A method and device for identifying hidden dangers in narrow working space of distribution network

Abstract

The present invention provides a method and device for identifying hidden dangers in a narrow working space of a distribution network, wherein the method comprises: obtaining patrol data of the distribution network working space, pre-processing and saving to a patrol database, the patrol data including power equipment and narrow working spaces that may exist in the distribution network working space; judging whether there are hidden dangers of narrow working spaces in the working space of the patrol database according to a hidden danger feature expert database, and when there are hidden dangers of narrow working spaces in the distribution network working space, grading the severity of the hidden dangers, and saving the grading results in the hidden danger database; obtaining a digital map of hidden dangers of narrow working spaces in the distribution network according to the patrol database, the hidden danger database and the basic information of the distribution network, and visually controlling the hidden dangers of narrow working spaces in the distribution network according to the digital map. The present invention can automatically and efficiently identify hidden dangers of narrow working spaces in the distribution network, and facilitates visual control of hidden dangers of narrow working spaces in the distribution network.

Description

Technical Field

[0001] The present invention relates to the technical field of power distribution network, and in particular to a method and device for identifying hidden dangers in a narrow working space of a power distribution network. Background Art

[0002] The personal safety hazards in the distribution network equipment environment are prominent. In history, there have been many personal accidents closely related to the personal safety hazards in the equipment environment, posing serious threats to the life and health of on-site workers.

[0003] At present, equipment environment-related personal safety hazards are mainly checked and controlled by manual and semi-manual methods, which consumes a lot of manpower, material and financial resources. In addition, due to the complexity of distribution network equipment, structure and environment, and the overall low quality of personnel, the efficiency of hazard discovery is low, some hidden hazards are difficult to find, and the results of hazard investigation cannot effectively support on-site operation risk control. The problems of being unable to patrol, identify and control are prominent. The distribution network lines are located in a complex environment, including mountainous areas, hills and other places that are difficult for people and vehicles to reach. The safety hazard of narrow working space requires manual climbing to be discovered. Therefore, the current manual method often cannot discover the existence of this hazard. The semi-manual method relies on manual identification of pictures, videos and other materials to discover whether the hazard exists. The workload is large, and the efficiency and accuracy of hazard discovery are not ideal.

[0004] In addition to being unable to patrol and identify, the current hidden danger investigation and control results are mainly managed in the form of tables, which makes it difficult to clearly review the historical investigation and control of hidden dangers of the same equipment; on the other hand, the hidden danger records lack detailed geographical and image information, and operators cannot obtain intuitive safety risk information from the hidden danger record table, resulting in insufficient support for the linkage between the hidden danger investigation and control situation and the on-site operation risk management.

[0005] The shortcomings of the existing technology are as follows: relying on manual and semi-manual methods for inspection requires a lot of manpower, material and financial resources; the efficiency of discovering hidden dangers in narrow working spaces that require manual climbing is low; the management and control methods are backward and cannot clearly display the hidden danger inspection and control situation. Summary of the invention

[0006] The purpose of the present invention is to provide a method and device for identifying hidden dangers in narrow working spaces of distribution networks, so as to solve the technical problem in the prior art that manual participation is required to check hidden dangers in narrow working spaces of distribution networks, resulting in low inspection efficiency.

[0007] The purpose of the present invention can be achieved through the following technical solutions:

[0008] A method for identifying hidden dangers in a narrow working space of a distribution network, comprising:

[0009] Obtaining patrol data of the distribution network operation space, pre-processing the patrol data and saving it to a patrol database, wherein the patrol data includes power equipment and narrow operation spaces existing in the distribution network operation space;

[0010] Judging whether there is a hidden danger of narrow working space in the working space of the inspection database according to the feature expert database of hidden dangers of narrow working space in the distribution network, wherein the feature expert database is obtained according to the historical data of the working space of the distribution network and the results of manual review, and the hidden danger of narrow working space refers to the hidden danger existing when the position relationship between the power equipment in the working space of the distribution network constitutes a narrow working space;

[0011] When there is a hidden danger of narrow working space in the distribution network working space, the severity of the hidden danger is graded and the classification results are saved in the hidden danger database;

[0012] A digital map of hidden dangers in narrow working spaces of the distribution network is obtained based on the inspection database, the hidden danger database and basic information of the distribution network, and visual management and control of hidden dangers in narrow working spaces of the distribution network is performed based on the digital map.

[0013] Optionally, judging whether there is a hidden danger of narrow working space in the working space of the patrol database according to the feature expert database includes:

[0014] The patrol data in the patrol database are divided into a training set and a test set, and the matching degree between the power equipment and the narrow working space in the test set and the hidden danger feature expert database is determined respectively, and a first matching score and a second matching score are obtained, wherein the first matching score indicates the matching degree between the power equipment in the test set and the hidden danger feature expert database, and the second matching score indicates the matching degree between the narrow working space in the test set and the hidden danger feature expert database;

[0015] The first matching score is compared with the first confidence threshold, and the second matching score is compared with the second confidence threshold, and whether there are hidden dangers in the inspection database is determined according to the comparison results; the first confidence threshold and the second confidence threshold are determined according to the hidden danger identification model trained by the training set, and are respectively used to determine whether there are hidden dangers in power equipment and narrow working spaces.

