model as high-risk feedback rewards and punishments, and the value function is updated, and then the risk prediction model is updated to realize the online dynamic update of the risk prediction model.

[0100] 2.5. Use the updated risk prediction model to evaluate system risk for the current multimodal data collected subsequently, and adaptively update the data collection strategy. The updated data collection strategy is used for multimodal data collection.

[0101] The above-mentioned power data collection method obtains the risk prediction result through the multimodal data of the power line and the preset risk prediction model, and then re-collects the target multimodal data according to the data collection strategy corresponding to the risk prediction result. The target multimodal data is re-collected in combination with the risk prediction result of the power circuit, so it is conducive to improving the utilization rate of the collection resources; the risk prediction model can be continuously updated according to the actual fault, hidden danger elimination situation and the target multimodal data, which is conducive to improving the risk detection accuracy of the risk prediction model for the power line; the risk detection result detected by the updated risk prediction model is used to update the data collection strategy, and the dynamic adjustment of the data collection strategy is realized. The updated data collection strategy is used for multimodal data collection to form an adaptive collection closed loop of power data, which improves the utilization rate of collection resources. The established risk-driven dynamic adaptive collection architecture achieves a balance between low-power active perception and high-reliability risk prevention. Since there are not many actual fault data of power lines, and the association with risk data before the fault is even more scarce, this online reinforcement learning method can realize the dynamic update of the risk prediction model, and realize the dual closed loop of adaptive collection of environmental data and adaptive update of risk model.

[0102] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0103] Based on the same inventive concept, the embodiment of the present application also provides a power data acquisition device for implementing the power data acquisition method involved above. The implementation solution provided by the device to solve the problem is similar to the implementation solution recorded in the above method, so the specific limitations in one or more power data acquisition device embodiments provided below can refer to the limitations of the power data acquisition method above, and will not be repeated here.

[0104] In an exemplary embodiment, as shown in FIG5 , a power data acquisition device 100 is provided, including: an acquisition module 120, a first update module 140 and a second update module 160, wherein:

[0105] The acquisition module 120 is used to determine the risk prediction result based on the multimodal data of the power line and the preset risk prediction model; the multimodal data includes image data, temperature data and electrical quantity data;

[0106] The first updating module 140 is used to re-collect target multimodal data according to the data collection strategy corresponding to the risk prediction result, and update the risk prediction model based on the actual fault, hidden danger elimination situation and the target multimodal data;

[0107] The second update module 160 is used to obtain the current multimodal data, perform risk prediction on the current multimodal data based on the updated risk prediction model, obtain the risk detection result of the power line, and update the data collection strategy according to the risk detection result. The updated data collection strategy is used for multimodal data collection.

[0108] The above-mentioned power data acquisition device obtains the risk prediction result through the multimodal data of the power line and the preset risk prediction model, and then re-collects the target multimodal data according to the data acquisition strategy corresponding to the risk prediction result. The target multimodal data is re-collected in combination with the risk prediction result of the power circuit. Therefore, it is beneficial to improve the utilization rate of the acquisition resources; the risk prediction model can be continuously updated according to the actual fault, hidden danger elimination situation and the target multimodal data, which is beneficial to improve the risk detection accuracy of the risk prediction model for the power line; the risk detection result detected by the updated risk prediction model is used to update the data acquisition strategy, so as to realize the dynamic adjustment of the data acquisition strategy. The updated data acquisition strategy is used for multimodal data acquisition to form an adaptive acquisition closed loop of power data, thereby improving the utilization rate of acquisition resources.

[0109] In one embodiment, a risk prediction result is determined based on the multimodal data of the power line and a preset risk prediction model, and the acquisition module 120 is also used to: use a preset identification and analysis system to identify hidden dangers on the multimodal data of the power line to obtain a hidden danger identification result corresponding to at least one hidden danger of the power line; input the hidden danger identification result corresponding to at least one hidden danger into the preset risk prediction model to obtain a risk prediction result.

