Electric energy meter wiring judgment method and system

Abstract

The invention provides an electric energy meter wiring judgment method and system. The method comprises the steps of collecting an electric energy meter wiring process image of a student; based on the electric energy meter wiring process image, color feature separation is carried out, and whether wires of all colors are connected according to a positive sequence or not is judged; through template matching and edge detection, whether the bending angle of the wire meets the specification or not is judged; the height difference between the wiring gap and the horizontal line is detected through an image vision algorithm, and the wire tightening state is judged; the positions of the ribbons are positioned, the distance value between every two adjacent ribbons is calculated, and whether the binding distance meets the standard or not is judged; utilizing an OCR character recognition model to recognize character information of a wiring number mark, and judging whether the number is correct or not; the tightening pressure value of the wiring screw is obtained through the annular force sensor, and whether the crimping state is qualified or not is judged in combination with a deep learning algorithm. According to the invention, various operations of students in the wiring process of the electric energy meter can be automatically monitored, and the efficiency and accuracy of subsequent wiring scoring of the electric energy meter are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of the electric power industry, and in particular to an electric energy meter wiring arbitration method and system. Background Art

[0002] In the training and teaching practice of the power industry, the wiring of electric energy meters is a crucial skill. Traditionally, the assessment of trainees' wiring skills often relies on manual proctoring and grading, which is not only inefficient but also easily affected by subjective factors, resulting in unfair and inaccurate grading results. In addition, manual proctoring is difficult to fully capture every detail of the trainees' wiring process, especially key safety operations and wiring processes, which may miss important assessment information.

[0003] Although the existing automated assessment systems have improved the assessment efficiency to a certain extent, they often have problems such as single function, low recognition accuracy, and imperfect scoring mechanism. Summary of the invention

[0004] The technical problem to be solved by the present invention is to provide a method and system for judging the connection of electric energy meters, which can automatically monitor various operations of trainees during the connection process of electric energy meters, thereby improving the efficiency and accuracy of subsequent scoring of electric energy meter connection.

[0005] In order to solve the above technical problems, the technical solution adopted by the present invention is: A method for judging the connection of an electric energy meter comprises the following steps: S1. Collect images of students’ electric energy meter wiring process; S2. Based on the wiring process image of the electric energy meter: S21, performing color feature separation to determine whether the wires of each color are connected in a positive order; S22, judging whether the bending angle of the wire is in compliance with the regulations through template matching and edge detection; S23, detecting the height difference between the wiring gap and the horizontal line through an image vision algorithm to determine the tightening state of the wires; S24, locate the cable tie position, calculate the spacing between adjacent cable ties, and determine whether the bundling spacing is compliant; S25, using an OCR character recognition model to recognize character information of the wiring number line label, and determining whether the number is correct; S26. Obtain the tightening pressure value of the wiring screw through the annular force sensor, and determine whether the crimping state is qualified by combining the deep learning algorithm.

[0006] In order to solve the above technical problems, another technical solution adopted by the present invention is: A power meter wiring referee system includes a control module, an image acquisition module, an image recognition module, a scoring module and an interaction module. A computer program is stored on the control module. When the control module executes the computer program, the steps in the power meter wiring referee method as described above are implemented.

[0007] The beneficial effects of the present invention are as follows: a method and system for judging the wiring of an electric energy meter are provided, which realizes the automated judging of the wiring process of the electric energy meter by collecting images of the wiring process of the trainee's electric energy meter and performing color feature separation, template matching and edge detection, image vision algorithm detection, image annotation, character recognition and crimping status detection on the images, and can accurately identify the access sequence of wires of different colors, the horizontality and verticality of the wiring process, the tightening state of the wires, the compliance of the bundling spacing, the correctness of the wiring number line mark, and the crimping state of the wire wiring screws, thereby jointly ensuring the accuracy, standardization and efficiency of the evaluation of the wiring of the electric energy meter, and can comprehensively cover all aspects of the wiring work, from wire access to screw crimping, thereby realizing all-round monitoring of the wiring of the trainee's electric energy meter and improving the efficiency and accuracy of the subsequent scoring of the wiring of the electric energy meter. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Figure 1 A flow chart of an electric energy meter wiring arbitration method according to an embodiment of the present invention; Figure 2 A schematic diagram of the structure of an electric energy meter wiring referee system according to an embodiment of the present invention; Figure 3 It is a structural schematic diagram of a face recognition unit in an electric energy meter wiring referee system according to an embodiment of the present invention; Figure 4 It is a structural schematic diagram of a protective equipment identification unit in an electric energy meter wiring referee system according to an embodiment of the present invention; Figure 5 It is a structural schematic diagram of a behavior recognition unit in an electric energy meter wiring referee system according to an embodiment of the present invention; Figure 6 The present invention is a schematic diagram of the structure of a line identification unit in an electric energy meter wiring referee system according to an embodiment of the present invention.

[0009] Description of labels: 1. Control module; 2. Image acquisition module; 3. Image recognition module; 31. Face recognition unit; 311. Face acquisition subunit; 312. Face image preprocessing subunit; 313. Face image feature extraction subunit; 314. Face image matching and recognition subunit; 32. Protective equipment recognition unit; 321. Labeling subunit; 322. Model training subunit; 323. Detection subunit; 33. Behavior recognition unit; 331. Dangerous area entry recognition subunit; 332. Personnel fall recognition subunit ; 333. Smoking identification subunit in the work area; 34. Line identification unit; 341. Wire access sequence identification subunit; 342. Wiring process judgment subunit; 343. Wire tightening identification subunit; 344. Bundling spacing identification subunit; 345. Wiring number identification subunit; 346. Wire wiring screw pressure point identification subunit; 4. Scoring module; 41. Deduction judgment unit; 42. Deduction execution unit; 43. Verification unit; 44. Recording unit; 45. Scoring summary unit; 5. Interaction module. DETAILED DESCRIPTION

[0010] In order to explain the technical content, achieved objectives and effects of the present invention in detail, the following is an explanation in conjunction with the implementation modes and the accompanying drawings.

[0011] Please refer to Figure 1 , a method for judging the connection of an electric energy meter, comprising the steps of: S1. Collect images of students’ electric energy meter wiring process; S2. Based on the wiring process image of the electric energy meter: S21, performing color feature separation to determine whether the wires of each color are connected in a positive order; S22, judging whether the bending angle of the wire is in compliance with the regulations through template matching and edge detection; S23, detecting the height difference between the wiring gap and the horizontal line through an image vision algorithm to determine the tightening state of the wires; S24, locate the cable tie position, calculate the spacing between adjacent cable ties, and determine whether the bundling spacing is compliant; S25, using an OCR character recognition model to recognize character information of the wiring number line label, and determining whether the number is correct; S26. Obtain the tightening pressure value of the wiring screw through the annular force sensor, and determine whether the crimping state is qualified by combining the deep learning algorithm.

[0012] From the above description, it can be seen that the beneficial effects of the present invention are: providing an electric energy meter wiring judging method, by collecting images of the trainee's electric energy meter wiring process, and performing color feature separation, template matching and edge detection, image vision algorithm detection, image annotation, character recognition and crimping status detection on the image, to realize the automated judging of the electric energy meter wiring process, and can accurately identify the access sequence of wires of different colors, the horizontality and verticality of the wiring process, the tightening state of the wires, the compliance of the bundling spacing, the correctness of the wiring number line mark, and the crimping state of the wire wiring screws, thereby ensuring the accuracy, standardization and efficiency of the electric energy meter wiring evaluation, and can comprehensively cover all aspects of the wiring work, from wire access to screw crimping, to realize all-round monitoring of the trainee's electric energy meter wiring, and improve the efficiency and accuracy of subsequent electric energy meter wiring scoring.