[0016] Optionally, training the hidden danger identification model according to the training set includes:

[0017] Respectively determining the degree of matching between the power equipment and the narrow working space in the training set and the hidden danger feature expert database and obtaining a third matching score and a fourth matching score, wherein the third matching score indicates the degree of matching between the power equipment in the training set and the hidden danger feature expert database, and the fourth matching score indicates the degree of matching between the narrow working space in the training set and the hidden danger feature expert database;

[0018] A first matching threshold and a second matching threshold are set for the power equipment, and a third matching threshold and a fourth matching threshold are set for the narrow working space. The second matching threshold is greater than the first matching threshold, and the fourth matching threshold is greater than the third matching threshold. When formula 1 is established, it is judged that there is no hidden danger; when formula 2 or formula 3 is established, it is judged that there is a hidden danger; when formula 4 is established, further manual judgment is required;

[0019] Wherein, Formula 1 is: the third matching score < the first matching threshold or the fourth matching score < the third matching threshold; Formula 2 is: the third matching score ≥ the second matching threshold and the fourth matching score ≥ the third matching threshold;

[0020] Formula 3 is: the fourth matching score ≥ the fourth matching threshold and the third matching score ≥ the first matching threshold;

[0021] Formula 4 is: first matching threshold ≤ third matching score ≤ second matching threshold and third matching threshold ≤ fourth matching score ≤ fourth matching threshold;

[0022] The results of manual judgment are input into the hidden danger identification model as part of the training set for learning. The first matching threshold and the third matching threshold are continuously increased manually according to the performance of the hidden danger identification model, and the second matching threshold and the fourth matching threshold are continuously lowered. The process of inputting the results of manual judgment into the hidden danger identification model is repeated until the first matching threshold and the second matching threshold are equal to obtain the first confidence threshold, and the third matching threshold and the fourth matching threshold are equal to obtain the second confidence threshold.

[0023] Optionally, respectively determining the matching degree between the power equipment in the test set and the narrow working space and the hidden danger feature expert database and obtaining a first matching score and a second matching score comprises:

[0024] Two trained neural network models are used to respectively obtain the hidden danger features of the power equipment and the narrow working space in the inspection database, and the hidden danger features of the power equipment and the narrow working space are respectively compared with the hidden danger features in the hidden danger feature expert database, and the first matching score and the second matching score are respectively obtained according to the matching degree.

[0025] Optionally, respectively comparing the first matching score with a first confidence threshold, and the second matching score with a second confidence threshold, and judging whether there is a hidden danger in the patrol database according to the comparison results includes:

[0026] It is determined whether the first matching score is greater than or equal to a first confidence threshold and whether the second matching score is greater than or equal to a second confidence threshold. If so, it is determined that there is a hidden danger in the narrow operating space of the distribution network; otherwise, there is no hidden danger.

[0027] Optionally, the operation of grading the severity of hidden dangers includes:

[0028] The point with the first confidence threshold on the horizontal axis and the second confidence threshold on the vertical axis is taken as the origin, and the point with the highest score on the horizontal axis and the highest score on the vertical axis is taken as the end point. The rectangular area formed by the origin and the end point is evenly divided into 9 areas, and the 9 areas are numbered in sequence from top to bottom and from left to right.

[0029] The hidden danger corresponding to the third area is a very high-level hidden danger, the hidden danger corresponding to the second, fifth and sixth areas are high-level hidden dangers, and the hidden dangers corresponding to other areas are medium-level hidden dangers.

[0030] Optionally, obtaining a digital map of hidden dangers in narrow working spaces of a distribution network according to the patrol database, the hidden danger database and basic information of the distribution network includes:

[0031] A holographic model database of hidden dangers in narrow working spaces of distribution networks is obtained by modeling based on the patrol database and the hidden danger database; a digital map of distribution networks is obtained by modeling based on basic information of distribution networks; the holographic model database is embedded in the digital map of distribution networks to obtain a digital map of hidden dangers in narrow working spaces of distribution networks.

[0032] Optionally, the hidden danger feature expert database is obtained based on historical data of the distribution network operation space and manual review results and includes:

[0033] The characteristics of power equipment and narrow working space in the historical data of distribution network operating space are identified respectively, and the hidden danger feature library of power equipment and narrow working space is obtained respectively. Combined with the manual review results, the feature expert library of hidden dangers of distribution network narrow working space is obtained.

[0034] The present invention also provides a device for identifying hidden dangers in narrow working spaces of a distribution network, comprising:

[0035] A patrol database establishment module is used to obtain patrol data of the distribution network operation space, and save the patrol data to the patrol database after pre-processing, wherein the patrol data includes power equipment and narrow operation spaces existing in the distribution network operation space;

[0036] A hidden danger judgment module is used to judge whether there is a hidden danger of narrow working space in the working space of the patrol database according to the feature expert database of hidden dangers of narrow working space of the distribution network, wherein the feature expert database is obtained according to the historical data of the working space of the distribution network and the results of manual review, and the hidden danger of narrow working space refers to the hidden danger existing when the position relationship between the power equipment in the working space of the distribution network constitutes a narrow working space;

[0037] A hidden danger database establishment module is used to classify the severity of hidden dangers when there are hidden dangers of narrow working space in the distribution network working space, and save the classification results in the hidden danger database;

[0038] The hidden danger automatic identification module is used to obtain a digital map of hidden dangers in the narrow working space of the distribution network based on the patrol database, the hidden danger database and the basic information of the distribution network, and to perform visual management and control of the hidden dangers in the narrow working space of the distribution network based on the digital map.

[0039] Optionally, the hidden danger judgment module judges whether there is a hidden danger of narrow working space in the working space of the patrol database according to the feature expert database, including:

[0040] The inspection data in the inspection database are divided into a training set and a test set, and the hidden danger judgment module judges the matching degree between the power equipment and the narrow working space in the test set and the hidden danger feature expert database respectively, and obtains a first matching score and a second matching score, wherein the first matching score indicates the matching degree between the power equipment in the test set and the hidden danger feature expert database, and the second matching score indicates the matching degree between the narrow working space in the test set and the hidden danger feature expert database;

[0041] The first matching score is compared with the first confidence threshold, and the second matching score is compared with the second confidence threshold, and whether there are hidden dangers in the inspection database is determined according to the comparison results; the first confidence threshold and the second confidence threshold are determined according to the hidden danger identification model trained by the training set, and are respectively used to determine whether there are hidden dangers in power equipment and narrow working spaces.