[0110] In one embodiment, the target multimodal data is re-collected according to the data collection strategy corresponding to the risk prediction result, and the first update module 140 is also used to: adjust the collection period of the collection device according to the data collection strategy corresponding to the risk prediction result; control the collection device to re-collect according to the adjusted collection period to obtain the target multimodal data.

[0111] In one embodiment, the risk prediction model is updated based on actual failures, hidden danger elimination conditions and target multimodal data, and the first update module 140 is also used to: update the risk prediction results based on the target multimodal data and the risk prediction model; check the correctness of the updated risk prediction results according to the actual failures and hidden danger elimination conditions; input the correctness check results, actual failures and hidden danger elimination conditions into a preset reinforcement learning model to obtain parameter changes, and update the risk prediction model based on the parameter changes.

[0112] In one embodiment, the correctness check results, actual faults and hidden danger elimination situations are input into a preset reinforcement learning model to obtain parameter changes. The first update module 140 is also used to: use the reward and punishment function in the preset reinforcement learning model to preprocess the correctness check results, actual faults and hidden danger elimination situations to obtain preprocessing results; and convert the preprocessing results into parameter changes based on the value function in the preset reinforcement learning model.

[0113] In one embodiment, the electric power data acquisition device further includes a third update module, which is used to: acquire the multimodal data to be processed according to the updated data acquisition strategy; and update the updated risk prediction model based on the multimodal data to be processed.

[0114] Each module in the above-mentioned power data acquisition device can be implemented in whole or in part by software, hardware and their combination. Each of the above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the corresponding operations of each of the above modules.

[0115] In an exemplary embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be shown in FIG6. The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit, and an input device. The processor, the memory, and the input / output interface are connected via a system bus, and the communication interface, the display unit, and the input device are connected to the system bus via the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange data between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner may be implemented through WIFI, a mobile cellular network, near field communication (NFC) or other technologies. When the computer program is executed by the processor, a method for collecting power data is implemented. The display unit of the computer device is used to form a visually visible picture, which may be a display screen, a projection device, or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device shell, or an external keyboard, touchpad or mouse.

[0116] Those skilled in the art will understand that the structure shown in FIG. 6 is merely a block diagram of a partial structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different arrangement of components.

[0117] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:

[0118] Based on the multimodal data of the power line and the preset risk prediction model, the risk prediction result is determined; the multimodal data includes image data, temperature data and electrical quantity data; the target multimodal data is re-collected according to the data collection strategy corresponding to the risk prediction result, and the risk prediction model is updated based on the actual fault, hidden danger elimination situation and the target multimodal data; the current multimodal data is obtained, and the risk prediction is performed on the current multimodal data based on the updated risk prediction model to obtain the risk detection result of the power line, and the data collection strategy is updated according to the risk detection result. The updated data collection strategy is used for multimodal data collection.

[0119] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0120] A preset identification and analysis system is used to identify hidden dangers in the multimodal data of the power line, and a hidden danger identification result corresponding to at least one hidden danger of the power line is obtained; the hidden danger identification result corresponding to at least one hidden danger is input into a preset risk prediction model to obtain a risk prediction result.

[0121] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0122] The collection cycle of the collection device is adjusted according to the data collection strategy corresponding to the risk prediction result; the collection device is controlled to re-collect the target multimodal data according to the adjusted collection cycle.

[0123] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0124] Based on the target multimodal data and risk prediction model, the risk prediction results are updated; the correctness of the updated risk prediction results is checked according to the actual failure and hidden danger elimination situation; the correctness check results, actual failure and hidden danger elimination situation are input into the preset reinforcement learning model to obtain the parameter change, and the risk prediction model is updated based on the parameter change.

[0125] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0126] The correctness check results, actual faults and hidden danger elimination situations are preprocessed using the reward and punishment function in the preset reinforcement learning model to obtain preprocessing results; the preprocessing results are converted into parameter changes based on the value function in the preset reinforcement learning model.