[0013] Furthermore, in step S21, color feature separation is performed to determine whether the wires of each color are connected in a positive order, specifically: The image of the wiring process of the electric energy meter is separated into three channels of color images through an artificial intelligence visual algorithm, and yellow, green, red and black wires are obtained according to color differences. It is then determined whether the wires of each color are connected in the correct sequence based on the electric energy metering instrument installation wiring scoring standard.

[0014] From the above description, it can be seen that the color separation of the wires can be effectively achieved through the artificial intelligence visual algorithm, thereby improving the accuracy of the judgment of the positive sequence access of the wires.

[0015] Furthermore, the step S22 comprises the steps of: S221, extracting the edge straight line of the wire in the wiring process image of the electric energy meter by an edge detection method; S222, performing fitting analysis on the bending angle of the wire; S223: If the bending angle is within the preset compliance range, it is judged as qualified, otherwise it is marked as unqualified.

[0016] From the above description, it can be seen that after edge detection of the wiring, straight line detection and angle fitting are performed to effectively realize the horizontality and verticality detection of the wire bending angle, thereby improving the accuracy of wire bending detection.

[0017] Furthermore, the step S24 comprises the steps of: S241, performing image annotation and target detection on the cable tie positions identified in the wiring process image of the electric energy meter; S242, measuring the pixel distance between adjacent cable ties, and converting the actual spatial distance based on the equipment calibration data; S243: Compare the actual spatial distance with a preset upper limit of the bundling spacing threshold to determine compliance of the bundling spacing.

[0018] From the above description, it can be seen that the position of the cable tie when it is tightened is marked through image annotation, and the actual distance between adjacent cable ties is effectively calculated through target detection, thereby improving the accuracy of judging whether the spacing between ties is compliant.

[0019] Furthermore, the step S25 is specifically as follows: S251, extracting the wiring number characters in the wiring process image of the electric energy meter through an OCR character recognition model; S252, verifying the height consistency, orientation consistency and completeness of the wiring number characters, and comparing them with the wiring template. If there is an error, back-to-back or occlusion, it is judged as non-standard.

[0020] From the above description, it can be seen that the OCR character recognition model trained by the deep learning architecture can realize accurate recognition of wiring number characters and improve the accuracy of judgment of wiring number character specifications.

[0021] Furthermore, the step S26 is specifically as follows: S261, obtaining the tightening pressure data of the connection screws by fixing the annular force sensor under the screws by identifying the connection process image of the electric energy meter; S262. Call the deep learning model to determine whether the pressure value of the tightening pressure data of the wiring screw is within the standard range. If the pressure value is not within the standard range or touches the insulation layer of the wire, it is marked as unqualified.

[0022] From the above description, it can be seen that the sensor can effectively determine whether the wire has been crimped reliably and whether it has been crimped to the wire insulation resulting in no signal transmission by sensing the degree of tightening of the wiring screw. The sensor can also use the trained data through the deep learning architecture to identify the normal pressure value of the wire crimping, thereby further accurately determining the unqualified crimping mode for subsequent accurate evaluation.

[0023] Furthermore, after step S1, the method further includes: Based on the electric energy meter wiring process image, trainee identity information recognition, trainee protective equipment wearing status recognition, and trainee behavior recognition during the electric energy meter wiring process; The student identity information recognition includes face collection, face image preprocessing, face image feature extraction and face image feature matching; The identification of the wearing status of the trainees' protective equipment includes labeling the wearing status of the protective equipment, training a target detection model according to the labeling results of the wearing status of the protective equipment, and detecting the wearing status of the protective equipment; The trainee's behavioral action recognition during the process of wiring the electric energy meter includes recognition of entry into a dangerous area, recognition of a person falling to the ground, and recognition of smoking in a work area.

[0024] From the above description, it can be seen that through image acquisition and image recognition technology, the system can automatically and quickly capture and analyze the key features of the trainees in the process of wiring the electricity meter, such as identity information, the wearing of protective equipment, and behavioral actions, etc., which improves the scoring efficiency and reduces the subjectivity and errors that may be caused by manual scoring, thereby ensuring the accuracy and fairness of the scoring. The identification of the wearing of protective equipment helps to ensure that the trainees comply with safety regulations during the wiring process and reduce the risk of accidents. At the same time, the identification of the trainees' behavioral actions can identify whether the trainees' wiring actions comply with standard operating procedures, further improving the standardization of operations.

[0025] Furthermore, after step S2, the method further includes: S3. Based on the judgment result of step S2, the wiring process of each student is scored in combination with the built-in scoring mechanism, specifically: S31. Determine whether the connection process of each student triggers the deduction mechanism according to the judgment result and the built-in scoring mechanism, match the deduction standard, calculate the deduction score according to the deduction standard, and obtain the deduction result; S32, verifying the deduction result; S33, recording the deduction results and reasons for the deduction; S34, summarizing the deduction results and generating a final score and outputting it visually.

[0026] From the above description, it can be seen that based on the judgment results and the built-in scoring mechanism, through the five links of deduction judgment, execution, verification, recording and summary, the functions of accurate triggering of deductions, strict verification of results, transparent recording of deduction reasons and efficient summary of scores are realized. It has the technical advantages of high accuracy, traceability and intelligence, and through automatic matching of deduction standards and intelligent score calculation, the deduction process is precise, efficient and standardized, which significantly improves the fairness, efficiency and reliability of scoring.

[0027] Furthermore, the step S2 also includes: Unsafe behaviors are detected in real time, and when the unsafe behaviors are detected, real-time alarms are triggered.

[0028] From the above description, it can be seen that unsafe behaviors are monitored in real time during the trainees' electricity meter wiring operations. Once potential safety hazards or illegal operations are identified, an alarm is triggered to improve safety.

[0029] Please refer to Figures 2 to 6 A power meter wiring referee system includes a control module, an image acquisition module, an image recognition module, a scoring module and an interaction module. A computer program is stored on the control module. When the control module executes the computer program, the steps in the power meter wiring referee method as described above are implemented.

[0030] From the above description, it can be seen that the beneficial effects of the present invention are: based on the same technical concept, in conjunction with the above-mentioned electric energy meter wiring judging method, an electric energy meter wiring judging system is provided, which realizes the automated judging of the electric energy meter wiring process by collecting images of the trainee's electric energy meter wiring process and performing color feature separation, template matching and edge detection, image vision algorithm detection, image annotation, character recognition and crimping status detection on the image, and can accurately identify the access sequence of wires of different colors, the horizontality and verticality of the wiring process, the tightening state of the wires, the compliance of the bundling spacing, the correctness of the wiring number line mark, and the crimping state of the wire wiring screws, thereby ensuring the accuracy, standardization and efficiency of the electric energy meter wiring evaluation, and can comprehensively cover all aspects of the wiring work, from wire access to screw crimping, to achieve all-round monitoring of the trainee's electric energy meter wiring, and improve the efficiency and accuracy of subsequent electric energy meter wiring scoring.