[0042] The present invention provides a method and device for identifying hidden dangers in a narrow working space of a distribution network, wherein the method comprises: obtaining patrol data of patrolling the working space of the distribution network, pre-processing the patrol data and saving them in a patrol database, wherein the patrol data includes power equipment and narrow working spaces existing in the working space of the distribution network; judging whether there are hidden dangers of narrow working spaces in the working space of the patrol database according to a feature expert database of hidden dangers of narrow working spaces in the distribution network, wherein the feature expert database is obtained according to historical data of the working space of the distribution network and manual review results, and the hidden dangers of narrow working spaces refer to hidden dangers existing when the positional relationship between power equipment in the working space of the distribution network constitutes a narrow working space; when there are hidden dangers of narrow working spaces in the working space of the distribution network, grading the severity of the hidden dangers, and saving the grading results in a hidden danger database; obtaining a digital map of hidden dangers of narrow working spaces in the distribution network according to the patrol database, the hidden danger database and basic information of the distribution network, and visually controlling the hidden dangers of narrow working spaces in the distribution network according to the digital map.

[0043] The method and device for identifying hidden dangers in narrow working spaces of distribution networks provided by the present invention have the following beneficial effects:

[0044] The present invention aims at the hidden danger of narrow working space in distribution network which is difficult to find. The inspection data of the distribution network working space is obtained by automatic inspection. The feature expert database of the hidden danger of narrow working space in distribution network is obtained according to the historical data of the distribution network working space and the manual review result. The historical accumulated data can provide data support for the effect of the optimization method. According to the hidden danger feature expert database, it is judged whether there is a hidden danger of narrow working space in the working space of the inspection database. Compared with the pure manual scheme, it can effectively reduce the manual workload and ensure the accuracy of the identification effect. The hidden danger database is obtained by grading the severity of the hidden danger, and the digital map of the hidden danger of narrow working space is further obtained. The formed hidden danger database, hidden danger feature expert database and hidden danger digital map can all be updated in real time, and can automatically and efficiently identify the hidden danger of narrow working space in distribution network, which is convenient for visual control of the hidden danger of narrow working space in distribution network. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is a schematic diagram of the flow chart of the identification method of the present invention;

[0046] Figure 2 It is a structural schematic diagram of the identification device of the present invention;

[0047] Figure 3 It is a flow chart of an embodiment of the identification method of the present invention;

[0048] Figure 4 A schematic diagram of power identification and a narrow working space in an embodiment of the identification method of the present invention;

[0049] Figure 5 A schematic diagram of hidden danger identification model training in an embodiment of the identification method of the present invention;

[0050] Figure 6 A schematic diagram of hidden danger judgment in an embodiment of the identification method of the present invention;

[0051] Figure 7 Schematic diagram of hidden danger classification operation in an embodiment of the identification method of the present invention. DETAILED DESCRIPTION

[0052] The embodiment of the present invention provides a method and device for identifying hidden dangers in a narrow working space of a distribution network, so as to solve the technical problem in the prior art that manual participation is required to check hidden dangers in a narrow working space of a distribution network, resulting in low inspection efficiency.

[0053] In order to facilitate the understanding of the present invention, the present invention will be described more fully below with reference to the relevant drawings. The preferred embodiments of the present invention are shown in the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, the purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive.

[0054] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used herein in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more related listed items.

[0055] A method and device for identifying hidden dangers in a small working space of a distribution network provided in an embodiment of the present invention can realize automatic inspection and investigation of hidden dangers in a small working space of a distribution network, reduce the workload of inspection; automatically identify hidden dangers in a small working space, improve the efficiency of discovering hidden dangers; and perform visual management and control of hidden dangers in a small working space.

[0056] See also Figure 1 The present invention provides an embodiment of a method for identifying hidden dangers of narrow working space in a distribution network, comprising:

[0057] S100: Obtain patrol data of the distribution network operation space, pre-process the patrol data and save it to a patrol database, wherein the patrol data includes power equipment and narrow operation spaces in the distribution network operation space;

[0058] S200: judging whether there is a hidden danger of narrow working space in the working space of the inspection database according to the feature expert database of hidden dangers of narrow working space in the distribution network, wherein the feature expert database is obtained according to the historical data of the working space of the distribution network and the result of manual review, and the hidden danger of narrow working space refers to the hidden danger existing when the position relationship between the power equipment in the working space of the distribution network constitutes a narrow working space;

[0059] S300: When there is a hidden danger of narrow working space in the distribution network working space, the severity of the hidden danger is graded and the classification result is saved in the hidden danger database;

[0060] S400: A digital map of hidden dangers in narrow working spaces of the distribution network is obtained according to the inspection database, the hidden danger database and basic information of the distribution network, and visual management and control of hidden dangers in narrow working spaces of the distribution network is performed according to the digital map.

[0061] In step S100, the distribution network operation space is patrolled or video monitored to obtain patrol data, and the obtained patrol data is pre-processed and saved in a patrol database, that is, the patrol database contains patrol images of the distribution network operation space, on-site video monitoring images and other information. The specific process is:

[0062] First, based on the patrol device such as a drone dynamically patrolling the distribution network operation space, a preset patrol program is installed in the patrol device to control the patrol path and obstacle avoidance strategy of the patrol device, so as to ensure that the patrol device can automatically patrol and obtain patrol data of the distribution network operation space. After obtaining the patrol data, if conditions are met, the patrol device is connected to the server where the patrol database is located, and the patrol data is transmitted to the server where the patrol database is located in real time. If conditions are not met, the patrol data is temporarily stored in the patrol device.

[0063] Specifically, after obtaining the patrol data, the patrol device transmits the patrol image, on-site video surveillance image data to the database server in real time and offline backhaul based on the WAPI standard wireless LAN and other channels, realizing the safe and fast transmission of the original image data. WAPI (Wireless LAN Authentication and Privacy Infrastructure) is a secure transmission protocol for wireless LANs. The main authentication process of WAPI is as follows:

[0064] (1) The wireless client, i.e. the patrol device, first performs 802.11 link negotiation with the WLAN (Wireless Local Area Network) device;

[0065] (2) The WLAN device triggers the authentication process of the wireless client;

[0066] (3) The authentication server performs certificate authentication to complete identity authentication;

[0067] (4) The wireless client and WLAN device negotiate keys;

[0068] (5) The WLAN device allows the wireless client to access the network based on the authentication result.

[0069] In this embodiment, after the transmission of the patrol data (image or video) is completed, image denoising, defogging, anti-shake, enhancement, restoration and other technologies are used to improve the quality of the patrol image data, and finally the processed image data is stored in a database to construct a patrol database.

