A method and system for generating a secondary drawing safety ticket for relay protection

Abstract

The present invention relates to the technical field of power system maintenance, and discloses a method and system for generating safety measures tickets for secondary drawings of relay protection. The method comprises: collecting original relay protection secondary drawings in multiple source formats, and pre-processing the original relay protection secondary drawings to obtain target relay protection secondary drawings; extracting features from the target relay protection secondary drawings to obtain drawing feature tensors, and performing multi-task header parsing on the drawing feature tensors to obtain drawing structured parsing results; and performing safety measures rule reasoning on the drawing structured parsing results based on a pre-built relay protection safety measures rule library to obtain safety measures tickets. The present invention can realize the full process intelligence of secondary drawing processing and safety measures ticket generation, and fundamentally improve the efficiency and safety of relay protection maintenance operations.

Description

Technical Field

[0001] The present invention relates to the technical field of power system maintenance, and in particular to a method and system for generating a relay protection secondary drawing installation ticket. Background Art

[0002] During power system relay protection maintenance operations, the correct placement of secondary safety measures is crucial for ensuring grid security. Secondary safety measures tickets (SAMTs) are the core documents guiding their implementation, and their preparation relies on accurate analysis of substation secondary drawings. Traditional maintenance operations require maintenance personnel to bring paper drawings to site, manually verify terminal blocks, mark isolation terminals, and prepare SAMTs. This presents drawbacks such as inefficient paper drawing management, error-prone manual analysis, and time-consuming SAMT preparation.

[0003] With the digital transformation of power systems, intelligent drawing processing technology has gained widespread adoption. Currently, existing technologies primarily rely on general-purpose optical character recognition (OCR) to extract text from drawings or use single-target detection for equipment location. However, these technologies all have shortcomings. For example, general-purpose OCR lacks a dictionary optimized for specialized power terminology, resulting in low accuracy for densely packed terminal numbering and an inability to distinguish similar symbols. Single-target detection models, trained using general-purpose datasets, have poor recognition performance for terminal block matrices and multi-layered cable connections, often found on power drawings, resulting in a high rate of missed detections. Furthermore, existing technologies suffer from fragmented functionality, failing to correlate "terminal coordinates, text numbers, and cable connections," making it impossible to automatically construct device connection logic. Furthermore, they fail to generate compliant safety precautions based on industry standards. Furthermore, existing technologies are poorly adaptable to commonly encountered field conditions, such as blurred drawings, handwritten modifications, and multi-color overprints, and lack error checking mechanisms. This results in low safety precautions compilation efficiency and high error rates, posing a serious threat to power grid maintenance safety.

[0004] Therefore, there is an urgent need for an efficient security ticket generation method to overcome the shortcomings of the existing technology. Summary of the Invention

[0005] The present invention aims to overcome the shortcomings of existing safety ticket generation technologies in terms of accuracy, compliance, adaptability, and automation, realize the full intelligent process of secondary drawing processing and safety ticket generation, and fundamentally improve the efficiency and safety of relay protection maintenance operations. To achieve the above objectives, the present invention provides a method and system for generating safety ticket for secondary drawings of relay protection.

[0006] In a first aspect, an embodiment of the present invention provides a method for generating a secondary drawing safety ticket for relay protection, comprising:

[0007] Collecting original relay protection secondary drawings in multiple source formats, and preprocessing the original relay protection secondary drawings to obtain target relay protection secondary drawings;

[0008] Performing feature extraction on the target relay protection secondary drawing to obtain a drawing feature tensor, and performing multi-task header parsing on the drawing feature tensor to obtain a drawing structured parsing result, wherein the drawing structured parsing result includes a secondary element coordinate matrix, a secondary circuit text recognition result, and a secondary circuit connection relationship diagram;

[0009] Based on the pre-built relay protection safety measure rule library, the safety measure rule reasoning is performed on the structured analysis results of the drawing to obtain the safety measure ticket.