[0031] The present invention provides an electric energy meter wiring adjudication method and system, which are mainly used in the scenario of electric energy meter wiring in the training and teaching practice of the power industry, and are specifically described below in conjunction with specific embodiments: Please refer to Figure 1 , Embodiment 1 of the present invention is: A method for judging the connection of an electric energy meter, such as Figure 1 As shown, the steps include: S1. Collect images of students’ electric energy meter wiring process; S2, based on the wiring process image of the energy meter: S21, performing color feature separation to determine whether the wires of each color are connected in a correct sequence.

[0032] S22. Determine whether the bending angle of the wire is in compliance with regulations through template matching and edge detection.

[0033] S23. Detect the height difference between the wiring gap and the horizontal line through an image vision algorithm to determine the tightening state of the wire.

[0034] S24, locate the cable tie position, calculate the spacing between adjacent cable ties, and determine whether the bundling spacing is compliant.

[0035] S25. Using the OCR character recognition model to recognize the character information of the wiring number line label, determine whether the number is correct.

[0036] S26. Obtain the tightening pressure value of the wiring screw through the annular force sensor, and determine whether the crimping state is qualified by combining the deep learning algorithm.

[0037] That is, in this embodiment, by collecting images of the student's electricity meter wiring process and performing color feature separation, template matching and edge detection, image vision algorithm detection, image annotation, character recognition and crimping status detection on the image, automated judging of the electricity meter wiring process is achieved, and the access sequence of wires of different colors, the horizontality and verticality of the wiring process, the tightening state of the wires, the compliance of the bundling spacing, the correctness of the wiring number line markings, and the crimping state of the wire wiring screws can be identified with high precision, thereby ensuring the accuracy, standardization and efficiency of the electricity meter wiring evaluation, and being able to comprehensively cover all aspects of the wiring work, from wire access to screw crimping, thereby achieving all-round monitoring of the student's electricity meter wiring and improving the efficiency and accuracy of subsequent electricity meter wiring scoring.

[0038] Embodiment 2 of the present invention is: A method for judging the connection of an electric energy meter, based on the above-mentioned embodiment 1, in this embodiment, color feature separation is performed in step S21 to judge whether each color wire is connected in a positive sequence, specifically: The artificial intelligence visual algorithm is used to separate the wiring process image of the electric energy meter into three channels of color images. The yellow wire, green wire, red wire and black wire are detected based on the color differences in the separated three channels. Based on the requirements of the document "Electric Energy Metering Equipment Installation Wiring Scoring Standards": the wires are connected in positive sequence, the U phase of the electric energy meter voltage and current is connected to the yellow wire, the V phase is connected to the green wire, and the W phase is connected to the red wire. It is judged whether the wires of each color are connected in positive sequence. The artificial intelligence visual algorithm is used to effectively realize the color separation of the wires, thereby improving the accuracy of the judgment of positive sequence connection of the wires.

[0039] Meanwhile, in this embodiment, step S22 includes the following steps: S221. Extracting the wire edge straight line in the wiring process image of the electric energy meter by using an edge detection method.

[0040] S222. Perform fitting analysis on the bending angle of the conductor.

[0041] S223: If the bending angle is within the preset compliance range, it is judged as qualified, otherwise it is marked as unqualified. The preset compliance range is 80-100 degrees or 170-190 degrees.

[0042] That is, the wiring position is found through template matching positioning, and straight line detection and angle fitting are performed after edge detection of the wiring, and then the horizontal and vertical wires are detected for horizontal and vertical degrees. In this embodiment, the wire bending is defined as 90 degrees and 180 degrees as the standard, and the wire bending detection angle is 80~100 degrees and 170~190 degrees. The angle within this range is judged as qualified, and other angles can be judged as unqualified by deducting points. After edge detection of the wiring, straight line detection and angle fitting are performed to effectively realize the horizontality and verticality detection of the wire bending angle, and improve the accuracy of wire bending detection.

[0043] At the same time, in this embodiment, the tightening state of the wires can also be based on the requirements of the "Electricity Metering Equipment Installation Wiring Scoring Standards" document, and the tightening and loosening of the wires can be identified through the artificial intelligence visual algorithm. Specifically, the tightening state of the wires can be judged by the height difference between the horizontal wiring gap and the horizontal line.

[0044] Meanwhile, in this embodiment, step S24 is specifically as follows: S241. Perform image annotation and target detection on the cable tie positions identified in the wiring process image of the electric energy meter.

[0045] S242, measuring the pixel distance between adjacent cable ties, and converting the actual spatial distance based on the equipment calibration data.

[0046] S243: Compare the actual space distance with the preset upper limit of the bundling spacing threshold to determine compliance of the bundling spacing.

[0047] That is, according to the requirements of the document "Electricity Metering Equipment Installation and Wiring Scoring Standards", the bundling spacing can be identified through artificial intelligence visual algorithms, and the output is the number of pixels between two adjacent cable ties. Specifically, the position of the black or white rolled strap when it is tightened can be marked through image annotation tools, and the actual distance between adjacent cable ties can be effectively calculated through target detection. The compliance of the bundling spacing can be judged based on the pre-set upper limit of the bundling spacing threshold, thereby improving the accuracy of the judgment on whether the bundling spacing is compliant.

[0048] Meanwhile, in this embodiment, step S25 is specifically as follows: S251. Extract the wiring number characters in the wiring process image of the electric energy meter through an OCR character recognition model.

[0049] S252. Check the height consistency, orientation consistency and completeness of the wiring number characters, and compare them with the wiring template. If there is an error, back-to-back or obstruction, it is judged as non-standard.

[0050] That is, the wiring number is also identified based on the artificial intelligence visual algorithm. The OCR character recognition model trained by the deep learning architecture can accurately recognize the wiring number characters, and compare them with the wiring template to determine whether the number is correct, thereby improving the accuracy of the judgment of the wiring number character specification. In this embodiment, the wiring number character recognition input is a captured wiring number line mark picture. , the minimum length of a single character is .

[0051] Meanwhile, in this embodiment, step S26 is specifically as follows: S261. Obtain the tightening pressure data of the wiring screws by identifying the wiring process image of the electric energy meter and fixing the annular force sensor under the screws.

[0052] S262. Call the deep learning model to determine whether the pressure value of the tightening pressure data of the wiring screw is within the standard range. If the pressure value is not within the standard range or touches the insulation layer of the wire, it is marked as unqualified.

[0053] That is, the sensor can effectively determine whether the wire has been crimped reliably and whether it has been crimped to the wire insulation resulting in no signal transmission by sensing the degree of tightening of the wiring screws. The sensor can also use the trained data through a deep learning architecture to identify the normal pressure value for wire crimping, thereby further accurately determining unqualified crimping patterns for subsequent accurate evaluation.

[0054] That is, in this embodiment, through artificial intelligence visual algorithms and deep learning models, it is possible to accurately identify the access order of wires of different colors, the horizontality and verticality of the wiring process, the tightening state of the wires, the compliance of the bundling spacing, the correctness of the wiring number line markings, and the crimping state of the wire wiring screws. This ensures the accuracy and standardization of the electricity meter wiring evaluation, and can comprehensively cover all aspects of the wiring work, from wire access to screw crimping, thereby achieving all-round monitoring of the trainees' electricity meter wiring.

[0055] Embodiment 3 of the present invention is: A method for judging the connection of an electric energy meter, based on the above-mentioned embodiment 1 or embodiment 2, in this embodiment, after step S1, further includes: Based on the image of the electric energy meter wiring process, the trainee’s identity information, the trainee’s protective equipment wearing status, and the trainee’s behavioral actions during the electric energy meter wiring process are identified.

