Image recognition-based automatic checking method for electricity service cost anomaly

Abstract

The application provides an image recognition-based electricity service fee anomaly automatic checking method, relates to the technical field of data processing, and comprises the following steps: calling an electrical wiring image in a target user electricity supply contract file and a historical identification sequence of the target user, determining a candidate power supply mode type set according to the historical identification sequence; performing hierarchical geometric extraction on the electrical wiring image to obtain a geometric feature combination and an image quality evaluation result; matching the geometric feature combination with a preset standard power supply mode library within the candidate power supply mode type set to obtain a power supply mode type determination result; determining a processing path for the target user according to the power supply mode type determination result, the historical identification sequence and the image quality evaluation result, and outputting an abnormal identification result under the processing path. The application realizes accurate and automatic identification of power supply models and parameter configuration anomalies, significantly reduces omissions caused by manual checking, and greatly improves the overall efficiency of data processing and comparison.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to an automatic verification method for abnormal electricity service charges based on image recognition. Background Technology

[0002] In the electricity business management of power supply companies, the user's power supply method directly determines the basis for calculating electricity consumption parameters. The power supply contract file is the main carrier for recording the user's power supply method, which includes electrical wiring diagrams reflecting the actual wiring status. When there is a difference between the user's actual electricity structure and the power supply method recorded in the contract file, it will lead to deviations in the recording of electricity consumption parameters in the business system. Therefore, accurate identification of the power supply method is a fundamental step in the business anomaly verification work.

[0003] Currently, verification of power supply methods typically relies on manual review of contract files. Business personnel meticulously compare electrical wiring diagrams with the power supply type registered in the business system, verifying the number of incoming circuits and the main wiring structure to ensure consistency with system records, and conducting longitudinal comparisons with historical files. This process requires verification personnel to possess a certain level of drawing interpretation skills. During verification, they need to extract the incoming conductor connections, busbar node distribution, and wiring configuration of each functional unit from the drawings, and then compare them item by item with standard power supply methods. Given the inconsistent quality of archived images, some images may have issues such as blurred lines or incomplete outlines, requiring verification personnel to make judgments based on experience. Different personnel may arrive at different interpretations of the same image.

[0004] The above methods suffer from low identification efficiency and poor consistency of results when dealing with large-scale document verification. Summary of the Invention

[0005] This application provides an automatic verification method for abnormal electricity service charges based on image recognition, in order to solve the problem of how to automatically identify the power supply mode of unstructured electrical wiring images in power supply contract files and accurately determine the abnormality of the service based on the identification results.

[0006] A first aspect of this application provides an automatic verification method for abnormal electricity service charges based on image recognition, comprising: retrieving electrical wiring images from the electricity supply contract file of a target user and the historical identification sequence of the target user, and determining a set of candidate power supply mode types based on the historical identification sequence; Layered geometric extraction is performed on the electrical wiring image to obtain geometric feature combinations and image quality assessment results; The geometric features are combined within the candidate power supply type set and matched with a preset standard power supply type library to obtain the power supply type determination result; Based on the power supply type determination result, the historical recognition sequence, and the image quality assessment result, a processing path for the target user is determined, and anomaly recognition results are output under the processing path.

[0007] Optionally, in one possible implementation of the first aspect, determining the candidate power supply mode type set based on the historical identification sequence includes: Based on the historical identification sequence and the change application record of the target user, power supply mode types that do not appear in the historical identification sequence and do not have corresponding entries in the change application record are removed from the power supply mode types in the standard power supply mode library to obtain an initial candidate set; Based on the degree of matching between the drawing imaging characteristics and drawing type of the electrical wiring image and the standard construction period and drawing method of each power supply type, the power supply type in the initial candidate set is sorted to obtain the candidate power supply type set.

[0008] Optionally, in one possible implementation of the first aspect, the hierarchical geometric extraction of the electrical wiring image to obtain a combination of geometric features includes: Extract the coordinates of the endpoints of the conductor segments in the incoming line area of ​​the electrical wiring image, and obtain the first incoming line circuit number based on the number of the conductor segment endpoints; The first connection node in the bus region of the electrical wiring image is identified based on the first incoming circuit number, and the number and coordinates of the first connection node are extracted. The geometric range of the busbar is determined based on the coordinates of the first connecting node. The contours within the geometric range of the busbar are detected and divided into closed contours and non-closed contours. The contour geometric features and contour coordinate range of each closed contour are extracted, and the number of closed contours and the total number of contours are counted. The first incoming line circuit number, conductor segment endpoint coordinates, number of first connection nodes, coordinates of the first connection nodes, contour geometric features of each closed contour, and contour coordinate range are determined as the geometric feature combination.

[0009] Optionally, in one possible implementation of the first aspect, layered geometric extraction is performed on the electrical wiring image to obtain an image quality assessment result, including: The ratio of the number of interrupted conductor segments to the total number of conductor segments in the incoming line area is used as the segment integrity index; the ratio of the number of closed contours to the total number of contours is used as the contour closure index. When both the line segment integrity index and the contour closure index are not lower than the preset quality threshold, the image quality assessment result is determined to be of usable level. If either the line segment integrity index or the contour closure index is lower than the preset quality threshold, the image quality assessment result is determined to be of low quality. If both the line segment integrity index and the contour closure index are below the preset quality threshold, the image quality assessment result is determined to be unusable.

[0010] Optionally, in one possible implementation of the first aspect, the step of combining the geometric features within the candidate power supply type set and matching them with a preset standard power supply type library to obtain a power supply type determination result includes: The standard incoming circuit number corresponding to each power supply type in the candidate power supply type set is obtained from the standard power supply method library. A first-level candidate set is obtained by performing a first-level screening based on the standard incoming circuit number and the first incoming circuit number. A secondary selection is performed based on the number and coordinates of the first connection nodes and the number and coordinates of the standard connection nodes corresponding to each power supply type in the primary candidate set to obtain a secondary candidate set. The power supply type with the smallest deviation from the contour geometric features of each closed contour in the secondary candidate set is determined as the power supply type determination result.

[0011] Optionally, in one possible implementation of the first aspect, determining the processing path for the target user based on the power supply type determination result, the historical recognition sequence, and the image quality assessment result includes: When the image quality assessment result is determined to be unusable, the processing path is determined to be an auxiliary assessment path; If the image quality assessment result is determined to be of usable or low quality, and the power supply type determination result is inconsistent with the most recent type record in the historical identification sequence, and the target user does not have a corresponding change application record, then the processing path is determined to be a change anomaly marking path. If the image quality assessment result is determined to be of usable or low quality level, and the power supply type determination result is consistent with the most recent type record in the historical identification sequence, or if the target user has a corresponding change application record, then the processing path is determined to be the business verification path.