[0010] Preferably, after obtaining a safety measure ticket by performing safety measure rule reasoning on the structured analysis result of the drawing based on the pre-built relay protection safety measure rule library, the method further includes:

[0011] A consistency check is performed on the actual execution result of the security ticket to obtain a check result, and the drawing structured analysis result is optimized based on the check result.

[0012] Preferably, the collecting of original relay protection secondary drawings in multiple source formats and preprocessing the original relay protection secondary drawings to obtain target relay protection secondary drawings includes:

[0013] Collect original relay protection secondary drawings in multiple source formats, including paper scans, on-site copies, and electronic documents;

[0014] The original relay protection secondary drawing is preprocessed to obtain a target relay protection secondary drawing, wherein the preprocessing includes illumination correction, denoising, binarization and image normalization.

[0015] Preferably, the step of extracting features from the target relay protection secondary drawing to obtain a drawing feature tensor, and performing multi-task header parsing on the drawing feature tensor to obtain a drawing structured parsing result, includes:

[0016] Extracting features of the target relay protection secondary drawing based on the first residual network to obtain a drawing feature tensor;

[0017] Performing secondary component detection on the drawing feature tensor based on a pre-built target detection model to obtain a secondary component coordinate matrix;

[0018] Performing secondary loop recognition on the drawing feature tensor based on a pre-built text recognition model to obtain a secondary loop text recognition result;

[0019] Based on the graph neural network, the secondary element coordinate matrix and the secondary circuit text recognition result are associated with reasoning to obtain a secondary circuit connection relationship diagram.

[0020] Preferably, the method of performing secondary component detection on the drawing feature tensor based on the pre-built target detection model to obtain a secondary component coordinate matrix includes:

[0021] Performing structural optimization on the original YOLOv8 model to obtain an improved YOLOv8 model, wherein the structural optimization includes embedding channel attention and spatial attention in the backbone network of the original YOLOv8 model;

[0022] The improved YOLOv8 model is supervised and trained based on a pre-built power secondary circuit drawing dataset to obtain a target detection model;

[0023] The drawing feature tensor is input into the target detection model to perform secondary component detection to obtain a secondary component coordinate matrix.

[0024] Preferably, the performing of secondary loop recognition on the drawing feature tensor based on the pre-built text recognition model to obtain a secondary loop text recognition result includes:

[0025] A hybrid OCR model is constructed based on the second residual network, convolutional recurrent neural network and Transformer network;

[0026] Supervised training of the hybrid OCR model is performed based on a pre-built electric power drawing text sample set to obtain a text recognition model;

[0027] The drawing feature tensor is input into the text recognition model to perform secondary loop recognition to obtain a secondary loop text recognition result.

[0028] Preferably, the step of inputting the drawing feature tensor into the text recognition model to perform secondary circuit recognition and obtain a secondary circuit text recognition result includes:

[0029] Performing feature extraction on the drawing feature tensor based on the second residual network to obtain a feature map sequence;

[0030] Performing sequence modeling on the feature graph sequence based on the convolutional recurrent neural network to obtain a hidden state sequence;

[0031] Based on the Transformer network, attention decoding is performed on the hidden state sequence and the electric power terminology dictionary to obtain a secondary circuit text recognition result.

[0032] Preferably, the step of performing safety measure rule reasoning on the structured analysis result of the drawing based on the pre-built relay protection safety measure rule library to obtain a safety measure ticket includes:

[0033] Conduct structured modeling of power industry safety standard documents to obtain a relay protection safety rule base;

[0034] Based on the relay protection safety measure rule base, scenario recognition is performed on the structured analysis results of the drawing to determine the current maintenance scenario;

[0035] Based on the relay protection safety measure rule base, rule matching is performed on the current maintenance scenario to obtain a safety measure ticket.