[0056] The identification of trainees' identity information includes face collection, face image preprocessing, face image feature extraction and face image feature matching; the identification of trainees' protective equipment wearing status includes labeling of protective equipment wearing status, training of target detection model according to the labeling results of protective equipment wearing status, and detection of protective equipment wearing status; the identification of trainees' behavioral actions during the wiring process of electric meters includes identification of entry into dangerous areas, identification of people falling to the ground, and identification of smoking in the work area.

[0057] That is, in this embodiment, the collection and identification of the images of the electricity meter wiring process are based on image acquisition technology and image recognition technology. The system can automatically and quickly capture and analyze the key features of the trainees in the electricity meter wiring process, such as identity information, protective equipment wearing status, and behavioral actions, etc., which improves the scoring efficiency and reduces the subjectivity and errors that may be caused by manual scoring, thereby ensuring the accuracy and fairness of the scoring. The identification of the wearing status of protective equipment helps to ensure that trainees comply with safety regulations during the wiring process and reduce the risk of accidents. At the same time, the identification of trainees' behavioral actions can identify whether the trainees' wiring actions comply with standard operating procedures, further improving the standardization of operations.

[0058] In this embodiment, face collection can be performed by detecting the image collected by the shooting device to obtain a face feature image, then performing face image preprocessing and face image feature extraction to obtain face features, and finally matching the face features with the feature template to determine the student's identity information. The specific process can be as follows: The face model is used to detect and locate the face in the input image, and the detected face is aligned and corrected; the face key point detection model is used to extract features from the corrected face image, and the similarity is compared with the face key point feature information in the database, so as to achieve face recognition. The face images of different students are collected by the acquisition device, and the collected face images include static images and dynamic images. When the student is within the shooting range of the acquisition device, the acquisition device automatically searches and shoots the student's face image; then the face position and size are detected and calibrated to extract the face features, which include histogram features, color features, template features, structural features and Haar features. In this embodiment, in the face detection process, the Adaboost algorithm is used to select rectangular features (weak classifiers) that can represent the face, and the obtained weak classifiers are constructed as strong classifiers in a weighted voting manner. Then, several trained strong classifiers are connected in series to form a cascaded stacked classifier, which can effectively improve the detection speed of the classifier; and the face image preprocessing mainly includes light compensation, grayscale transformation, histogram equalization, normalization, geometric correction, filtering and sharpening of the face image; the face features extracted from the face image mainly include visual features, pixel statistical features, face image transformation coefficient features and face image algebraic features; finally, the face features to be identified are compared with the obtained face feature templates, and the identity information of the face is judged according to the similarity. The face image feature matching searches and matches the extracted face features with the feature templates stored in the database. When the similarity between the matched features and the face features exceeds a preset threshold, the matching results are output.

[0059] Among them, the wearing conditions of protective equipment include the wearing conditions of safety helmets, the wearing conditions of insulating gloves, the wearing conditions of insulating shoes (boots), and the wearing conditions of work clothes. Safety helmet wearing recognition can realize the detection of trainees wearing safety helmets and not wearing safety helmets. Among them, the colors of safety helmets include red, yellow, blue and white. In this embodiment, the targets wearing or not wearing safety helmets of different colors can be labeled by image annotation tools, and the corresponding target detection model can be called through the deep learning architecture to classify and train the labeled targets, so as to realize the detection of safety helmet wearing by personnel on the site; at the same time, insulating glove wearing recognition can realize the detection of wearing insulating gloves and not wearing insulating gloves. In this embodiment, the hand parts of people wearing or not wearing insulating gloves can be labeled by image annotation tools, and the corresponding target detection model can be called through the deep learning architecture to classify and train the labeled targets, so as to realize the detection of insulating gloves worn by personnel on the site, wherein insulating gloves include long-tube gel-like and white cotton-like; at the same time, insulating shoes (boots) wearing recognition It is possible to detect whether insulating shoes (boots) are worn or not. In this embodiment, the foot parts of people wearing or not wearing insulating shoes (boots) can be annotated by image annotation tools, and the corresponding target detection model is called through the deep learning architecture to classify and train the annotated targets, thereby realizing the detection of insulating shoes (boots) worn by personnel on the site, for example, including outdoor insulating shoes, military green insulating boots, white casual insulating shoes and not wearing insulating shoes; at the same time, the detection of wearing work clothes and not wearing work clothes can be realized through work clothes wearing recognition. In this embodiment, the body parts wearing or not wearing work clothes can be annotated by image annotation tools, and the corresponding target detection model is called through the deep learning architecture to classify and train the annotated targets, thereby realizing the detection of work clothes worn by personnel on the site.

[0060] That is, in this embodiment, by automatically identifying the wearing conditions of safety helmets, insulating gloves, insulating shoes (boots) and work clothes, the trainee's safety protection status can be monitored in real time to improve the accuracy of the scoring of the electric energy meter wiring process, and can improve the trainee's safety awareness and reduce safety accidents caused by not wearing or wearing improper protective equipment. Compared with the traditional manual inspection method, this embodiment uses deep learning technology for automatic identification, which can greatly improve the detection efficiency and accuracy and reduce human misjudgment and missed judgments; at the same time, this embodiment supports the identification of protective equipment of various colors and types, such as safety helmets of different colors, insulating gloves and insulating shoes (boots) of different materials, etc., and has good versatility. When new protective equipment needs to be detected, the recognition model can be easily updated and expanded. By combining the recognition results with the database, intelligent personnel management and safety protection decision support can be achieved.

[0061] In this embodiment, the identification of entry into dangerous areas can be achieved by monitoring dangerous areas and drawing areas in the monitoring image for human intrusion detection, and multi-area detection is allowed. When a person enters a dangerous area, the drawn area turns red for early warning, a screenshot is saved, and an event is recorded. In this embodiment, the frame difference method or background method can be used to find out whether there is an area with a large grayscale change in the current frame to determine whether there is target movement. When target movement is detected, the human body detection model is called to identify the type of moving target, and the human body positioning frame is used to determine whether a person has entered the designated dangerous area, and then an alarm is issued; at the same time, the identification of falling people can be achieved by drawing areas in the monitoring image for human falling detection, and multi-area detection is allowed. When a person falls, the drawn area turns red for early warning, a screenshot is saved, and an event is recorded. In this embodiment, image annotation tools can be used to identify various types of people falling. The posture is annotated, and the corresponding state detection model is called through the deep learning architecture to train such annotated targets, so as to realize the recognition and positioning of the posture of the person falling down in the real scene; at the same time, the working area smoking recognition sub-can be realized by monitoring the trainee's working area and using the smoking detection algorithm to recognize the trainee's smoking behavior. When smoking behavior occurs, the picture is captured and saved, and the event is recorded. In this embodiment, the image annotation tool can be used to annotate the hand holding cigarette or the mouth smoking part when the person smokes, and the corresponding target detection model is called through the deep learning architecture to perform category training on such annotated targets. When the human body detection model detects the human body, the smoking part of the person is detected, so as to realize the detection of smoking behavior of people in the real scene.