[0012] Optionally, in one possible implementation of the first aspect, when the processing path is determined to be a change anomaly marker path, the anomaly identification result is output, including: Query the number of standard incoming circuits and the number of standard connection nodes corresponding to the latest type record in the standard power supply method library; When it is determined that the number of standard incoming circuits is less than the number of first incoming circuits and the number of first connection nodes increases accordingly, the abnormal identification result of the main circuit level change not being reported is output.

[0013] Optionally, in one possible implementation of the first aspect, when the processing path is determined to be an auxiliary evaluation path, the anomaly identification result is output, including: The electrical wiring image is geometrically annotated according to the coordinates of the first connection node, the coordinates of the endpoints of the conductor segments, and the range of the contour coordinates. The geometric annotation results are packaged into evaluation task data and sent to the evaluation queue. The confirmation power supply mode type corresponding to the evaluation task data is received. Query the number of standard incoming circuits corresponding to the confirmed power supply method type in the standard power supply method library, and add the confirmed power supply method type to the historical identification sequence; The anomaly identification result is output based on the number of endpoints of the conductor segment and the number of standard incoming circuits.

[0014] Optionally, in one possible implementation of the first aspect, outputting the anomaly identification result based on the number of endpoints of the conductor segment and the number of standard incoming circuits includes: When it is determined that the number of conductor segment endpoints in the incoming line side area of ​​the electrical wiring image matches the standard number of incoming line circuits, the business verification path is entered, and the anomaly identification result is output. When it is determined that the number of conductor segment endpoints in the incoming line side area does not match the number of standard incoming line circuits, the abnormality marking path is entered, and the abnormality identification result is output.

[0015] The automatic verification method for abnormal electricity service charges based on image recognition provided in this application has the following beneficial effects: 1. This application automatically obtains the coordinates of the endpoints of the incoming conductor segments, the connection nodes of the busbar area, and the geometric features of the closed contour by performing layered geometric extraction on electrical wiring images. It constructs a combination of geometric features and narrows down the range of candidate power supply types by combining historical recognition sequences. Within the candidate range, it performs step-by-step matching according to the number of incoming circuits, the number of connection nodes, and the geometric features of the contour. In other words, this application breaks down the image recognition process into a multi-level screening from coarse to fine, gradually narrowing down the matching range, and determining the power supply type based on this. Compared with the manual image reading method, it can automatically determine the power supply type without human intervention, improving the efficiency of processing large batches of files. 2. Based on a comprehensive comparison of image quality assessment results, power supply type determination results, and historical recognition sequences, this application automatically assigns auxiliary assessment paths, change anomaly marker paths, or business verification paths to target users. In other words, by incorporating the consistency of image quality status with historical records into path decision-making, it outputs corresponding structured anomaly recognition results for different situations. Compared with verification methods that rely on human experience, this application can ensure the consistency of anomaly recognition conclusions under different image quality conditions and improve the reliability of business anomaly recognition results. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating the automatic verification method for abnormal electricity service charges based on image recognition provided in the embodiments of this application; Figure 2 This is an application environment diagram of the automatic verification method for abnormal electricity service charges based on image recognition provided in the embodiments of this application; Figure 3 This is a schematic diagram of the power supply type determination process of the automatic verification method for abnormal electricity service fees based on image recognition provided in the embodiments of this application; Figure 4 This is a schematic diagram of the hardware structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0018] The technical solutions of this application will be described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0019] In the electricity business management of power supply companies, the power supply contract archives record the user's power supply method type and corresponding electrical wiring diagram, which is an important basis for verifying whether the user's actual power consumption structure is consistent with the information registered in the business system. With the increase in the number of electricity users, the number of archives continues to expand. Batch identification of electrical wiring images in the contract archives and output of anomaly identification results are key steps to improve the efficiency of business verification.

[0020] The existing verification method usually involves business personnel manually reviewing contract files, reading electrical wiring diagrams one by one, and comparing them with the registration information in the business system. When the quality of the file images varies, the reading conclusions are prone to deviation due to differences in personnel experience, and consistency is difficult to guarantee when processing in batches.

[0021] To address the aforementioned issues, this application proposes an automatic verification method for abnormal electricity service charges based on image recognition, which can be applied to applications such as... Figure 2 In the application environment shown, the file management server stores power supply contract files, and the business processing terminal communicates with the file management server and the business system server via the network. The business processing terminal retrieves electrical wiring images and historical identification sequences from the target user's power supply contract file from the file management server. Based on the historical identification sequences, it determines a set of candidate power supply mode types. It performs layered geometric extraction on the electrical wiring images to obtain geometric feature combinations and image quality assessment results. It matches the geometric feature combinations within the candidate power supply mode type set with a preset standard power supply mode image library to obtain a power supply mode type determination result. Based on the power supply mode type determination result, historical identification sequences, and image quality assessment results, it determines the processing path for the target user and outputs anomaly identification results within that processing path.

[0022] The business processing terminal can be, but is not limited to, various desktop computers, laptops, workstations, and other terminal devices with image processing capabilities. The file management server can be a standalone physical server, a server cluster consisting of multiple physical servers, or a cloud server providing cloud storage services. The business system server stores the target user's electricity consumption parameter data and historical data, allowing the business processing terminal to retrieve parameter values ​​specifically within the business verification path. The standard power supply mode library can be integrated locally on the business processing terminal or deployed on the file management server or a separate database server, accessible via the network.

[0023] See Figure 1 This is a flowchart illustrating the automatic verification method for abnormal electricity service charges based on image recognition provided in this application embodiment. Figure 1The execution entity of the method shown can be a software and / or hardware device. The execution entity of this application can include, but is not limited to, at least one of the following: user equipment, network equipment, etc. User equipment can include, but is not limited to, computers, smartphones, personal digital assistants (PDAs), and the aforementioned electronic devices. Network equipment can include, but is not limited to, a single network server, a server group consisting of multiple network servers, or a cloud based on cloud computing consisting of a large number of computers or network servers. Cloud computing is a type of distributed computing, consisting of a super virtual computer composed of a group of loosely coupled computers. This embodiment does not limit this. Steps 100 to 400 are detailed below: Step 100: Retrieve the electrical wiring images and historical identification sequences of the target user from the power supply contract file, and determine the set of candidate power supply mode types based on the historical identification sequences.