[0070] The problem to be solved in image denoising is to estimate an unknown clean image given a noisy image, using a sparse coding method called sparse and redundant representation based on dictionary training. This method obtains an overcomplete dictionary through learning, and assumes that each image block in a natural image can be approximately represented by a linear combination of atoms in the dictionary.

[0071] The main task of image defogging technology is to remove the impact of weather factors on image quality, so as to enhance image visibility and improve image quality. The foggy image restoration method is based on the physical process of foggy image degradation, establishes a model to describe image degradation, and uses the dark channel prior algorithm to obtain the relevant parameters of the foggy image imaging model through relevant algorithms, combined with the degradation model, and then infers the real information of the scene, inverts the image degradation process, and restores the foggy blur caused by fog. This processing method is truly defogging in the physical sense, restoring the image to be real and natural, and generally there is no information loss.

[0072] Video jitter refers to the jitter and blur of the video sequence caused by inconsistent motion noise of the camera during shooting. In order to eliminate these jitters, it is necessary to extract the real global motion parameters of the camera, and then use appropriate transformation technology to compensate for the camera's motion to make the video smooth and stable. This technology is usually called video de-jittering or video stabilization. The geometric analysis method is a video de-jittering algorithm based on motion vectors. A new fast and robust estimation method is used to obtain the camera's global motion parameter set; the parameter set is filtered to filter out the motion noise caused by random jitter.

[0073] Image enhancement is to make an image clear or convert it into a form more suitable for human or machine analysis. It can highlight the details of the image and improve the image contrast according to the specific application requirements, thereby improving the visual effect of the image. Image restoration is to use the prior knowledge that causes image degradation to establish an effective mathematical model to describe the image degradation process, and then follow the inverse process of image degradation to design a corresponding solution method to restore the original appearance of the degraded image.

[0074] In step S200, it is determined whether there is a narrow working space hazard in the working space of the patrol database based on the feature expert database of the hidden dangers of narrow working spaces in the distribution network. The feature expert database is obtained based on the historical data of the working space of the distribution network and the results of manual review. The narrow working space hazard refers to the hazard that exists when the positional relationship between the power equipment in the working space of the distribution network constitutes a narrow working space.

[0075] The process of establishing the feature expert database of hidden dangers in narrow working space of distribution network is as follows:

[0076] Based on historical data, we obtain historically accumulated samples of hidden dangers in narrow working spaces, study them, extract feature quantities that can accurately characterize the hidden danger information in narrow working spaces, establish a feature library for hidden dangers in narrow working spaces, and continuously update it based on the latest samples obtained in real time.

[0077] The learning features adopt the convolutional neural network model, which is a kind of deep learning model and the most commonly used model structure in image recognition. The convolutional neural network model adopts a local connection method, and each convolution kernel shares a set of weights, which greatly reduces the number of parameters in the network. The network also uses spatial downsampling to make the learned features invariant to the input samples such as translation, scaling, and distortion, thereby improving the classification accuracy and the generalization ability of the model. Based on deep learning, the original image can be directly input into the neural network model without the need for manual feature design. The deep learning system can learn and summarize the features by itself and achieve a high recognition accuracy.

[0078] In this embodiment, the distribution network operating space of the patrol database is identified by elements, and the power equipment and narrow operating space in the distribution network operating space are identified respectively. There are two factors that cause hidden dangers in the narrow operating space of the distribution network: first, there is power equipment that needs to be operated, and second, the power equipment and other equipment and the on-site environment form a narrow operating space available for operation in space. Therefore, the feature learning of this section is also mainly divided into two parts. One part is to identify the power equipment in the distribution network operating space hidden danger scene, and mark and classify the power equipment in the picture; the other part is to identify the spatial relationship in the distribution network operating space hidden danger scene, such as the narrow space and angle formed by the power equipment and other power equipment, the on-site environment, etc. According to the requirements of the two parts of learning features, the specific structure and parameter settings of the neural network used may be different. Figure 4 As shown, the dotted-line box represents the identified power equipment (element 1), and the solid-line box represents the identified narrow working space (element 2).

[0079] Automatically identify the hidden danger scenarios contained in the historical data, and obtain the hidden danger feature libraries corresponding to power equipment and narrow working spaces respectively. After obtaining these two parts of the hidden danger feature library, manually review the results and supplement typical working pictures to form a hidden danger feature expert library.

[0080] In this embodiment, it is determined whether there are hidden dangers in the operating scenes of the patrol database based on the feature expert database of hidden dangers in the narrow operating space of the distribution network. The specific process is: the patrol data in the patrol database is divided into a training set and a test set, and the patrol pictures, on-site video surveillance pictures and other information in the training set are automatically scanned. Based on the artificial intelligence algorithm and the established feature expert database of hidden dangers in the narrow operating space of the distribution network, it is identified whether there are hidden dangers in the patrol database.

[0081] The artificial intelligence algorithm identification in this embodiment is also divided into two parts. Two trained neural network models are used to respectively identify the power equipment and the narrow working space in the distribution network operating space. The first neural network model is responsible for identifying the characteristics of the power equipment, and the second neural network model is responsible for identifying the characteristics of the narrow working space. The matching degree between the power equipment characteristics and the narrow working space characteristics obtained according to the test set and the hidden danger feature expert library is judged respectively, and the first matching score and the second matching score are obtained respectively; wherein, the first matching score indicates the matching degree between the power equipment in the test set and the hidden danger feature expert library, and the second matching score indicates the matching degree between the narrow working space in the test set and the hidden danger feature expert library.

[0082] In this embodiment, two trained neural network models are used to respectively obtain the hidden danger features of the power equipment and the narrow working space in the test set, and the hidden danger features of the power equipment and the narrow working space are respectively compared with the hidden danger features in the hidden danger feature expert database, and the first matching score and the second matching score are respectively obtained according to the matching degree.

[0083] It is worth noting that the values ​​of the first matching score and the second matching score range from 0 to 100, that is, the lowest score is 0 and the highest score is 100, representing a matching degree ranging from extremely low to extremely high, and the size of the score reflects the degree of matching; of course, other scoring systems can also be used, such as a 10-point system.