[0062] That is, in this embodiment, through the danger zone entry recognition, the intrusion of personnel in the danger zone can be monitored in real time, timely warning and measures can be taken to effectively prevent the occurrence of safety accidents; through the personnel fall recognition, the personnel fall situation can be quickly identified and warned, providing valuable time for emergency rescue and reducing the degree of injury; through the work area smoking recognition, smoking behavior in the work area can be discovered and stopped in time to avoid safety hazards such as fire caused by illegal operations; at the same time, the deep learning model and smoking detection algorithm are used to automatically and intelligently identify various behaviors, which greatly improves the monitoring efficiency and accuracy. In addition, this embodiment allows multi-area detection and can monitor multiple areas at the same time to achieve all-round and no-dead-angle monitoring.

[0063] Meanwhile, after step S2, the method further includes: S3. Based on the judgment result of step S2, the wiring process of each student is scored in combination with the built-in scoring mechanism, specifically: S31. According to the judgment result and the built-in scoring mechanism, it is judged whether the connection process of each student triggers the deduction mechanism, and the deduction standard is matched. The deduction score is calculated according to the deduction standard to obtain the deduction result.

[0064] Among them, accurate deduction judgment can be made through key features to ensure that the deduction mechanism is triggered accurately, which helps to avoid unnecessary deductions and maintain the fairness and accuracy of the scoring. At the same time, after the deduction mechanism is triggered, the deduction mechanism can be quickly executed through the deduction execution, and the deduction results can be accurately calculated according to the deduction standards.

[0065] S32, verifying the deduction result.

[0066] That is, the deduction results are strictly checked to ensure the correctness of the deductions, thereby improving the efficiency and accuracy of scoring.

[0067] S33. Record the deduction results and reasons for the deduction.

[0068] That is, record the deduction results in detail and mark the reasons for the deduction so that students can clearly understand the deduction situation.

[0069] S34, summarize the deduction results and generate a final score and output it visually.

[0070] That is, the final score is automatically summarized according to the deduction results, and the score results are output visually, so that managers can easily obtain the score results and analyze them.

[0071] That is, based on the judgment results and the built-in scoring mechanism, through the five links of deduction judgment, execution, verification, recording and summary, it realizes the functions of accurately triggering deductions, strictly verifying results, transparently recording reasons for deductions and efficiently summarizing scores. It has the technical advantages of high accuracy, traceability and intelligence, and through automatic matching of deduction standards and intelligent score calculation, it realizes the precision, efficiency and standardization of the deduction process, which significantly improves the fairness, efficiency and reliability of scoring.

[0072] In addition, in this embodiment, step S2 also includes: Detect unsafe behaviors in real time and trigger real-time alarms when unsafe behaviors are detected.

[0073] That is, during the trainees' electricity meter wiring operation, unsafe behaviors are monitored in real time. Once potential safety hazards or illegal operations are identified, alarms are triggered to improve safety. The alarms can be issued through sound, light or other visual prompts.

[0074] Please refer to Figure 2 , Embodiment 4 of the present invention is: An electric energy meter wiring referee system, such as Figure 2As shown, it includes a control module 1, an image acquisition module 2, an image recognition module 3, a scoring module 4 and an interaction module 5. The control module 1 stores a computer program. When the control module 1 executes the computer program, it implements the steps in an electric energy meter wiring judgment method in any one of the above-mentioned embodiments 1 to 3.

[0075] Among them Figure 2 As shown, in this embodiment, the control module 1 is electrically connected to the image acquisition module 2, the image recognition module 3, the scoring module 4 and the interaction module 5, wherein the control module 1 is used to control the data interaction between the modules; the image acquisition module 2 is used to acquire the image of the wiring process of the electric energy meter of each trainee and send it to the image recognition module 3; the image recognition module 3 is used to identify key features according to the image of the wiring process of the electric energy meter and send it to the scoring module 4; the scoring module 4 is used to score the wiring process of each trainee according to the key features combined with the built-in scoring mechanism, and feed back the scoring results to the control module 1; the interaction module 5 is used to perform human-computer interaction according to the scoring results sent by the control module 1.

[0076] That is, in this embodiment, by integrating multiple modules such as image acquisition, image recognition, scoring and human-computer interaction, automated judging of the wiring process of the electric energy meter is achieved, wherein the image recognition module 3 can accurately identify the key features in the wiring process of the electric energy meter, greatly improving the accuracy and efficiency of the judging, and the scoring module 4 performs refined scoring of the wiring process of each student based on the key features combined with the built-in scoring mechanism to ensure the fairness and accuracy of the scoring, and finally the control module 1 outputs the scoring results to the interactive module 5, so that the students can easily view their own wiring process and scoring results, and the coach can also provide guidance and feedback to the students through the interactive module 5, which promotes the improvement of students' skills and the improvement of teaching effects.

[0077] At the same time, Figure 2 As shown, in this embodiment, an alarm module is also included, wherein the alarm module is electrically connected to the control module 1 and is used to identify unsafe behaviors during the wiring process and issue an alarm.

[0078] That is, the alarm module can monitor unsafe behaviors during the wiring process of the electricity meter in real time. Once potential safety hazards or illegal operations are identified, the alarm mechanism is triggered and an alarm is issued through sound, light or other visual prompts.

[0079] In addition, in this embodiment, the control module 1 includes a microprocessor or an embedded system, which provides a stable and reliable control core for the system with its high performance, low power consumption and high integration characteristics; the interactive module 5 includes a touch screen, which significantly improves the interactive experience and convenience between the user and the system through an intuitive and easy-to-use touch operation interface.

[0080] Please refer to Figures 2 to 6 , Embodiment 5 of the present invention is: An electric energy meter wiring referee system, based on the above-mentioned fourth embodiment, in this embodiment, as Figure 2 As shown, the image recognition module 3 includes a face recognition unit 31 , a protective equipment recognition unit 32 , a behavior recognition unit 33 and a line recognition unit 34 .

[0081] The face recognition unit 31 is used to identify the trainee's identity information; the protective equipment recognition unit 32 is used to identify the trainee's wearing of protective equipment; the behavior recognition unit 33 is used to identify the trainee's behavior during the wiring process of the electric energy meter; and the line recognition unit 34 is used to identify the electric energy meter line.

[0082] That is, in this embodiment, through image acquisition and image recognition technology, the system can automatically and quickly capture and analyze the key features of the trainees in the process of wiring the electric energy meter, such as identity information, wearing of protective equipment, behavioral actions, and line connection conditions, etc., which improves the scoring efficiency and reduces the subjectivity and errors that may be caused by manual scoring, thereby ensuring the accuracy and fairness of the scoring; and through the protective equipment identification unit 32, it can detect in real time whether the trainees are wearing necessary protective equipment, which helps to ensure that the trainees comply with safety regulations during the wiring process and reduce the risk of accidents; at the same time, the behavior identification unit 33 and the line identification unit 34 can identify whether the trainees' wiring actions comply with standard operating procedures, further improving the standardization of the operation.

[0083] At the same time Figure 3 As shown, in this embodiment, the face recognition unit 31 includes: The face acquisition subunit 311 is used to detect the image acquired by the shooting device to obtain a face feature image; the face image preprocessing subunit 312 is electrically connected to the face acquisition subunit 311, and is used to receive the face feature image sent by the face acquisition subunit 311 and preprocess it; the face image feature extraction subunit 313 is electrically connected to the face image preprocessing subunit 312, and is used to extract face features based on the preprocessed face feature image; the face feature matching and recognition subunit is electrically connected to the face image feature extraction subunit 313, and is used to determine the student identity information based on the extracted face features.