[0024] It should be noted that the power supply contract archive stores the user's electrical wiring diagram in image form, and the historical identification sequence records the identification results of the target user's power supply mode type in each period. Both are retrieved from the archive management system. The candidate power supply mode type set is a subset obtained by filtering all power supply mode types included in the standard power supply mode library in combination with the target user's historical identification sequence. This subset is used to constrain the comparison range of subsequent matching and avoid invalid comparisons of types in the library that are irrelevant to the target user.

[0025] In some embodiments, step 100 includes steps 110 and 120: Step 110: Based on the historical identification sequence and the change application records of the target user, remove the power supply mode types that do not appear in the historical identification sequence and do not have corresponding entries in the change application records from the power supply mode types in the standard power supply mode library to obtain the initial candidate set.

[0026] It should be noted that step 110 performs dual filtering by referring to both the historical identification sequence and the change application record. That is, if a power supply method type has never appeared in the historical identification sequence and does not exist in the change application record, it can be considered that the target user is highly unlikely to use that type, and it is removed from the standard power supply method library, resulting in an initial candidate set. This approach narrows the scope of the candidate set to types that are actually related to the target user's historical electricity consumption, providing a more accurate starting point for the sorting in step 120.

[0027] Step 120: Based on the matching degree between the drawing imaging characteristics and drawing type of the electrical wiring image and the standard construction period and drawing method of each power supply type, sort the power supply type in the initial candidate set to obtain the candidate power supply type set.

[0028] It should be noted that electrical wiring diagrams for power supply facilities built in different eras differ in imaging characteristics and drawing methods. For example, hand-drawn diagrams and computer-aided diagrams differ in line regularity and symbol styles. Furthermore, each power supply method type is marked with its corresponding standard construction period and drawing method in the standard power supply method diagram library. Step 120 uses the above correspondence to sort the power supply method types in the initial candidate set, placing types with higher matching degrees between imaging characteristics and drawing types at the top, so that subsequent matching steps prioritize the more likely types.

[0029] Preferably, step 100 uses two steps—historical record filtering and drawing feature sorting—to gradually narrow down the full range of power supply types in the standard power supply method library to a set of candidate power supply method types that are highly relevant to the target user. This effectively controls the comparison range of subsequent geometric feature matching and reduces the overall computational load for recognition in batch file processing scenarios.

[0030] Step 200: Perform layered geometric extraction on the electrical wiring image to obtain the geometric feature combination and image quality assessment results.

[0031] It should be noted that hierarchical geometric extraction refers to extracting geometric features sequentially from the incoming line area and the busbar area according to the structural hierarchy of the electrical wiring diagram, rather than performing a one-time recognition of the entire image. The electrical wiring diagram has a clear hierarchical structure; that is, the incoming line area reflects the connection between the external power supply and the user, while the busbar area reflects the distribution of circuits within the user's premises. Their positional distribution and graphic features in the image differ, therefore, extracting features hierarchically ensures the specificity and completeness of features for each area. Steps 210 and 220 share the same hierarchical extraction process, outputting geometric feature combinations and image quality assessment results, respectively. These two outputs play different roles in subsequent steps.

[0032] In some embodiments, step 200 includes steps 210 and 220: Step 210: Perform layered geometric extraction on the electrical wiring image to obtain a combination of geometric features.

[0033] In some embodiments, step 210 includes steps 211 to 214: Step 211: Extract the coordinates of the endpoints of the conductor segments in the incoming line area of ​​the electrical wiring image, and obtain the number of the first incoming line circuits based on the number of conductor segment endpoints.

[0034] The incoming line side area is the region located at the top or left of the electrical wiring diagram, reflecting the location of the external power supply. It can be located from the electrical wiring diagram using contour detection or region segmentation based on the layout characteristics of the drawing. The coordinates of the conductor segment endpoints are the coordinate values ​​extracted after endpoint detection of the segments within the incoming line side area. Since each incoming circuit corresponds to a set of conductor segments in the incoming line side area, and the number of conductor segment endpoints corresponds to the number of incoming circuits, the number of the first incoming circuit can be obtained based on the number of conductor segment endpoints.

[0035] It is easy to understand that the number of incoming circuits is the primary feature that distinguishes different power supply types. For example, single-circuit power supply, dual-circuit power supply, and multi-circuit power supply have clear differences in the number of line segment endpoints in the incoming side area. Step 211 converts this difference into a quantifiable first incoming circuit number, providing a basis for the first-level screening in the subsequent matching steps.

[0036] Step 212: Identify the first connection node in the bus region of the electrical wiring image based on the number of the first incoming circuits, and extract the number and coordinates of the first connection node.

[0037] The first connection node is a node within the busbar area used to connect different circuits or realize circuit switching functions, including tie switch nodes and automatic transfer switch nodes. The tie switch node corresponds to the switch symbol between two busbar segments in the image, and the automatic transfer switch node corresponds to the device symbol with automatic switching function in the image. Both types of nodes present specific symbol outlines in the graphic and can be detected from the busbar area through symbol recognition.

[0038] It is understandable that step 212 uses the number of the first incoming line circuits as a prerequisite for identifying nodes in the busbar area because different numbers of incoming line circuits correspond to different busbar structures, and the types and number ranges of the first connection nodes that may exist in the busbar area also differ. Taking dual-circuit power supply as an example, the busbar area usually has tie switch nodes; taking a power supply method with automatic transfer switch function as an example, the busbar area will have automatic transfer switch device nodes. Using the number of the first incoming line circuits to narrow down the range of node identification can reduce misidentification caused by image noise.

[0039] Step 213: Determine the geometric range of the busbar based on the coordinates of the first connecting node, detect the contours within the geometric range of the busbar and divide them into closed contours and non-closed contours, extract the contour geometric features and contour coordinate range of each closed contour, and count the number of closed contours and the total number of contours.

[0040] The busbar geometric range is a rectangular or polygonal area defined by extending a certain region outward from the coordinates of the first connecting node. This range is used to constrain subsequent contour detection within the image area containing the busbar and its associated components. A closed contour is a contour whose ends meet, forming a complete closed boundary. In electrical wiring diagrams, this typically corresponds to components with clearly defined symbolic boundaries, such as transformers and circuit breakers. A non-closed contour is a contour with breaks, not forming a complete closed boundary; this may correspond to line segments or contours broken due to poor image quality. Contour geometric features include parameters describing the shape of the closed contour, such as its area, perimeter, and aspect ratio, calculated from each detected closed contour. The contour coordinate range is the coordinates of the circumscribed rectangle of each closed contour in the image, extracted from the contour detection results.

[0041] It should be noted that in traditional manual image recognition, the image recognizer relies on experience to determine which symbol regions in the image belong to functional components. Step 213 distinguishes the contours within the busbar region into closed contours and non-closed contours, and uses closed contours to correspond to functional component symbols, thus transforming this recognition process into a computable geometric detection operation, eliminating the reliance on experience in manual image recognition.