[0084] The process of training the hidden danger identification model includes:

[0085] Respectively determining the degree of matching between the power equipment and the narrow working space in the training set and the hidden danger feature expert database and obtaining a third matching score and a fourth matching score, wherein the third matching score indicates the degree of matching between the power equipment in the training set and the hidden danger feature expert database, and the fourth matching score indicates the degree of matching between the narrow working space in the training set and the hidden danger feature expert database;

[0086] A first matching threshold and a second matching threshold are set for the power equipment, and a third matching threshold and a fourth matching threshold are set for the narrow working space. The second matching threshold is greater than the first matching threshold, and the fourth matching threshold is greater than the third matching threshold. When formula 1 is established, it is judged that there is no hidden danger; when formula 2 or formula 3 is established, it is judged that there is a hidden danger; when formula 4 is established, further manual judgment is required;

[0087] Wherein, Formula 1 is: the third matching score < the first matching threshold or the fourth matching score < the third matching threshold; Formula 2 is: the third matching score ≥ the second matching threshold and the fourth matching score ≥ the third matching threshold;

[0088] Formula 3 is: the fourth matching score ≥ the fourth matching threshold and the third matching score ≥ the first matching threshold;

[0089] Formula 4 is: first matching threshold ≤ third matching score ≤ second matching threshold and third matching threshold ≤ fourth matching score ≤ fourth matching threshold;

[0090] The results of manual judgment are input into the hidden danger identification model as part of the training set for learning. The first matching threshold and the third matching threshold are continuously increased manually according to the performance of the hidden danger identification model, and the second matching threshold and the fourth matching threshold are continuously lowered. The process of inputting the results of manual judgment into the hidden danger identification model is repeated until the first matching threshold and the second matching threshold are equal to obtain the first confidence threshold, and the third matching threshold and the fourth matching threshold are equal to obtain the second confidence threshold.

[0091] It is worth noting that the first confidence threshold and the second confidence threshold are two parameters of the hidden danger identification model. The output result of the hidden danger identification model is affected by the first confidence threshold and the second confidence threshold. The result of manual judgment is input into the hidden danger identification model. The first matching threshold and the third matching threshold are continuously increased and the second matching threshold and the fourth matching threshold are continuously reduced by manual means. These thresholds will affect the output result of the hidden danger identification model after adjustment. People can see the output result of the hidden danger identification model. People adjust the thresholds according to the output result of the hidden danger identification model, and the cycle continues until the first matching threshold and the second matching threshold are equal to obtain the first confidence threshold, and the third matching threshold and the fourth matching threshold are equal to obtain the second confidence threshold. When the first confidence threshold and the second confidence threshold are obtained in this embodiment, it means that the hidden danger identification model has been trained at this time, and the trained hidden danger identification model has been obtained, and the training of the hidden danger identification model is ended.

[0092] It is worth noting that the first matching threshold, the second matching threshold, the third matching threshold and the fourth matching threshold are artificially set based on experience. The first matching threshold and the third matching threshold are suspected thresholds. In a preferred implementation, the initial value of the suspected threshold can be set to 60±5. The second matching threshold and the fourth matching threshold are sure thresholds. In a preferred implementation, the initial value of the sure preset can be set to 80±5. The specific offset values ​​of these thresholds can be determined based on the performance of the trained recognition network model. If the performance of the recognition network model is good, the suspected threshold can be increased and the sure threshold can be lowered; if the performance of the recognition network model is poor, the opposite operation is performed.

[0093] It should be noted that, according to actual needs, the first matching threshold and the third matching threshold may be equal or unequal; similarly, the second matching threshold and the fourth matching threshold may be equal or unequal.

[0094] In this embodiment, after obtaining the first confidence threshold and the second confidence threshold, the first matching score is compared with the first confidence threshold, and the second matching score is compared with the second confidence threshold, and it is determined whether there are hidden dangers in the patrol database according to the comparison results, specifically including:

[0095] Determine whether the first matching score is greater than or equal to the first confidence threshold and whether the second matching score is greater than or equal to the second confidence threshold. If so, it is determined that there is a hidden danger in the narrow operating space of the distribution network, otherwise there is no hidden danger; wherein the first confidence threshold and the second confidence threshold are determined when the hidden danger identification model is trained, and correspond to the power equipment and narrow operating space in the narrow operating space of the distribution network respectively.

[0096] In step S300, when there is a hidden danger of narrow working space in the distribution network working space, the severity of the hidden danger is graded and the classification result is stored in the hidden danger database.

[0097] Specifically, when grading the severity of hidden dangers in the narrow working space of the distribution network, the point with the first confidence threshold on the horizontal axis and the second confidence threshold on the vertical axis is taken as the origin, and the point with the highest score on the horizontal axis and the highest score on the vertical axis is taken as the end point. The horizontal axis interval between the first confidence threshold and the highest score is evenly divided into three sections, and the vertical axis interval between the second confidence threshold and the highest score is also evenly divided into three sections. The rectangular area formed by the origin and the end point is evenly divided into 9 areas, and the 9 areas are numbered in sequence from top to bottom and from left to right. The hidden danger corresponding to the third area is a very high-level hidden danger, the hidden danger corresponding to the second, fifth and sixth areas are high-level hidden dangers, and the hidden dangers corresponding to other areas are intermediate-level hidden dangers.

[0098] It is understandable that the horizontal and vertical values ​​corresponding to the third area (i.e., the upper right corner area) are the largest. At this time, the possibility of hidden dangers in the power equipment and the narrow working space of the distribution network is extremely high.

[0099] In step S400, a digital map of hidden dangers in narrow working spaces of the distribution network is obtained based on the inspection database, the hidden danger database and basic information of the distribution network, and visual management and control of hidden dangers in narrow working spaces of the distribution network is performed based on the digital map.

[0100] Specifically, a holographic model database of hidden dangers in narrow working spaces of distribution networks is obtained by modeling based on the patrol database and the hidden danger database, a digital map of distribution networks is obtained by modeling based on basic information of distribution networks, and the holographic model database is embedded in the digital map of distribution networks to obtain a digital map of hidden dangers in narrow working spaces of distribution networks.