[0084] Specifically in this embodiment, the face model is used to detect and locate the face in the input image, and the detected face is aligned and corrected; and the face key point detection model is used to extract features from the corrected face image, and the similarity is compared with the face key point feature information in the database, so as to achieve face recognition. The face images of different students are collected by the acquisition device, and the collected face images include static images and dynamic images. When the student is within the shooting range of the acquisition device, the acquisition device automatically searches and shoots the student's face image; then, the face position and size are detected and calibrated to extract face features, which include histogram features, color features, template features, structural features, Haar features, etc. In the face detection process, this embodiment uses the Adaboost algorithm to select rectangular features (weak classifiers) that can represent faces, and constructs the obtained weak classifiers into strong classifiers in a weighted voting manner, and then connects several trained strong classifiers in series to form a cascaded stacked classifier, which can effectively improve the detection speed of the classifier; and the preprocessing by the face image preprocessing subunit 312 mainly includes light compensation, grayscale transformation, histogram equalization, normalization, geometric correction, filtering and sharpening of the face image; and the face features extracted by the face image feature extraction subunit 313 mainly include visual features, pixel statistical features, face image transformation coefficient features and face image algebraic features; face recognition is to compare the face features to be identified with the obtained face feature templates, and judge the identity information of the face according to the similarity. The face image matching and recognition subunit 314 searches and matches the extracted face features with the feature templates stored in the database. When the similarity between the matched features and the face features exceeds the preset threshold, the matching results are output.

[0085] That is, the face recognition unit 31 achieves high-precision, fast response, stable and reliable face recognition effects through the coordinated work of various sub-units and the optimization of the four stages of image acquisition, preprocessing, feature extraction and matching recognition.

[0086] At the same time Figure 4 As shown, in this embodiment, the protective equipment identification unit 32 includes: The labeling subunit 321 is used to label the wearing status of protective equipment through the built-in image labeling tool; the model training subunit 322 is electrically connected to the labeling subunit 321, and is used to train the target detection model according to the labeling results of the wearing status of protective equipment; the detection subunit 323 is electrically connected to the model training subunit 322, and is used to detect the wearing status of protective equipment.

[0087] Among them, the wearing conditions of protective equipment include the wearing conditions of safety helmets, the wearing conditions of insulating gloves, the wearing conditions of insulating shoes (boots), and the wearing conditions of work clothes. Safety helmet wearing recognition can realize the detection of trainees wearing safety helmets and not wearing safety helmets. Among them, the colors of safety helmets include red, yellow, blue and white. In this embodiment, the image annotation tool can be used to annotate the targets wearing or not wearing safety helmets of different colors, and the corresponding target detection model can be called through the deep learning architecture to classify and train the annotated targets, so as to realize the detection of safety helmet wearing by personnel on site; at the same time, insulating glove wearing recognition can realize the detection of wearing insulating gloves and not wearing insulating gloves. In this embodiment, the hand parts of people wearing or not wearing insulating gloves can be annotated through the image annotation tool, and the corresponding target detection model can be called through the deep learning architecture to classify and train the annotated targets, so as to realize the detection of insulating gloves wearing by personnel on site. Insulating gloves include long-tube gel and white cotton; at the same time, insulating shoe (boot) wearing recognition can It is possible to detect whether insulating shoes (boots) are worn or not. In this embodiment, the foot parts of people wearing or not wearing insulating shoes (boots) can be annotated by image annotation tools, and the corresponding target detection model is called through the deep learning architecture to classify and train the annotated targets, so as to detect whether the insulating shoes (boots) are worn by the personnel on the site, for example, outdoor insulating shoes, military green insulating boots, white casual insulating shoes and not wearing insulating shoes, etc.; at the same time, the detection of wearing work clothes and not wearing work clothes is realized through the work clothes wearing recognition. In this embodiment, the body parts wearing or not wearing work clothes can be annotated by image annotation tools, and the corresponding target detection model is called through the deep learning architecture to classify and train the annotated targets, so as to detect whether the work clothes are worn by the personnel on the site.

[0088] That is, in this embodiment, by automatically identifying the wearing conditions of safety helmets, insulating gloves, insulating shoes (boots) and work clothes, the trainee's safety protection status can be monitored in real time to improve the accuracy of the scoring of the electric energy meter wiring process, and can improve the trainee's safety awareness and reduce safety accidents caused by not wearing or wearing improper protective equipment. Compared with the traditional manual inspection method, this embodiment uses deep learning technology for automatic identification, which can greatly improve the detection efficiency and accuracy and reduce human misjudgment and missed judgments; at the same time, this embodiment supports the identification of protective equipment of various colors and types, such as safety helmets of different colors, insulating gloves and insulating shoes (boots) of different materials, etc., and has good versatility. When new protective equipment needs to be detected, the recognition model can be easily updated and expanded. By combining the recognition results with the database, intelligent personnel management and safety protection decision support can be achieved.

[0089] That is, the protective equipment identification unit 32 can monitor the trainee's safety protection status in real time to improve the accuracy of the electric energy meter wiring process score, and can improve the trainee's safety awareness and reduce safety accidents caused by not wearing or wearing improper protective equipment.

[0090] At the same time Figure 5 As shown, in this embodiment, the behavior recognition unit 33 includes: The danger zone entry identification subunit 331 is electrically connected to the control module 1, and is used to detect whether a person enters the danger zone; the person fall identification subunit 332 is electrically connected to the control module 1, and is used to identify the posture of a person falling and determine whether a person has fallen; the work area smoking identification subunit 333 is electrically connected to the control module 1, and is used to identify smoking behavior in the work area.

[0091] That is, in this embodiment, the dangerous area can be monitored through the dangerous area entry identification subunit 331, and the area in the monitoring image is used for human intrusion detection, and multi-area detection is allowed. When a person enters the dangerous area, the area in the area turns red for early warning, a screenshot is saved, and an event is recorded. In this embodiment, the frame difference method or background method can be used to find out whether there is an area with a large grayscale change in the current frame to determine whether there is target movement. When target movement is detected, the human body detection model is called to identify the type of moving target, and the human body positioning frame is used to determine whether a person enters the designated dangerous area, and then an alarm is issued; at the same time, the person fall identification subunit 332 is used to detect the person fall in the area in the monitoring image, and multi-area detection is allowed. When a person falls, the area in the area turns red for early warning, a screenshot is saved, and an event is recorded. In this embodiment, various person falls can be identified through image annotation tools. The posture when a person smokes is annotated, and the corresponding state detection model is called through the deep learning architecture to train such annotated targets, thereby realizing the recognition and positioning of the posture of the person falling to the ground in the real scene; at the same time, the working area smoking recognition subunit 333 monitors the trainee's working area, and uses the smoking detection algorithm to recognize the trainee's smoking behavior. When smoking behavior occurs, a picture is captured and saved, and an event record is made. In this embodiment, the image annotation tool can be used to annotate the hand holding cigarette or the mouth smoking part when a person smokes, and the corresponding target detection model is called through the deep learning architecture to perform category training on such annotated targets. When the human body detection model detects the human body, the smoking part of the person is detected, thereby realizing the detection of smoking behavior of people in the real scene.