[0042] Step 214: Determine the first incoming line loop number, conductor segment endpoint coordinates, first connection node number, first connection node coordinates, contour geometric features of each closed contour, and contour coordinate range as a geometric feature combination.

[0043] It is easy to understand that steps 211 to 213 extract geometric quantities at different levels in the incoming line side area and the busbar area, respectively. Step 214 summarizes the extraction results of the above levels into a unified geometric feature combination, which serves as the standardized input for matching with the standard power supply mode library in the subsequent step 300.

[0044] Preferably, step 210 extracts geometric features layer by layer in the order from the incoming line side region to the bus region. There is a correlation between the extraction results of each layer. That is, the number of the first incoming line loop extracted in the incoming line side region guides the node recognition range of the bus region, and the coordinates of the first connecting node in the bus region further constrain the contour detection range. This layer-by-layer correlation extraction method makes the components in the final summarized geometric feature combination have an inherent structural consistency. Compared with a one-time feature extraction of the entire image, the geometric information of each region is more accurate.

[0045] Step 220: Perform layered geometric extraction on the electrical wiring image to obtain the image quality assessment results.

[0046] In some embodiments, step 220 includes steps 221 to 224: Step 221: Use the ratio of the number of interrupted conductor segments in the incoming line area to the total number of conductor segments as the segment integrity index; use the ratio of the number of closed contours to the total number of contours as the contour closure index.

[0047] The number of interrupted conductor segments refers to the number of conductor segments with obvious breaks and failure to extend continuously within the incoming line area, calculated from the segment detection results in step 211. The total number of conductor segments refers to the total number of conductor segments detected within the incoming line area, also calculated from the detection results in step 211. The number of closed contours and the total number of contours are calculated from the contour detection results in step 213. The segment integrity index reflects the integrity of conductor segments within the incoming line area, while the contour closure index reflects the integrity of component symbol contours within the busbar area. These two indicators correspond to the two areas in the electrical wiring diagram most susceptible to image quality issues.

[0048] The preset quality threshold is a reference value for judging whether the line segment integrity index and the contour closure index meet the subsequent recognition requirements. Those skilled in the art can select it according to the imaging quality distribution of the actual archival image. That is to say, if there are a large number of images with low scanning quality in the archive, the preset quality threshold can be appropriately relaxed; if the overall imaging of the archival image is clear, the preset quality threshold can be tightened to improve the recognition accuracy requirements.

[0049] Step 222: When both the line segment integrity index and the contour closure index are not lower than the preset quality threshold, the image quality assessment result is determined to be of usable level.

[0050] Step 223: If either the line segment integrity index or the contour closure index is lower than the preset quality threshold, the image quality assessment result is determined to be of low quality.

[0051] Step 224: If both the line segment integrity index and the contour closure index are below the preset quality threshold, the image quality assessment result is determined to be unusable.

[0052] It should be noted that the classification of image quality assessment results into three levels—usable, low-quality, and unusable—is based on three combinations: both indicators are met, only one is met, and neither is met. The reason for distinguishing between low-quality and unusable levels, rather than simply dividing into usable and unusable categories, is that when only one indicator is low, the image still retains some usable geometric information, and subsequent steps may still be able to complete the identification based on this information. However, when both indicators are low, the image has significant deficiencies in both the incoming line and busbar regions. In this case, the reliability of directly relying on geometric extraction results for power supply mode matching cannot be guaranteed, requiring manual intervention in the auxiliary evaluation path. This three-level classification provides a more precise and specific basis for the selection of processing paths, avoiding deviations in identification conclusions caused by improper handling of image quality boundary conditions.

[0053] It is understandable that steps 222 to 224 have fully covered all combinations of the line segment integrity index and the contour closure index: if both are not lower than the preset quality threshold, the level is usable; if both are lower than the preset quality threshold, the level is unusable; if exactly one is lower than the preset quality threshold, the level is low quality. The three cases do not overlap and are fully covered.

[0054] It should be noted that step 220 reuses the line segment detection and contour detection results already completed in step 210, without requiring reprocessing of the electrical wiring image. The line segment integrity index and contour closure index are directly obtained from the detection results of steps 211 and 213. During the scanning, storage, or transmission of archival images, line segments may be broken or contours may be blurred. If the geometric extraction results are directly used for power supply matching without evaluating image quality, the reliability of matching conclusions for low-quality images cannot be guaranteed. Step 220, by quantifying the geometric integrity of two key regions, transforms the judgment of image quality from relying on human experience into quantifiable index calculations, providing an objective basis for the selection of processing paths in subsequent step 400.

[0055] Preferably, step 200 simultaneously outputs the geometric feature combination and image quality assessment results within the same set of hierarchical geometric extraction processes. Both outputs share intermediate detection results from the incoming line region and the busbar region, avoiding redundant processing of the electrical wiring images. The geometric feature combination is used for power supply type matching, and the image quality assessment results are used for subsequent processing path selection. Both play their respective roles in subsequent steps, jointly supporting the accurate identification and anomaly determination of the target user's power supply mode.

[0056] Step 300: Combine the geometric features within the candidate power supply type set and match them with the preset standard power supply type library to obtain the power supply type determination result.

[0057] It should be noted that the standard power supply method library includes standard geometric parameters corresponding to various power supply method types, including the number of standard incoming circuits, the number and coordinates of standard connection nodes, and the standard outline geometric features of each functional component. These are all reference values ​​pre-compiled and stored in the library according to the standard wiring specifications of various power supply methods. Step 300 compares the candidate power supply method type set with the corresponding entries in the standard power supply method library, rather than traversing all types in the library. This further controls the amount of matching calculation based on the narrowed candidate range already achieved in step 100.

[0058] In some embodiments, step 300 includes steps 310 to 330: Step 310: Obtain the number of standard incoming circuits corresponding to each power supply type in the candidate power supply type set from the standard power supply method library, and perform first-level filtering based on the number of standard incoming circuits and the number of first incoming circuits to obtain the first-level candidate set.

[0059] The standard number of incoming circuits is the standard value of the number of incoming circuits corresponding to each power supply method type in the standard power supply method library, which is obtained by querying the standard power supply method library. The first-level screening rule is as follows: power supply method types whose standard number of incoming circuits is consistent with the first number of incoming circuits are retained in the first-level candidate set, and all power supply method types whose standard number of incoming circuits is inconsistent with the first number of incoming circuits are excluded, regardless of whether the standard number of incoming circuits is greater than or less than the first number of incoming circuits, they are not included in the first-level candidate set.