[0127] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0128] The multimodal data to be processed is collected according to the updated data collection strategy; and the updated risk prediction model is updated based on the multimodal data to be processed.

[0129] In one embodiment, a computer readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

[0130] Based on the multimodal data of the power line and the preset risk prediction model, the risk prediction result is determined; the multimodal data includes image data, temperature data and electrical quantity data; the target multimodal data is re-collected according to the data collection strategy corresponding to the risk prediction result, and the risk prediction model is updated based on the actual fault, hidden danger elimination situation and the target multimodal data; the current multimodal data is obtained, and the risk prediction is performed on the current multimodal data based on the updated risk prediction model to obtain the risk detection result of the power line, and the data collection strategy is updated according to the risk detection result. The updated data collection strategy is used for multimodal data collection.

[0131] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0132] A preset identification and analysis system is used to identify hidden dangers in the multimodal data of the power line, and a hidden danger identification result corresponding to at least one hidden danger of the power line is obtained; the hidden danger identification result corresponding to at least one hidden danger is input into a preset risk prediction model to obtain a risk prediction result.

[0133] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0134] The collection cycle of the collection device is adjusted according to the data collection strategy corresponding to the risk prediction result; the collection device is controlled to re-collect the target multimodal data according to the adjusted collection cycle.

[0135] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0136] Based on the target multimodal data and risk prediction model, the risk prediction results are updated; the correctness of the updated risk prediction results is checked according to the actual failure and hidden danger elimination situation; the correctness check results, actual failure and hidden danger elimination situation are input into the preset reinforcement learning model to obtain the parameter change, and the risk prediction model is updated based on the parameter change.

[0137] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0138] The correctness check results, actual faults and hidden danger elimination situations are preprocessed using the reward and punishment function in the preset reinforcement learning model to obtain preprocessing results; the preprocessing results are converted into parameter changes based on the value function in the preset reinforcement learning model.

[0139] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0140] The multimodal data to be processed is collected according to the updated data collection strategy; and the updated risk prediction model is updated based on the multimodal data to be processed.

[0141] In one embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the following steps:

[0142] Based on the multimodal data of the power line and the preset risk prediction model, the risk prediction result is determined; the multimodal data includes image data, temperature data and electrical quantity data; the target multimodal data is re-collected according to the data collection strategy corresponding to the risk prediction result, and the risk prediction model is updated based on the actual fault, hidden danger elimination situation and the target multimodal data; the current multimodal data is obtained, and the risk prediction is performed on the current multimodal data based on the updated risk prediction model to obtain the risk detection result of the power line, and the data collection strategy is updated according to the risk detection result. The updated data collection strategy is used for multimodal data collection.

[0143] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0144] A preset identification and analysis system is used to identify hidden dangers in the multimodal data of the power line, and a hidden danger identification result corresponding to at least one hidden danger of the power line is obtained; the hidden danger identification result corresponding to at least one hidden danger is input into a preset risk prediction model to obtain a risk prediction result.

[0145] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0146] The collection cycle of the collection device is adjusted according to the data collection strategy corresponding to the risk prediction result; the collection device is controlled to re-collect the target multimodal data according to the adjusted collection cycle.

[0147] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0148] Based on the target multimodal data and risk prediction model, the risk prediction results are updated; the correctness of the updated risk prediction results is checked according to the actual failure and hidden danger elimination situation; the correctness check results, actual failure and hidden danger elimination situation are input into the preset reinforcement learning model to obtain the parameter change, and the risk prediction model is updated based on the parameter change.

[0149] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0150] The correctness check results, actual faults and hidden danger elimination situations are preprocessed using the reward and punishment function in the preset reinforcement learning model to obtain preprocessing results; the preprocessing results are converted into parameter changes based on the value function in the preset reinforcement learning model.

[0151] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0152] The multimodal data to be processed is collected according to the updated data collection strategy; and the updated risk prediction model is updated based on the multimodal data to be processed.