[0036] Preferably, performing consistency verification on the actual execution result of the security ticket to obtain a verification result, and optimizing the drawing structured analysis result based on the verification result, includes:

[0037] Comparing the theoretical execution result and the actual execution result of the security ticket to obtain a verification result;

[0038] The drawing structured parsing result is optimized by reverse iteration based on the verification result.

[0039] In a second aspect, an embodiment of the present invention provides a system for generating a secondary drawing safety measure ticket for relay protection, including:

[0040] A data acquisition and preprocessing module is used to acquire original relay protection secondary drawings in multiple source formats and preprocess the original relay protection secondary drawings to obtain target relay protection secondary drawings;

[0041] an intelligent parsing module, configured to extract features from the target relay protection secondary drawing to obtain a drawing feature tensor, and perform multi-task header parsing on the drawing feature tensor to obtain a drawing structured parsing result, wherein the drawing structured parsing result includes a secondary element coordinate matrix, a secondary circuit text recognition result, and a secondary circuit connection relationship diagram;

[0042] The safety measure ticket generation module is used to perform safety measure rule reasoning on the structured analysis results of the drawing based on a pre-built relay protection safety measure rule library to obtain a safety measure ticket.

[0043] Compared with the prior art, the method and system for generating safety measures tickets for secondary drawings of relay protection in the embodiment of the present invention have the following beneficial effects: through multi-source drawing preprocessing, feature tensor extraction and multi-task header parsing, precise positioning of secondary component coordinates, high-accuracy recognition of secondary circuit text and topological modeling of connection relationships are achieved. Combined with the reasoning of the pre-built safety measures rule base, safety measures tickets that meet industry standards can be automatically generated, which significantly reduces the labor cost and risk of human error in the preparation of safety measures tickets. Especially in the processing of complex protection panel cabinet drawings, the logical verification and step generation of safety measures steps can be completed quickly, providing technical support for the standardized and intelligent execution of relay protection maintenance operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 This is a flow chart of a method for generating a secondary drawing installation ticket for relay protection according to an embodiment of the present invention;

[0045] Figure 2 It is a schematic diagram of a process for obtaining a structural analysis result of a drawing according to an embodiment of the present invention;

[0046] Figure 3 This is another flow chart of a method for generating a secondary drawing safety ticket for relay protection according to an embodiment of the present invention;

[0047] Figure 4 This is a structural diagram of a system for generating a secondary drawing safety measure ticket for relay protection according to an embodiment of the present invention;

[0048] Reference numerals:

[0049] 1. Data collection and preprocessing module; 2. Intelligent analysis module; 3. Security ticket generation module. DETAILED DESCRIPTION

[0050] The following embodiments of the present invention are described in further detail with reference to the accompanying drawings and examples. The following examples are used to illustrate the present invention but are not intended to limit the scope of the present invention.

[0051] In the description of the present invention, it should be understood that the terms "first" and "second" etc. are used in the present invention to distinguish different objects rather than to describe a specific order.

[0052] In describing the present invention, it should be noted that, unless otherwise defined, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art. The terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. Those skilled in the art will understand the specific meanings of the above terms in the present invention in specific circumstances.

[0053] like Figure 1 As shown in FIG, it is a flow chart of a method for generating a secondary drawing safety ticket for relay protection according to an embodiment of the present invention. Figure 1 The embodiment of the present invention provides a method for generating a secondary drawing installation ticket for relay protection, comprising the steps of:

[0054] S1. Collecting original relay protection secondary drawings in multiple source formats and preprocessing the original relay protection secondary drawings to obtain target relay protection secondary drawings;

[0055] Specifically, step S1 includes:

[0056] 1) Collect original relay protection secondary drawings in multiple source formats;

[0057] Multiple source formats include paper scans, on-site photocopies, and electronic documents. In other words, the drawing collection method supports paper scans, on-site photocopies, and electronic document import.

[0058] 2) Preprocess the original relay protection secondary drawing to obtain the target relay protection secondary drawing.