[0060] It should be noted that the number of incoming circuits is one of the most distinguishing features between different power supply types. Single-circuit, dual-circuit, and multi-circuit power supplies have fundamental differences in wiring structure, and their corresponding standard incoming circuit counts are different. Using the number of incoming circuits as the first screening dimension can quickly narrow down the candidate range with minimal computation, allowing step 320 to only compare nodes and coordinates among power supply types with the same number of circuits, thus eliminating unnecessary comparisons for types with obviously inconsistent circuit counts.

[0061] Step 320: Perform secondary screening based on the number and coordinates of the first connection nodes and the number and coordinates of the standard connection nodes corresponding to each power supply type in the primary candidate set to obtain the secondary candidate set.

[0062] The number and coordinates of standard connection nodes are obtained from the standard power supply method library, which contains node parameters corresponding to each type in the primary candidate set. The secondary screening process involves the following steps: First, the number of first connection nodes is compared with the number of standard connection nodes; power supply method types with inconsistent numbers are directly excluded. For power supply method types with consistent numbers, the spatial deviation between the coordinates of the first and standard connection nodes is further compared. Power supply method types with spatial deviations within a preset coordinate tolerance range are retained in the secondary candidate set, while those with deviations exceeding the preset coordinate tolerance range are excluded. The preset coordinate tolerance range is an upper limit for coordinate deviation pre-set based on the scanning resolution of the archival image and the drawing error. Those skilled in the art can select this range based on the actual imaging resolution of the archival image and the drawing specifications; lower imaging resolution or larger drawing error typically corresponds to a more lenient preset coordinate tolerance range.

[0063] Understandably, among power supply types with the same number of incoming circuits, the number and location distribution of tie switch nodes and automatic transfer switch nodes can further distinguish types with similar wiring structures but different node configurations. For example, different types of dual-circuit power supply, one with a tie switch and the other without, will differ in the number and coordinate distribution of nodes in the busbar area. Step 320 transforms these differences into screening criteria through a joint comparison of node number and coordinates, further narrowing the secondary candidate set to the type range most closely matching the node configuration of the target user's electrical wiring image.

[0064] Step 330: Determine the power supply type with the smallest deviation from the contour geometric features of each closed contour in the secondary candidate set as the power supply type determination result.

[0065] The calculation method for the profile geometric feature deviation is as follows: For each power supply type in the secondary candidate set, the standard profile geometric features of each corresponding standard functional element are obtained from the standard power supply type library, including standard area, standard perimeter, and standard aspect ratio. These are compared item by item with the profile geometric features of each closed profile extracted in step 213. The absolute values ​​of the differences between each feature are summed to obtain the total profile geometric feature deviation for that power supply type. After completing the above calculation for all power supply types in the secondary candidate set, the power supply type with the smallest total profile geometric feature deviation is determined as the power supply type determination result.

[0066] It should be noted that after the two rounds of screening in steps 310 and 320, the remaining power supply types in the secondary candidate set are highly consistent with the geometric feature combination in terms of the number of incoming circuits and the configuration of connection nodes. Step 330 uses the minimum total deviation of the contour geometric features as the final judgment criterion to determine the type that is closest to the overall electrical wiring image of the target user within the secondary candidate set. The contour geometric features reflect the shape characteristics of the symbols of each functional element in the busbar area and are a refined basis for distinguishing the differences in wiring structure details. They are used at the end of the tertiary screening to ensure the accuracy of the judgment results at the element symbol level.

[0067] See Figure 3 The power supply mode type determination process in step 300 adopts a three-level progressive screening structure. S1 is the candidate power supply mode type set, which is the range of power supply mode types related to the target user output in step 100. A first-level screening is performed based on the standard number of incoming circuits, retaining types with the same number of circuits, resulting in the first-level candidate set S2. A second-level screening is performed based on the number and coordinates of connection nodes, retaining types with consistent node configurations, resulting in the second-level candidate set S3. Within the range of S3, the minimum deviation of the contour geometric features is used as the judgment criterion, resulting in the power supply mode type determination result S4. Each level of screening excludes types that do not meet the conditions, gradually narrowing the candidate range, and finally completing a refined determination within a set with relatively low computational complexity.

[0068] Preferably, step 300 completes the power supply type determination through a three-level screening process, from coarse to fine. The three screening dimensions are, in order, the number of incoming circuits, the configuration of connection nodes, and the geometric features of the profile, with the granularity of differentiation increasing progressively. The granularity of differentiation based on the number of incoming circuits is the coarsest but has the highest screening efficiency, while the granularity of differentiation based on the geometric features of the profile is the finest but also has the largest computational load. The three-level screening concentrates the computational load on a gradually narrowed candidate set, so that the overall computational load is effectively controlled while ensuring the accuracy of the determination.

[0069] Step 400: Determine the processing path for the target user based on the power supply type determination result, historical recognition sequence, and image quality assessment result, and output the anomaly recognition result under the processing path.

[0070] It should be noted that the determination of the processing path integrates three inputs: the image quality assessment result from step 220, the consistency comparison conclusion between the power supply type determination result from step 330 and the most recent type record in the historical identification sequence, and the query result of the target user's change application record. The combination of these three inputs corresponds to three processing paths: the auxiliary evaluation path, the change anomaly marking path, and the business verification path. Each path outputs different types of anomaly identification results.

[0071] In some embodiments, step 400 includes steps 410 to 430: Step 410: When the image quality assessment result is determined to be unusable, the processing path is determined to be the auxiliary assessment path.

[0072] Step 420: If the image quality assessment result is determined to be of usable or low quality level, and the power supply type determination result is inconsistent with the most recent type record in the historical recognition sequence, and the target user does not have a corresponding change application record, then the processing path is determined to be the change anomaly marking path.

[0073] Step 430: If the image quality assessment result is determined to be of usable or low quality level, and the power supply type determination result is consistent with the most recent type record in the historical recognition sequence, or if the target user has a corresponding change application record, then the processing path is determined to be the business verification path.

[0074] It is understandable that steps 410 to 430 have fully covered all cases of the combination of image quality assessment results and power supply type determination results: when the image quality assessment result is unusable, regardless of the power supply type determination result, the auxiliary assessment path is entered; when the image quality assessment result is usable or low quality, the change anomaly marking path or business verification path is entered respectively, depending on the consistency between the power supply type determination result and the historical recognition sequence and the existence of the change application record. The three paths do not overlap and are fully covered.