[0101] In this embodiment, based on the inspection data and hidden danger identification results, a 360-degree holographic model of the hidden danger scene of the distribution network operation space and its surrounding equipment and environment is established by using three-dimensional modeling technology to form a holographic model database of hidden dangers in narrow operation spaces. Similarly, the basic information of the distribution network, such as the power grid, environment, and equipment, is modeled to form a digital map of the distribution network, and the obtained holographic model database of hidden dangers in narrow operation spaces is embedded in the digital map of the distribution network to form a digital map of hidden dangers in narrow operation spaces of the distribution network, thereby realizing panoramic visualization management and control of hidden dangers in narrow operation spaces.

[0102] The method for identifying hidden dangers in the narrow working space of the distribution network provided by the embodiment of the present invention aims at the hidden dangers of the narrow working space of the distribution network which are difficult to find. The patrol data of the narrow working space of the distribution network is obtained by automatic patrol. The feature expert database of the hidden dangers in the narrow working space is obtained according to the historical data of the distribution network working space and the manual review results. The historical accumulated data can provide data support for the effect of the optimization method; according to the hidden danger feature expert database, it is judged whether there are hidden dangers in the working space of the patrol database, which can effectively reduce the manual workload compared with the purely manual scheme, and at the same time ensure the accuracy of the identification effect; the hidden danger database is obtained by grading the severity of the hidden danger, and the digital map of the hidden dangers in the narrow working space is further obtained. The formed hidden danger database, hidden danger feature expert database and hidden danger digital map can all be updated in real time, and can automatically and efficiently identify the hidden dangers in the narrow working space of the distribution network, so as to facilitate the visual management and control of the hidden dangers in the narrow working space of the distribution network.

[0103] See also Figure 3 Another embodiment of the method for identifying hidden dangers in a narrow working space of a distribution network of the present invention includes: using a dynamic patrol device to patrol the operating space of the distribution network to obtain patrol data, transmitting the patrol data to a server, preprocessing the patrol data and storing it in a patrol database, extracting feature quantities from historical data of the operating space of the distribution network to form a feature expert database of hidden dangers in the narrow working space of the distribution network, using the hidden danger feature expert database to automatically identify hidden dangers in the narrow working space in the operating space of the patrol data, and then constructing a holographic model database of hidden dangers in the narrow working space of the distribution network, further obtaining a digital map of hidden dangers in the narrow working space of the distribution network, and using the digital map of hidden dangers in the narrow working space of the distribution network to visualize the hidden dangers in the narrow working space of the distribution network.

[0104] Power equipment, small working space such as Figure 4 As shown, the information of the patrol database such as patrol images and on-site video surveillance images is automatically scanned, and based on the artificial intelligence algorithm and the established expert database of hidden danger characteristics in small working spaces, it is identified whether there are hidden dangers in the patrol database. The artificial intelligence algorithm identification is also divided into two parts, which are carried out separately by two neural networks. The first neural network is responsible for identifying the elements of power equipment, and the second neural network is responsible for identifying the spatial relationship elements. The degree of match between the data in the patrol database and the case characteristics in the expert database is scored respectively. The matching degree ranges from 0-100, representing from extremely low matching degree to extremely high matching degree. The subsequent operation is determined according to the results identified by the two networks.

[0105] The specific logic for subsequent operation decisions is as follows: when the matching degree of a certain element is lower than the suspected threshold, it is judged that the hidden danger does not exist (area without patterns); when the matching degree of a certain element is higher than the certain threshold and the confidence of another element is not lower than the suspected threshold, it is judged that the hidden danger exists (twill area); when the matching degree of both elements is between the suspected threshold and the certain threshold (the middle square area), the process of the corresponding scene will be handed over to manual identification for processing.

[0106] At the same time, the hidden danger data after manual identification is input into the network again as learning material, so that the accuracy of the network is continuously improved. With the accumulation of experience and continuous learning of the hidden danger identification model, the suspected threshold will continue to increase, and the confident threshold will continue to decrease until the two are equal. The two thresholds are combined into one to become the final confident threshold, so that it is believed that the identification results of the network can be fully trusted, thereby achieving the effect of completely eliminating the need for manual intervention, such as Figure 6 shown.

[0107] Based on the inspection data and hidden danger identification results, a 360-degree holographic model of the hidden dangers in the narrow working space and its surrounding equipment and environment is established using three-dimensional modeling technology to form a holographic model database of hidden dangers in the narrow working space. Similarly, the basic information of the distribution network, such as the power grid, environment, and equipment, is modeled to form a digital map of the distribution network. The obtained holographic model database of hidden dangers in the narrow working space is embedded in the digital map of the distribution network to form a digital map of hidden dangers in the narrow working space, thereby realizing panoramic visualization management and control of hidden dangers in the narrow working space.

[0108] The method for identifying hidden dangers in narrow working spaces of distribution networks provided in the embodiments of the present invention aims at the hidden dangers in narrow working spaces of distribution networks which are difficult to find. An automatic inspection, hidden danger identification and visualization process of automatic inspection, feature extraction, automatic identification and holographic display is constructed. The hidden danger database, hidden danger feature expert database and hidden danger digital map formed are all updated in real time, and the historical accumulated data can provide data support for the effect of the optimization method. The identification is carried out by factors, and measures are selectively taken according to the degree of matching. Compared with pure manual solutions, the intelligent identification solution in this embodiment effectively reduces the manual workload and ensures the accuracy of the identification effect. After mature deployment, no manual intervention is required, which can effectively reduce the workload compared with the current manual and semi-manual solutions.

[0109] The method for identifying hidden dangers in the narrow working space of the distribution network provided in the present embodiment performs automatic inspections and investigations on the hidden dangers in the narrow working space, thereby reducing the inspection workload compared with manual and semi-manual solutions; a hidden danger database and a hidden danger feature expert database are established, and hidden dangers in the narrow working space are automatically identified through artificial intelligence algorithms, thereby increasing the efficiency of discovering hidden dangers; a digital map of hidden dangers in the narrow working space is established, which facilitates the visual management and control of hidden dangers in the narrow working space of the distribution network.