[0059] To address quality issues such as blurring, wrinkling, and multi-color overprinting in the original relay protection secondary drawings, this step uses a preprocessing process called "adaptive enhancement + noise suppression." Specifically, the preprocessing includes illumination correction, denoising, binarization, and image normalization.

[0060] The following is a step-by-step description of the preprocessing process:

[0061] 21) Illumination correction: Considering the characteristics of the original relay protection secondary drawings, which are mainly black and white with color as a supplement, an improved Retinex algorithm is used to perform illumination correction to solve the problem of uneven illumination.

[0062] Specifically, the following formula is used to characterize the improved Retinex algorithm:

[0063]

[0064] in, represents the enhanced image, represents the input image, represents the logarithmic transformation coefficient, Represents the guided filter weight. It should be noted that this formula uses logarithmic transformation Extraction of simulated reflection components using guided filtering Simulate the extraction of illumination components, and then and Balancing the two allows the image to retain details while correcting uneven lighting, thereby achieving an enhanced effect.

[0065] 22) Denoising: Median filtering is used to suppress random noise, such as handwritten corrections such as pencil markings, present in the original relay protection secondary drawings. Specifically, this embodiment uses a 3×3 window size for median filtering, which effectively covers the noise range of pencil marks while preserving the text structure.

[0066] 23) Binarization: To address the multi-color overprints present in the original relay protection secondary drawings, such as those covered by red and blue pencils, we first extract the main color channel through color space conversion, separate the interference color layer, and then perform binarization to accurately extract the text and filter out the multi-color overprints.

[0067] 24) Image Normalization: In this embodiment, the output size of the target relay protection secondary drawings is unified to 2480×3508 pixels (standard resolution of A3 paper), the color space is grayscale, and the pixel values ​​are normalized to the range [0,1] to reduce the amount of calculation.

[0068] It should be noted that steps 21) to 24) are logically executed sequentially. That is, the output of each step becomes the input of the next step, and the steps are processed sequentially until the final target relay protection secondary drawing is output.

[0069] S2. Extract features from the target relay protection secondary drawing to obtain a drawing feature tensor, and perform multi-task header parsing on the drawing feature tensor to obtain a drawing structured parsing result;

[0070] The structural analysis results of the drawings include the secondary component coordinate matrix, the secondary circuit text recognition results and the secondary circuit connection diagram.

[0071] Specifically, if Figure 2 As shown, it is a flow chart of step S2. Figure 2 , step S2 includes:

[0072] S201, extracting features of a target relay protection secondary drawing based on a first residual network to obtain a drawing feature tensor;

[0073] In this example, ResNet-50 was selected as the first residual network. ResNet-50, with its 50-layer depth and balanced performance-computational complexity, has become a classic backbone network for image feature extraction. ResNet-50 was used to extract features from the target relay protection secondary drawings, generating a drawing feature tensor of H×W×C (where H represents height, W represents width, and C represents the number of channels).

[0074] S202, performing secondary component detection on the drawing feature tensor based on a pre-built target detection model to obtain a secondary component coordinate matrix;

[0075] Specifically, step S202 includes:

[0076] 1) Optimize the structure of the original YOLOv8 model to obtain an improved YOLOv8 model;

[0077] To address the problem that traditional target detection models have poor recognition performance for small targets and dense elements in power drawings, this step optimizes the structure of the original YOLOv8 model to obtain an improved YOLOv8 model.

[0078] Specifically, the structural optimization includes embedding channel attention and spatial attention in the backbone network of the original YOLOv8 model, that is, adding a small target attention module to the backbone network of the original YOLOv8 model, and enhancing the feature response to dense terminals (size ≤ 5×5 pixels) by fusing the weights of channel attention and spatial attention.

[0079] Furthermore, the channel attention weight and spatial attention weight are calculated using the following formulas:

[0080]

[0081]

[0082] in, represents the channel attention weight, represents the spatial attention weight, represents the Sigmoid activation function, and represents the weight matrix of the fully connected layer, represents global average pooling, Represents the feature map output by the backbone network, represents the feature concatenation operation, represents maximum pooling, represents average pooling.