[0075] It should be noted that image quality assessment results are used as the primary criterion for path selection because image quality directly determines the reliability of the power supply type determination result in step 300. When the image quality assessment result is unusable, both the line segment integrity index and the contour closure index in the geometric feature combination are significantly missing. The reliability of the power supply type determination result output in step 300 cannot be guaranteed. If anomaly marking is directly based on the power supply type determination result, there is a risk of misjudging image quality problems as power supply type change anomalies. Therefore, the auxiliary assessment path is prioritized for manual confirmation, rather than directly outputting anomaly identification conclusions.

[0076] In some embodiments, when the determined processing path in step 420 is a change anomaly marker path, the anomaly identification result is output, including steps 421 and 422: Step 421: Query the latest type record in the standard power supply method library for the number of standard incoming circuits and the number of standard connection nodes.

[0077] The most recent type record is the most recent power supply mode type identification record in the historical identification sequence, extracted from the historical identification sequence. The number of standard incoming circuits and standard connection nodes corresponding to the most recent type record are the standard parameter values ​​corresponding to the most recent type record in the standard power supply mode diagram library, obtained by querying the standard power supply mode diagram library. The purpose of step 421 is to obtain the number of circuits and nodes corresponding to the power supply mode type confirmed by the target user in the previous identification cycle at the standard level, providing a comparison benchmark for determining the nature of the change in step 422.

[0078] Step 422: When the number of standard incoming circuits is less than the number of first incoming circuits and the number of first connection nodes increases accordingly, output the abnormal identification result of the main circuit level change not being reported.

[0079] The corresponding increase in the number of first connection nodes means that the number of first connection nodes is greater than the number of standard connection nodes found in step 421, and the increase in the number corresponds to the increase in the number of incoming circuits in the wiring specifications. Specifically, for each additional incoming circuit, there is usually an additional tie switch node or automatic transfer switch node in the busbar area. If the increase in the number of first connection nodes and the increase in the number of incoming circuits conform to the above correspondence, it is determined that the number of nodes has increased accordingly; if the number of first connection nodes has not increased or the increase does not conform to the above correspondence, it is determined that the number of nodes has not increased accordingly.

[0080] In other embodiments, it also includes: If the number of standard incoming circuits is less than the number of first incoming circuits, and the number of first connection nodes has not increased accordingly, the business verification path is entered.

[0081] When it is determined that the number of standard incoming line circuits is the same as the number of first incoming line circuits and the number of first connection nodes is greater than the number of standard connection nodes, the business verification path is entered.

[0082] Understandably, the above classification is based on the correspondence between changes in the number of incoming line circuits and changes in the number of connected nodes. If the standard number of incoming line circuits is less than the first number of incoming line circuits and the number of the first connected nodes increases accordingly, it indicates a complete expansion of the main circuit structure, with changes consistent with the increase in circuits on both the incoming and bus sides. This is a main circuit-level change, and no anomalies are reported for the output main circuit-level change. If the standard number of incoming line circuits is less than the first number of incoming line circuits but the number of the first connected nodes does not increase accordingly, it indicates a change in the number of circuits on the incoming side, but the node configuration in the bus area has not been adjusted accordingly. The nature of the change is uncertain, and further verification is required through the business verification path. If the standard number of incoming line circuits is the same as the first number of incoming line circuits, but the number of the first connected nodes is greater than the number of standard connected nodes, it indicates that the number of main circuits has not changed, but the node configuration within the bus area has been adjusted. This is a node-level change, and verification is also required through the business verification path. The three scenarios do not overlap. An increase in the number of incoming circuits and a corresponding increase in nodes are classified as an anomaly due to a change not reported at the main circuit level. An increase in the number of incoming circuits but no corresponding increase in nodes, and an increase in the number of incoming circuits but no change in the number of nodes are all classified as business verification paths. All scenarios have been covered.

[0083] It should be noted that steps 421 and 422 further locate the power supply mode change anomaly to the main circuit level by comparing the standard parameters corresponding to the most recent type record with the geometric features extracted from the current electrical wiring image. The output anomaly identification result contains the hierarchical information of the change, enabling business processing personnel to directly obtain the change location conclusion without having to manually judge the nature of the anomaly.

[0084] In some embodiments, when the processing path determined in step 410 is an auxiliary evaluation path, the anomaly identification result is output, including steps 411 to 413: Step 411: Perform geometric annotation on the electrical wiring image based on the coordinates of the first connection node, the coordinates of the endpoints of the conductor segments, and the range of contour coordinates. Package the geometric annotation results into evaluation task data and send them to the evaluation queue. Receive the confirmation power supply mode type corresponding to the evaluation task data.

[0085] Geometric annotation refers to overlaying the coordinates of the first connection node, the endpoint coordinates of the conductor segment, and the contour coordinate range extracted in step 210 onto the electrical wiring image in the form of graphic markers, forming a geometric annotation result. Specifically, endpoint markers are marked on the coordinate positions of each conductor segment endpoint in the incoming line area, node markers are marked on the coordinate positions of each first connection node in the busbar area, and border markers are marked on the contour coordinate range of each closed contour, enabling assessors to visually locate the positions of each geometric feature on the image. The assessment task data is a packaged collection of geometric annotation results and target user profile information, which is sent to the assessment queue for manual processing by assessors. The confirmed power supply mode type is the power supply mode type manually confirmed by the assessors after reviewing the geometric annotation results, and is received from the return results of the assessment queue.

[0086] It should be noted that when the image quality assessment result is unusable, the electrical wiring image shows significant missing geometric information in both the incoming line and busbar areas, making it impossible for automatic identification to provide a reliable power supply type determination. Step 411 does not directly submit the original image to manual evaluation; instead, it sends the geometrically annotated image to the evaluators. With the aid of the annotation results, evaluators can more quickly determine the power supply type without having to reread the drawings from scratch. This reduces the reliance of manual evaluation on the evaluators' drawing interpretation experience and minimizes discrepancies in conclusions caused by different evaluators using varying interpretation standards.

[0087] Step 412: Query and confirm the number of standard incoming circuits corresponding to the power supply method type in the standard power supply method library, and add the confirmed power supply method type to the historical identification sequence.

[0088] It is easy to understand that step 412 adds the manually confirmed power supply mode type to the historical identification sequence, so that the output conclusion of the auxiliary evaluation path can participate in the candidate set generation of step 100 in the next identification cycle, maintaining the continuous integrity of the historical identification sequence. At the same time, the number of standard incoming circuits corresponding to the confirmed power supply mode type is queried from the standard power supply mode library, providing a benchmark value for the anomaly identification comparison in step 413.

[0089] Step 413: Output the anomaly identification result based on the number of conductor segment endpoints and the number of standard incoming circuits.