[0110] See also Figure 2 The present invention also provides an embodiment of a device for identifying hidden dangers in narrow working spaces of a distribution network, including:

[0111] A patrol database establishment module is used to obtain patrol data of the distribution network operation space, and save the patrol data to the patrol database after pre-processing, wherein the patrol data includes power equipment and narrow operation spaces existing in the distribution network operation space;

[0112] A hidden danger judgment module is used to judge whether there is a hidden danger of narrow working space in the working space of the patrol database according to the feature expert database of hidden dangers of narrow working space of the distribution network, wherein the feature expert database is obtained according to the historical data of the working space of the distribution network and the results of manual review, and the hidden danger of narrow working space refers to the hidden danger existing when the position relationship between the power equipment in the working space of the distribution network constitutes a narrow working space;

[0113] A hidden danger database establishment module is used to classify the severity of hidden dangers when there are hidden dangers of narrow working space in the distribution network working space, and save the classification results in the hidden danger database;

[0114] The hidden danger automatic identification module is used to obtain a digital map of hidden dangers in the narrow working space of the distribution network based on the patrol database, the hidden danger database and the basic information of the distribution network, and to perform visual management and control of the hidden dangers in the narrow working space of the distribution network based on the digital map.

[0115] In a preferred embodiment, the hidden danger judgment module judges whether there is a hidden danger of narrow working space in the working space of the patrol database according to the feature expert database, including:

[0116] The inspection data in the inspection database are divided into a training set and a test set, and the hidden danger judgment module judges the matching degree between the power equipment and the narrow working space in the test set and the hidden danger feature expert database respectively, and obtains a first matching score and a second matching score, wherein the first matching score indicates the matching degree between the power equipment in the test set and the hidden danger feature expert database, and the second matching score indicates the matching degree between the narrow working space in the test set and the hidden danger feature expert database;

[0117] The first matching score is compared with the first confidence threshold, and the second matching score is compared with the second confidence threshold, and whether there are hidden dangers in the inspection database is determined according to the comparison results; the first confidence threshold and the second confidence threshold are determined according to the hidden danger identification model trained by the training set, and are respectively used to determine whether there are hidden dangers in power equipment and narrow working spaces.

[0118] The device for identifying hidden dangers of narrow working spaces in distribution networks provided in the embodiments of the present invention aims at the hidden dangers of narrow working spaces in distribution networks which are difficult to find. The patrol data of the working spaces of distribution networks are obtained by automatic patrol. The feature expert database of hidden dangers of narrow working spaces in distribution networks is obtained according to the historical data of the working spaces of distribution networks and the results of manual review. The historical accumulated data can provide data support for the effect of the optimization method. The device can judge whether there are hidden dangers of narrow working spaces in the working spaces of patrol databases according to the hidden danger feature expert database. Compared with pure manual schemes, the manual workload can be effectively reduced, and the accuracy of the identification effect can be ensured at the same time. The hidden danger database is obtained by grading the severity of the hidden dangers, and a digital map of the hidden dangers of narrow working spaces is further obtained. The formed hidden danger database, hidden danger feature expert database and hidden danger digital map can all be updated in real time, and can automatically and efficiently identify the hidden dangers of narrow working spaces in distribution networks, so as to facilitate the visual management and control of the hidden dangers of narrow working spaces in distribution networks.

[0119] The hidden danger identification device for the narrow working space of the distribution network provided in the present embodiment performs automatic inspections and investigations on the hidden dangers in the narrow working space, thereby reducing the inspection workload compared with manual and semi-manual solutions; a hidden danger database and a hidden danger feature expert database are established, and hidden dangers in the narrow working space are automatically identified through artificial intelligence algorithms, thereby increasing the efficiency of discovering hidden dangers; a digital map of hidden dangers in the narrow working space is established, which facilitates the visual management and control of hidden dangers in the narrow working space of the distribution network.