[0153] It should be noted that the user data (including but not limited to user device data, user personal data, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all data and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0154] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, an artificial intelligence (AI) processor, etc., but are not limited to this.

[0155] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0156] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A method for collecting power data, characterized in that: The method comprises: Determine the risk prediction result based on the multimodal data of the power line and a preset risk prediction model; the multimodal data includes image data, temperature data and electrical quantity data; Re-collect target multimodal data according to the data collection strategy corresponding to the risk prediction result, and update the risk prediction model based on the actual fault, hidden danger elimination situation and the target multimodal data; Acquire current multimodal data, perform risk prediction on the current multimodal data based on an updated risk prediction model to obtain a risk prediction result of the power line, and update the data acquisition strategy according to the risk prediction result, wherein the updated data acquisition strategy is used for multimodal data acquisition; The updating of the risk prediction model based on actual faults, hidden danger elimination conditions and the target multimodal data includes: Based on the target multimodal data and the risk prediction model, updating the risk prediction result; Check the correctness of the updated risk prediction results based on the actual failures and hidden danger elimination situations; Input the correctness check results, actual faults and hidden danger elimination situations into a preset reinforcement learning model to obtain parameter changes, and update the risk prediction model based on the parameter changes; The correctness check results, actual faults and hidden danger elimination situations are input into the preset reinforcement learning model to obtain parameter changes, including: The reward and punishment function in the preset reinforcement learning model is used to pre-process the correctness check results, actual faults and hidden danger elimination situations to obtain the pre-processing results; The preprocessing result is converted into a parameter change based on a value function in a preset reinforcement learning model.

2. The method according to claim 1, characterized in that The method of determining the risk prediction result based on the multimodal data of the power line and the preset risk prediction model includes: Using a preset identification and analysis system to identify hidden dangers on the multimodal data of the power line, and obtaining hidden danger identification results corresponding to at least one hidden danger of the power line; The hidden danger identification result corresponding to at least one hidden danger is input into a preset risk prediction model to obtain a risk prediction result.

3. The method according to claim 1, characterized in that The re-collecting target multimodal data according to the data collection strategy corresponding to the risk prediction result includes: Adjusting the collection cycle of the collection device according to the data collection strategy corresponding to the risk prediction result; The acquisition device is controlled to re-acquire the target multimodal data according to the adjusted acquisition cycle.

4. The method according to claim 1, characterized in that: The method further comprises: Collecting the multimodal data to be processed according to the updated data collection strategy; The updated risk prediction model is updated based on the to-be-processed multimodal data and the corresponding actual fault and hidden danger elimination conditions.

5. A power data acquisition device, characterized in that: The device comprises: An acquisition module, used to determine a risk prediction result based on multimodal data of the power line and a preset risk prediction model; the multimodal data includes image data, temperature data and electrical quantity data; A first updating module is used to re-collect target multimodal data according to the data collection strategy corresponding to the risk prediction result, and update the risk prediction model based on the actual fault, hidden danger elimination situation and the target multimodal data; A second updating module is used to obtain current multimodal data, perform risk prediction on the current multimodal data based on an updated risk prediction model to obtain a risk prediction result of the power line, and update the data acquisition strategy according to the risk prediction result, wherein the updated data acquisition strategy is used for multimodal data acquisition; The first updating module is further used to update the risk prediction result based on the target multimodal data and the risk prediction model; perform a correctness check on the updated risk prediction result according to the actual failure and hidden danger elimination situation; input the correctness check result, the actual failure and hidden danger elimination situation into a preset reinforcement learning model to obtain a parameter change, and update the risk prediction model based on the parameter change; The first update module is also used to pre-process the correctness check results, actual faults and hidden danger elimination situations using the reward and punishment function in the preset reinforcement learning model to obtain pre-processing results; and convert the pre-processing results into parameter changes based on the value function in the preset reinforcement learning model.

6. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 4 are implemented.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 4 are implemented.

8. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 4 are implemented.