[0092] That is, in this embodiment, through the dangerous area entry identification subunit 331, it is possible to monitor the intrusion of personnel in the dangerous area in real time, timely warn and take measures to effectively prevent the occurrence of safety accidents; through the personnel fall identification subunit 332, it is possible to quickly identify and warn the personnel fall, provide valuable time for emergency rescue, and reduce the degree of injury; through the work area smoking identification subunit 333, it is possible to timely discover and stop smoking in the work area, and avoid fire and other safety hazards caused by illegal operations; at the same time, the deep learning model and smoking detection algorithm are used to automatically and intelligently identify various behaviors, greatly improving the monitoring efficiency and accuracy. In addition, this embodiment allows multi-area detection, and can monitor multiple areas at the same time, realizing all-round and no-dead-angle monitoring.

[0093] That is, the behavior recognition unit 33 realizes real-time monitoring and rapid warning of potential dangerous behaviors through detection of entry into dangerous areas, identification of people falling down and identification of smoking behavior, effectively improving safety management efficiency, reducing safety risks, and ensuring the safety of personnel and the working environment.

[0094] At the same time Figure 6 As shown, in this embodiment, the line identification unit 34 includes: The wire access sequence identification subunit 341 is electrically connected to the control module 1, and is used to detect the positive sequence access of wires of different colors; the wiring process judgment subunit 342 is electrically connected to the control module 1, and is used to judge the horizontality and verticality of the wiring process; the wire tightening identification subunit 343 is electrically connected to the control module 1, and is used to identify the tightening state of the wire; the bundling spacing identification subunit 344 is electrically connected to the control module 1, and is used to judge whether the bundling spacing is compliant; the wiring number mark identification subunit is electrically connected to the control module 1, and is used to verify whether the wiring number line mark is correct. If there is a part of the numbering position that is backward or blocked, it is judged that the numbering of this part of the wiring is not standardized and no points are scored; the wire wiring screw pressure point identification subunit 346 is electrically connected to the control module 1, and is used to identify the tightening degree of the wiring screw. The tightening degree of the wiring screw is sensed by the annular force sensor fixed under the screw, and the normal pressure value of the wire crimping is identified in combination with the deep learning model, and the unqualified crimping mode is judged and recorded.

[0095] Among them, the wire access sequence identification subunit 341 detects the yellow wire, green wire, red wire and black wire according to the color difference in the three channels separated from the color image through an artificial intelligence visual algorithm. According to the requirements of the document "Electric Energy Metering Equipment Installation Wiring Scoring Standards": the wires are connected in positive order, the voltage and current U phase of the electric energy meter is connected to the yellow wire, the V phase is connected to the green wire, and the W phase is connected to the red wire; the wiring process judgment subunit 342 is also based on the requirements of the document "Electric Energy Metering Equipment Installation Wiring Scoring Standards". The system uses an artificial intelligence visual algorithm to judge the wiring process. In this embodiment, the wiring position can be found through template matching positioning, and the wiring is edge detected and then straight line detection and angle fitting are performed, and then the horizontal and vertical wires are judged. Horizontal and vertical detection, wherein in this embodiment, 90 degrees and 180 degrees of wire bending are defined as standards, and the wire bending detection angles are 80~100 degrees and 170~190 degrees. Angles within this range are judged as qualified, and other angles are judged as unqualified by deducting points; the wire tightening identification subunit 343 is also based on the requirements of the "Electric Energy Metering Equipment Installation and Wiring Scoring Standards" document. The system uses an artificial intelligence visual algorithm to identify whether the wires are tightened or not. In this embodiment, the tightening state of the wires can be judged by the height difference between the horizontal wiring gap and the horizontal line; the bundling spacing identification subunit 344 is also based on the requirements of the "Electric Energy Metering Equipment Installation and Wiring Scoring Standards" document. The system uses an artificial intelligence visual algorithm to identify the bundling spacing. The output of the algorithm is the number of pixels between two adjacent cable ties. In this embodiment, the position of the black or white strapping when it is tightened can be annotated through an image annotation tool, and the corresponding target detection model is called through a deep learning architecture to train such annotated targets. The shooting distance is fixed, and the device will be calibrated before image recognition. The measured spacing is converted and finally becomes the actual spatial distance, thereby realizing the positioning and identification of the tightening point of the strapping at the horizontal connection point, and calculating the spacing between adjacent strappings of the same type arranged from left to right, and judging whether the bundling spacing is compliant based on the set upper limit of the strapping spacing threshold; the wiring number line mark recognition subunit 345 recognizes the wiring number line mark through an artificial intelligence visual algorithm.The wiring number line mark has the same height, the character direction is consistent, and the character is fully visible. In this embodiment, the wiring number line mark can be identified by the trained OCR character recognition model through the deep learning architecture, and compared with the wiring template to determine whether the number is correct. In this embodiment, the Chinese wiring number line mark recognition input is a captured wiring number line mark image 260*720p, and the minimum single character is 25*30p; the wire wiring screw tightening recognition subunit also uses an artificial intelligence algorithm to transmit and identify the wire wiring screw pressure point. In this embodiment, the sensor can determine whether the wire has been reliably crimped and whether it is crimped to the wire insulation skin by sensing the degree of tightening of the wiring screw. There is no signal transmission. The trained data is used through the deep learning architecture to identify the normal pressure value of the wire crimping, thereby determining the unqualified crimping mode and recording and deducting the score.

[0096] That is, in this embodiment, through artificial intelligence visual algorithms and deep learning models, it is possible to accurately identify the access sequence of wires of different colors, the horizontality and verticality of the wiring process, the tightening state of the wires, the compliance of the bundling spacing, the correctness of the wiring number line mark, and the crimping state of the wire wiring screws, which jointly ensure the accuracy and standardization of the electric energy meter wiring evaluation; it integrates multiple functional sub-units, which can fully cover all aspects of the wiring work, from wire access to screw crimping, and realize all-round monitoring of the trainee's electric energy meter wiring; at the same time, through intelligent management, it can automatically record, analyze and report the quality of the wiring work, and improve the efficiency and accuracy of the electric energy meter wiring scoring. Through automated identification and detection, it can significantly reduce manual intervention and improve the efficiency of the electric energy meter wiring scoring.

[0097] That is, the line identification unit 34 can comprehensively identify and detect the wiring sequence, process level, wire tightening status, bundling spacing, number identification and screw tightening status, and has the technical advantages of high precision, high efficiency and high automation, which significantly improves the standardization and safety of line wiring and reduces the risk of human operation errors.

[0098] Please refer to Figure 2 , Embodiment 6 of the present invention is: An electric energy meter wiring referee system, based on the above-mentioned fourth or fifth embodiment, in this embodiment, as Figure 2 As shown, the scoring module 4 includes: The deduction judgment unit 41 is used to judge whether to trigger the deduction mechanism based on the key features; the deduction execution unit 42 is electrically connected to the deduction judgment unit 41 and is used to execute the deduction mechanism; the verification unit 43 is electrically connected to the deduction execution unit 42 and is used to verify the deduction result and send the final deduction result to the recording unit 44; the recording unit 44 is electrically connected to the verification unit 43 and is used to record the deduction result and mark the reason for the deduction; the scoring summary unit 45 is electrically connected to the recording unit 44 and is used to summarize the final score according to the deduction result and output it to the interaction module 5 through the control module 1.

[0099] That is, in this embodiment, the scoring module 4 realizes the functions of accurately triggering deductions, strictly verifying results, transparently recording reasons for deductions, and efficiently summarizing scores through the five links of deduction judgment, execution, verification, recording, and summary. It has the technical advantages of high accuracy, traceability, and intelligence, and effectively improves the fairness and reliability of the automated scoring system.