[0120] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0121] In the embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0122] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0123] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0124] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0125] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for identifying hidden dangers in narrow working spaces of distribution networks, characterized in that: include: Obtaining patrol data of the distribution network operation space, pre-processing the patrol data and saving it to a patrol database, wherein the patrol data includes power equipment and narrow operation spaces that may exist in the distribution network operation space; Judging whether there is a hidden danger of narrow working space in the working space of the inspection database according to the feature expert database of hidden dangers of narrow working space in the distribution network, wherein the feature expert database is obtained according to the historical data of the working space of the distribution network and the results of manual review, and the hidden danger of narrow working space refers to the hidden danger existing when the position relationship between the power equipment in the working space of the distribution network constitutes a narrow working space; When there is a hidden danger of narrow working space in the distribution network working space, the severity of the hidden danger is graded and the classification results are saved in the hidden danger database; A digital map of hidden dangers in narrow working spaces of the distribution network is obtained based on the inspection database, the hidden danger database and basic information of the distribution network, and visual management and control of hidden dangers in narrow working spaces of the distribution network is performed based on the digital map; Judging whether there is a hidden danger of narrow working space in the working space of the patrol database according to the feature expert database includes: The patrol data in the patrol database are divided into a training set and a test set, and the matching degree between the power equipment and the narrow working space in the test set and the hidden danger feature expert database is determined respectively, and a first matching score and a second matching score are obtained, wherein the first matching score indicates the matching degree between the power equipment in the test set and the hidden danger feature expert database, and the second matching score indicates the matching degree between the narrow working space in the test set and the hidden danger feature expert database; The first matching score is compared with a first confidence threshold, and the second matching score is compared with a second confidence threshold, and whether there are hidden dangers in the patrol database is determined according to the comparison results; the first confidence threshold and the second confidence threshold are determined according to the hidden danger identification model trained by the training set, and are used to judge whether there are hidden dangers in power equipment and narrow working space respectively; Training the hidden danger identification model according to the training set includes: Respectively determining the degree of matching between the power equipment and the narrow working space in the training set and the hidden danger feature expert database and obtaining a third matching score and a fourth matching score, wherein the third matching score indicates the degree of matching between the power equipment in the training set and the hidden danger feature expert database, and the fourth matching score indicates the degree of matching between the narrow working space in the training set and the hidden danger feature expert database; A first matching threshold and a second matching threshold are set for the power equipment, and a third matching threshold and a fourth matching threshold are set for the narrow working space. The second matching threshold is greater than the first matching threshold, and the fourth matching threshold is greater than the third matching threshold. When formula 1 is established, it is judged that there is no hidden danger; when formula 2 or formula 3 is established, it is judged that there is a hidden danger; when formula 4 is established, further manual judgment is required; Wherein, Formula 1 is: the third matching score < the first matching threshold or the fourth matching score < the third matching threshold; Formula 2 is: the third matching score ≥ the second matching threshold and the fourth matching score ≥ the third matching threshold; Formula 3 is: the fourth matching score ≥ the fourth matching threshold and the third matching score ≥ the first matching threshold; Formula 4 is: first matching threshold ≤ third matching score ≤ second matching threshold and third matching threshold ≤ fourth matching score ≤ fourth matching threshold; Inputting the result of manual judgment as part of the training set into the hidden danger identification model for learning, continuously increasing the first matching threshold and the third matching threshold by manual means according to the performance of the hidden danger identification model, and continuously reducing the second matching threshold and the fourth matching threshold, and repeatedly inputting the result of manual judgment into the hidden danger identification model until the first matching threshold and the second matching threshold are equal to obtain the first confidence threshold, and the third matching threshold and the fourth matching threshold are equal to obtain the second confidence threshold; Determining the matching degree between the power equipment and the narrow working space in the test set and the hidden danger feature expert database and obtaining a first matching score and a second matching score includes: Using two trained neural network models, respectively obtaining the hidden danger features of the power equipment and the narrow working space in the inspection database, respectively comparing the hidden danger features of the power equipment and the narrow working space with the hidden danger features in the hidden danger feature expert database, and respectively obtaining a first matching score and a second matching score according to the matching degree; Respectively comparing the first matching score with the first confidence threshold, and the second matching score with the second confidence threshold, and judging whether there is a hidden danger in the patrol database according to the comparison results includes: Determine whether the first matching score is greater than or equal to a first confidence threshold and whether the second matching score is greater than or equal to a second confidence threshold, if so, determine that there is a hidden danger in the narrow working space of the distribution network, otherwise there is no hidden danger; The severity of hidden dangers is graded as follows: The point with the first confidence threshold on the horizontal axis and the second confidence threshold on the vertical axis is taken as the origin, and the point with the highest score on the horizontal axis and the highest score on the vertical axis is taken as the end point. The rectangular area formed by the origin and the end point is evenly divided into 9 areas, and the 9 areas are numbered in sequence from top to bottom and from left to right. The hidden danger corresponding to the third area is a very high-level hidden danger, the hidden danger corresponding to the second, fifth and sixth areas are high-level hidden dangers, and the hidden dangers corresponding to other areas are medium-level hidden dangers.

2. The method for identifying hidden dangers of narrow working space in distribution network according to claim 1 is characterized in that: According to the inspection database, the hidden danger database and the basic information of the distribution network, a digital map of hidden dangers in narrow working spaces of the distribution network is obtained, including: A holographic model database of hidden dangers in narrow working spaces of distribution networks is obtained by modeling based on the patrol database and the hidden danger database; a digital map of distribution networks is obtained by modeling based on basic information of distribution networks; the holographic model database is embedded in the digital map of distribution networks to obtain a digital map of hidden dangers in narrow working spaces of distribution networks.

3. The method for identifying hidden dangers of narrow working space in distribution network according to claim 1 or 2, characterized in that: The hidden danger feature expert database is obtained based on the historical data of the distribution network operation space and the results of manual review, including: The characteristics of power equipment and narrow working space in the historical data of distribution network operating space are identified respectively, and the hidden danger feature library of power equipment and narrow working space is obtained respectively. Combined with the manual review results, the feature expert library of hidden dangers of distribution network narrow working space is obtained.

4. A device for identifying hidden dangers in narrow working spaces of distribution networks, characterized in that: include: A patrol database establishment module is used to obtain patrol data of the distribution network operation space, and save the patrol data to the patrol database after pre-processing, wherein the patrol data includes power equipment and narrow operation spaces existing in the distribution network operation space; A hidden danger judgment module is used to judge whether there is a hidden danger of narrow working space in the working space of the patrol database according to the feature expert database of hidden dangers of narrow working space of the distribution network, wherein the feature expert database is obtained according to the historical data of the working space of the distribution network and the results of manual review, and the hidden danger of narrow working space refers to the hidden danger existing when the position relationship between the power equipment in the working space of the distribution network constitutes a narrow working space; A hidden danger database establishment module is used to classify the severity of hidden dangers when there are hidden dangers of narrow working space in the distribution network working space, and save the classification results in the hidden danger database; The hidden danger automatic identification module is used to obtain a digital map of hidden dangers in the narrow working space of the distribution network based on the patrol database, the hidden danger database and the basic information of the distribution network, and to perform visual management and control of the hidden dangers in the narrow working space of the distribution network based on the digital map.

5. The device for identifying hidden dangers of narrow working space in distribution network according to claim 4, characterized in that: The hidden danger judgment module judges whether there is a hidden danger of narrow working space in the working space of the patrol database according to the feature expert database, including: The inspection data in the inspection database are divided into a training set and a test set, and the hidden danger judgment module judges the matching degree between the power equipment and the narrow working space in the test set and the hidden danger feature expert database respectively, and obtains a first matching score and a second matching score, wherein the first matching score indicates the matching degree between the power equipment in the test set and the hidden danger feature expert database, and the second matching score indicates the matching degree between the narrow working space in the test set and the hidden danger feature expert database; The first matching score is compared with the first confidence threshold, and the second matching score is compared with the second confidence threshold, and whether there are hidden dangers in the inspection database is determined according to the comparison results; the first confidence threshold and the second confidence threshold are determined according to the hidden danger identification model trained by the training set, and are respectively used to determine whether there are hidden dangers in power equipment and narrow working spaces.

Images