[0090] In some embodiments, step 413, which outputs anomaly identification results based on the number of conductor segment endpoints and the number of standard incoming circuits, includes steps A1 and A2:

[0091] Step A1: When the number of conductor segment endpoints in the incoming line area of ​​the electrical wiring image matches the standard number of incoming line circuits, proceed to the business verification path and output the anomaly identification result.

[0092] Step A2: When the number of conductor segment endpoints in the incoming line area does not match the standard number of incoming line circuits, proceed to the change anomaly marking path and output the anomaly identification result.

[0093] The condition that the number of conductor segment endpoints matches the standard number of incoming circuits means that the number of conductor segment endpoints extracted in step 211 matches the number of standard incoming circuits corresponding to the confirmed power supply type queried in step 412. The condition that they do not match means that the number of conductor segment endpoints does not match the standard number of incoming circuits, including both cases where the number of conductor segment endpoints is greater than the standard number of incoming circuits and cases where the number of conductor segment endpoints is less than the standard number of incoming circuits; both are categorized into step A2. Steps A1 and A2 completely cover both matching and non-matching cases, and there are no cases that cannot be categorized into any branch.

[0094] It should be noted that, based on the manual confirmation of the power supply method type, steps A1 and A2 further compare the number of conductor segment endpoints in the incoming line area with the standard number of incoming line circuits. Even if the image quality is insufficient to support automatic power supply method matching, the number of conductor segment endpoints in the incoming line area can usually still be extracted from the image. Therefore, steps A1 and A2 utilize this relatively reliable geometric quantity to connect the auxiliary evaluation path with the business verification path and the change anomaly marking path, enabling the identification process to provide a definitive anomaly identification conclusion even when the image quality is unavailable, rather than simply terminating the process.

[0095] It should be noted that when the automatic identification cannot directly provide a judgment conclusion, steps 411 to 413 assist manual confirmation through geometric annotation, add historical records, and output anomaly identification conclusion based on the confirmation result, thus forming a complete processing closed loop under the auxiliary evaluation path, ensuring that the anomaly identification result still has traceability even when the image quality is unavailable.

[0096] In some embodiments, when the processing path determined in step 430 is a business verification path, the anomaly identification result is output, including steps 431 to 433: Step 431: Based on the power supply mode type determination result, determine the corresponding power consumption parameter field set in the standard power supply mode library, extract the parameter values ​​of the target user from the business system according to the power consumption parameter field set, and obtain the power consumption parameter data.

[0097] The power consumption parameter field set is a list of parameter fields corresponding to the power supply mode type determination result from the standard power supply mode diagram library. It is obtained by querying the standard power supply mode diagram library and reflects the power consumption parameter items that need to be checked under the power supply mode type corresponding to the power supply mode type determination result. These can be parameter fields directly related to the wiring structure, such as circuit capacity and transformer capacity. The content of the power consumption parameter field set varies for different power supply mode types. The power consumption parameter data consists of target user parameter values ​​extracted from the business system according to the power consumption parameter field set and returned by the business system.

[0098] Step 431 uses the power supply mode type determination result as an index to retrieve the set of power consumption parameter fields corresponding to the power supply mode type determination result from the standard power supply mode library, and then extracts parameter values ​​from the business system in a targeted manner, avoiding indiscriminate comparison of all parameters in the business system. Different power supply mode types correspond to different sets of power consumption parameter fields, and targeted extraction can accurately limit the scope of business verification to parameter fields directly related to the power supply mode type.

[0099] Step 432: When it is determined that there is a circuit capacity parameter in the power consumption parameter data that corresponds to the number of the first incoming circuits, the circuit capacity parameter is compared with the corresponding historical circuit capacity parameter in the historical record data, and the comparison result is output as the anomaly identification result.

[0100] The circuit capacity parameter is the capacity parameter value corresponding to the circuit number indicated by the first incoming circuit number in the power consumption parameter data, obtained from the power consumption parameter data extracted in step 431; the historical circuit capacity parameter is the historical capacity record value corresponding to the circuit number in the historical record data, obtained from the historical record data of the business system. The comparison result is the difference between the circuit capacity parameter and the historical circuit capacity parameter. When the two are consistent, a comparison result without anomalies is output; when the two are inconsistent, a comparison result indicating an abnormal change in the circuit capacity parameter is output, both of which are output as anomaly identification results.

[0101] Step 433: If the circuit capacity parameter corresponding to the number of the first incoming circuit is not found in the power consumption parameter data, output the abnormal identification result of missing power consumption parameters.

[0102] It is understandable that steps 432 and 433 cover two scenarios in the power consumption parameter data: the presence or absence of a circuit capacity parameter corresponding to the number of the first incoming circuits. These two scenarios do not overlap. When a circuit capacity parameter exists, the parameter value is compared with historical circuit capacity parameters to determine if it has changed. When a circuit capacity parameter does not exist, it indicates that the capacity parameter corresponding to the number of incoming circuits is not registered in the business system, which is considered a missing parameter record, resulting in an output power consumption parameter missing anomaly. Both scenarios provide definitive anomaly identification results, and there are no cases where the anomaly cannot be categorized into either branch.

[0103] It should be noted that, under the premise that the power supply mode type determination result is reliable, step 430 transforms the identification conclusion into a basis for verifying the power consumption parameter data in the business system. By comparing the circuit capacity parameter with the historical circuit capacity parameter, parameter anomalies are identified, or by detecting whether there are any anomalies in the circuit capacity parameter identification record, so that the anomaly identification result under the business verification path and the geometric identification conclusion of the power supply mode type form a complete verification chain, rather than just stopping at the judgment at the level of power supply mode type.

[0104] Preferably, step 400 integrates three inputs—image quality assessment result, power supply type determination result, consistency with historical identification sequences, and existence of change application records—into the path decision. This ensures that different processing methods have clearly corresponding paths, and the output anomaly identification results cover various anomaly types such as unreported main circuit level changes, missing power parameters, and abnormal changes in circuit capacity parameters. Regardless of image quality, step 400 can output a definitive anomaly identification conclusion. Compared to relying on manual image-by-image verification, the consistency and traceability of the identification conclusions are guaranteed. See also Figure 4 This is a schematic diagram of the hardware structure of an electronic device provided in an embodiment of this application. The electronic device 40 includes: a processor 41, a memory 42, and a computer program; wherein, The memory 42 is used to store the computer program, and the memory may also be flash memory. The computer program is, for example, an application program or functional module that implements the above method.

[0105] The processor 41 is configured to execute the computer program stored in the memory to implement the various steps performed by the device in the above method. For details, please refer to the relevant descriptions in the preceding method embodiments.

[0106] Alternatively, the memory 42 can be either standalone or integrated with the processor 41.