[0100] The deduction execution unit 42 includes: The deduction standard matching unit is used to match the corresponding deduction standard; the deduction calculation unit is used to calculate the deduction score according to the matched deduction standard. That is, the deduction execution unit 42 realizes the precision, efficiency and standardization of the deduction process through automatic matching of deduction standards and intelligent score calculation, which significantly improves the fairness, efficiency and reliability of the scoring system.

[0101] In this embodiment, the deduction judgment unit 41 can make accurate judgments based on key features to ensure that the deduction mechanism is triggered accurately, which helps to avoid unnecessary deductions and maintain the fairness and accuracy of the scoring; the deduction execution unit 42 can quickly execute the deduction mechanism and accurately calculate the deduction results according to the deduction standards; at the same time, the verification unit 43 strictly checks the deduction results to ensure the correctness of the deductions, thereby improving the efficiency and accuracy of the scoring; the recording unit 44 records the deduction results in detail and marks the reasons for the deductions, so that the trainees can clearly understand the deductions; the scoring summary unit 45 can automatically summarize the final score based on the deduction results, and output the scoring results to the interactive module 5, so that the management personnel can easily obtain the scoring results.

[0102] In summary, the present invention provides a method and system for judging the connection of an electric energy meter. By integrating multiple modules such as image acquisition, image recognition, scoring and human-computer interaction, the present application realizes the automated judgment of the connection process of the electric energy meter. The image recognition module can accurately identify the key features of the trainee's identity information, the wearing of protective equipment, behavioral actions, and the electric energy meter line, greatly improving the accuracy and efficiency of the referee. The built-in protective equipment recognition unit and behavior recognition unit can monitor the trainee's safety protection measures and dangerous behaviors in real time. For example, by detecting whether there are people entering the dangerous area, whether the people fall to the ground, and whether there is smoking in the work area, etc., the occurrence of safety accidents is effectively prevented and the personal safety of the trainees is guaranteed. The scoring module performs refined scoring on the connection process of each trainee based on the key features combined with the built-in scoring mechanism. The deduction judgment unit and the execution unit can accurately judge and execute the deduction mechanism, and the verification unit strictly checks the deduction results to ensure the fairness and accuracy of the scoring. Finally, the scoring summary unit outputs the summarized scoring results to the interaction module, which is convenient for the trainees and coaches to view and analyze. The interaction module provides a friendly human-computer interaction interface, so that the trainees can easily view their own connection process and scoring results. At the same time, coaches can also provide guidance and feedback to trainees through interactive modules, which promotes the improvement of trainees' skills and teaching effectiveness.

[0103] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent transformations made using the contents of the present invention's specification and drawings, or directly or indirectly applied in related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for judging the connection of an electric energy meter, characterized in that: Includes steps: S1. Collect images of students’ electric energy meter wiring process; S2. Based on the wiring process image of the electric energy meter: S21, performing color feature separation to determine whether the wires of each color are connected in a positive order; S22, judging whether the bending angle of the wire is in compliance with the regulations through template matching and edge detection; S23, detecting the height difference between the wiring gap and the horizontal line through an image vision algorithm to determine the tightening state of the wires; S24, locate the cable tie position, calculate the spacing between adjacent cable ties, and determine whether the bundling spacing is compliant; S25, using an OCR character recognition model to recognize character information of the wiring number line label, and determining whether the number is correct; S26. Obtain the tightening pressure value of the wiring screw through the annular force sensor, and determine whether the crimping state is qualified by combining the deep learning algorithm.

2. The method for judging the connection of electric energy meters according to claim 1, characterized in that: In step S21, color feature separation is performed to determine whether the wires of each color are connected in a positive order, specifically: The image of the wiring process of the electric energy meter is separated into three channels of color images through an artificial intelligence visual algorithm, and yellow, green, red and black wires are obtained according to color differences. It is then determined whether the wires of each color are connected in the correct sequence based on the electric energy metering instrument installation wiring scoring standard.

3. The method for judging the connection of electric energy meters according to claim 1, characterized in that: The step S22 comprises the steps of: S221, extracting the edge straight line of the wire in the wiring process image of the electric energy meter by an edge detection method; S222, performing fitting analysis on the bending angle of the wire; S223: If the bending angle is within the preset compliance range, it is judged as qualified, otherwise it is marked as unqualified.

4. The method for judging the connection of electric energy meters according to claim 1, characterized in that: The step S24 comprises the steps of: S241, performing image annotation and target detection on the cable tie positions identified in the wiring process image of the electric energy meter; S242, measuring the pixel distance between adjacent cable ties, and converting the actual spatial distance based on the equipment calibration data; S243: Compare the actual spatial distance with a preset upper limit of the bundling spacing threshold to determine compliance of the bundling spacing.

5. The method for judging the connection of electric energy meters according to claim 1, characterized in that: The step S25 is specifically as follows: S251, extracting the wiring number characters in the wiring process image of the electric energy meter through an OCR character recognition model; S252, verifying the height consistency, orientation consistency and completeness of the wiring number characters, and comparing them with the wiring template. If there is an error, back-to-back or occlusion, it is judged as non-standard.

6. The method for judging the connection of electric energy meters according to claim 1, characterized in that: The step S26 is specifically as follows: S261, obtaining the tightening pressure data of the connection screws by fixing the annular force sensor under the screws by identifying the connection process image of the electric energy meter; S262. Call the deep learning model to determine whether the pressure value of the tightening pressure data of the wiring screw is within the standard range. If the pressure value is not within the standard range or touches the insulation layer of the wire, it is marked as unqualified.

7. The method for judging the connection of electric energy meters according to claim 1, characterized in that: After step S1, the method further includes: Based on the electric energy meter wiring process image, trainee identity information recognition, trainee protective equipment wearing status recognition, and trainee behavior recognition during the electric energy meter wiring process; The student identity information recognition includes face collection, face image preprocessing, face image feature extraction and face image feature matching; The identification of the wearing status of the trainees' protective equipment includes labeling the wearing status of the protective equipment, training a target detection model according to the labeling results of the wearing status of the protective equipment, and detecting the wearing status of the protective equipment; The trainee's behavioral action recognition during the process of wiring the electric energy meter includes recognition of entry into a dangerous area, recognition of a person falling to the ground, and recognition of smoking in a work area.

8. The method for judging the connection of electric energy meters according to claim 1, characterized in that: After step S2, the method further includes: S3. Based on the judgment result of step S2, the wiring process of each student is scored in combination with the built-in scoring mechanism, specifically: S31. Determine whether the connection process of each student triggers the deduction mechanism according to the judgment result and the built-in scoring mechanism, match the deduction standard, calculate the deduction score according to the deduction standard, and obtain the deduction result; S32, verifying the deduction result; S33, recording the deduction results and reasons for the deduction; S34, summarizing the deduction results and generating a final score and outputting it visually.

9. The method for judging the connection of electric energy meters according to claim 1, characterized in that: The step S2 also includes: Unsafe behaviors are detected in real time, and when the unsafe behaviors are detected, real-time alarms are triggered.

10. An electric energy meter wiring referee system, characterized in that: It includes a control module, an image acquisition module, an image recognition module, a scoring module and an interaction module. The control module stores a computer program. When the control module executes the computer program, the steps in the electric energy meter wiring judgment method as described in any one of claims 1 to 9 are implemented.

Images