[0107] When the memory 42 is a device independent of the processor 41, the device may further include: Bus 43 is used to connect the memory 42 and the processor 41.

[0108] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

Claims

1. An automatic verification method for abnormal electricity service charges based on image recognition, characterized in that, include: Retrieve electrical wiring images from the target user's power supply contract file and the target user's historical identification sequence, and determine a set of candidate power supply mode types based on the historical identification sequence; Layered geometric extraction is performed on the electrical wiring image to obtain geometric feature combinations and image quality assessment results; The geometric features are combined within the candidate power supply type set and matched with a preset standard power supply type library to obtain the power supply type determination result; Based on the power supply type determination result, the historical recognition sequence, and the image quality assessment result, a processing path for the target user is determined, and anomaly recognition results are output under the processing path.

2. The method according to claim 1, characterized in that, The step of determining the candidate power supply mode type set based on the historical identification sequence includes: Based on the historical identification sequence and the change application record of the target user, power supply mode types that do not appear in the historical identification sequence and do not have corresponding entries in the change application record are removed from the power supply mode types in the standard power supply mode library to obtain an initial candidate set; Based on the degree of matching between the drawing imaging characteristics and drawing type of the electrical wiring image and the standard construction period and drawing method of each power supply type, the power supply type in the initial candidate set is sorted to obtain the candidate power supply type set.

3. The method according to claim 1, characterized in that, The step of performing layered geometric extraction on the electrical wiring image to obtain a combination of geometric features includes: Extract the coordinates of the endpoints of the conductor segments in the incoming line area of ​​the electrical wiring image, and obtain the first incoming line circuit number based on the number of the conductor segment endpoints; The first connection node in the bus region of the electrical wiring image is identified based on the first incoming circuit number, and the number and coordinates of the first connection node are extracted. The geometric range of the busbar is determined based on the coordinates of the first connecting node. The contours within the geometric range of the busbar are detected and divided into closed contours and non-closed contours. The contour geometric features and contour coordinate range of each closed contour are extracted, and the number of closed contours and the total number of contours are counted. The first incoming line circuit number, conductor segment endpoint coordinates, number of first connection nodes, coordinates of the first connection nodes, contour geometric features of each closed contour, and contour coordinate range are determined as the geometric feature combination.

4. The method according to claim 3, characterized in that, Layered geometric extraction is performed on the electrical wiring image to obtain image quality assessment results, including: The ratio of the number of interrupted conductor segments to the total number of conductor segments in the incoming line area is used as the segment integrity index; the ratio of the number of closed contours to the total number of contours is used as the contour closure index. When both the line segment integrity index and the contour closure index are not lower than the preset quality threshold, the image quality assessment result is determined to be of usable level. If either the line segment integrity index or the contour closure index is lower than the preset quality threshold, the image quality assessment result is determined to be of low quality. If both the line segment integrity index and the contour closure index are below the preset quality threshold, the image quality assessment result is determined to be unusable.

5. The method according to claim 3, characterized in that, The step of combining the geometric features within the candidate power supply type set and matching them with a preset standard power supply type library to obtain a power supply type determination result includes: The standard incoming circuit number corresponding to each power supply type in the candidate power supply type set is obtained from the standard power supply method library. A first-level candidate set is obtained by performing a first-level screening based on the standard incoming circuit number and the first incoming circuit number. A secondary selection is performed based on the number and coordinates of the first connection nodes and the number and coordinates of the standard connection nodes corresponding to each power supply type in the primary candidate set to obtain a secondary candidate set. The power supply type with the smallest deviation from the contour geometric features of each closed contour in the secondary candidate set is determined as the power supply type determination result.

6. The method according to claim 1, characterized in that, The step of determining the processing path for the target user based on the power supply type determination result, the historical recognition sequence, and the image quality assessment result includes: When the image quality assessment result is determined to be unusable, the processing path is determined to be an auxiliary assessment path; If the image quality assessment result is determined to be of usable or low quality, and the power supply type determination result is inconsistent with the most recent type record in the historical identification sequence, and the target user does not have a corresponding change application record, then the processing path is determined to be a change anomaly marking path. If the image quality assessment result is determined to be of usable or low quality level, and the power supply type determination result is consistent with the most recent type record in the historical identification sequence, or if the target user has a corresponding change application record, then the processing path is determined to be the business verification path.

7. The method according to claim 6, characterized in that, When the processing path is determined to be a change anomaly marker path, the anomaly identification result is output, including: Query the number of standard incoming circuits and the number of standard connection nodes corresponding to the latest type record in the standard power supply method library; When it is determined that the number of standard incoming circuits is less than the number of first incoming circuits and the number of first connection nodes increases accordingly, the abnormal identification result of the main circuit level change not being reported is output.

8. The method according to claim 7, characterized in that, When the processing path is determined to be an auxiliary evaluation path, the anomaly identification result is output, including: The electrical wiring image is geometrically annotated according to the coordinates of the first connection node, the coordinates of the endpoints of the conductor segments, and the range of the contour coordinates. The geometric annotation results are packaged into evaluation task data and sent to the evaluation queue. The confirmation power supply mode type corresponding to the evaluation task data is received. Query the number of standard incoming circuits corresponding to the confirmed power supply method type in the standard power supply method library, and add the confirmed power supply method type to the historical identification sequence; The anomaly identification result is output based on the number of endpoints of the conductor segment and the number of standard incoming circuits.

9. The method according to claim 8, characterized in that, The step of outputting the anomaly identification result based on the number of endpoints of the conductor segment and the number of standard incoming circuits includes: When it is determined that the number of conductor segment endpoints in the incoming line side area of ​​the electrical wiring image matches the standard number of incoming line circuits, the business verification path is entered, and the anomaly identification result is output. When it is determined that the number of conductor segment endpoints in the incoming line side area does not match the number of standard incoming line circuits, the abnormality marking path is entered, and the abnormality identification result is output.

10. The method according to claim 1, characterized in that, When the processing path is determined to be a business verification path, the anomaly identification result is output, including: Based on the power supply mode type determination result, the corresponding set of power consumption parameter fields is determined in the standard power supply mode library, and the parameter values ​​of the target user are extracted from the business system according to the set of power consumption parameter fields to obtain the power consumption parameter data; When it is determined that there is a circuit capacity parameter in the power consumption parameter data that corresponds to the number of first incoming circuits, the circuit capacity parameter is compared with the corresponding historical circuit capacity parameter in the historical data, and the comparison result is output as the anomaly identification result. If it is determined that there is no circuit capacity parameter in the power consumption parameter data corresponding to the number of first incoming circuits, an abnormal identification result of missing power consumption parameters is output.

Images