[0083] 2) Supervised training of the improved YOLOv8 model based on a pre-built dataset of power secondary circuit drawings to obtain a target detection model;

[0084] Specifically, this example constructs a dataset of 50,000 power secondary circuit drawings based on the "Specifications for Relay Protection Secondary Circuit Drawings." Twelve annotation categories, including "terminal block," "cable outlet," "pressure plate," and "circuit breaker," are used. The annotation method is "bounding box + category + scene label." Data augmentation strategies include geometric augmentation and noise augmentation. Geometric augmentation simulates wrinkles in the drawings and changes in shooting angles to improve the model's adaptability to non-ideal images. Noise augmentation includes adding Gaussian noise, blurring, and overwriting handwritten terminal numbers to enhance model robustness.

[0085] Furthermore, this embodiment uses an improved YOLOv8 model to perform supervised training on a dataset of power secondary circuit drawings for 100 rounds, resulting in a target detection model. This model uses an AdamW optimizer with a learning rate of 1e-4. In one specific embodiment, compared to the traditional YOLOv8 model, the target detection model improves detection accuracy for terminal blocks and cable outlets by 7.7% and 8.5%, respectively, demonstrating its effectiveness in identifying small targets.

[0086] 3) Input the drawing feature tensor into the target detection model for secondary component detection to obtain the secondary component coordinate matrix.

[0087] The drawing feature tensor is fed into the object detection model, which detects secondary components, outputs bounding boxes and categories, and combines them with scene labels to construct a secondary component coordinate matrix. Secondary components detected by the object detection model include, but are not limited to, terminal blocks, cable outlets, and device symbols.

[0088] S203, performing secondary loop recognition on the drawing feature tensor based on a pre-built text recognition model to obtain a secondary loop text recognition result;

[0089] Specifically, step S203 includes:

[0090] 1) Build a hybrid OCR model based on the second residual network, convolutional recurrent neural network, and Transformer network;

[0091] This example uses ResNet-34 as the second residual network and the backbone of the hybrid OCR model. The convolutional recurrent neural network uses a bidirectional LSTM as the RNN layer, and the Transformer network includes a six-layer multi-head attention mechanism. A hybrid OCR model is constructed based on this combination of networks.

[0092] 2) Supervised training of the hybrid OCR model based on a pre-built set of text samples from power drawings to obtain a text recognition model;

[0093] Specifically, this embodiment constructs a sample set of electric power drawing text based on electric power terminology samples and small text enhancement samples. The electric power terminology samples include over 2,000 electric power terms, terminal numbers, and pressure plate markings, covering typical text types found on relay protection secondary drawings. The small text enhancement samples involve scaling the original samples (to 1 / 2 their original size) to simulate dense terminal numbers with a font height of ≤2mm. This scaling enhancement allows the model to learn the key features of small text, improving its ability to recognize text in densely packed areas of panels and cabinets.

[0094] Furthermore, the hybrid OCR model of this embodiment was trained on a sample set of text from electrical power drawings to produce a text recognition model. The loss function employed a weighted sum of a connectionist temporal classification loss and a cross-entropy loss, with a weight ratio of 7:3. This improves the recognition accuracy of long text. In one specific embodiment, compared to common OCR models (Tesseract and PaddleOCR), the text recognition model achieved an accuracy rate of 98.5% for terminal numbers and 99.2% for electrical power terminology, demonstrating the effectiveness of the text recognition model.

[0095] 3) Input the drawing feature tensor into the text recognition model for secondary circuit recognition to obtain the secondary circuit text recognition result.

[0096] Specifically, step 3) includes:

[0097] 31) Extract features from the drawing feature tensor based on the second residual network to obtain a feature map sequence;

[0098] ResNet-34 is used to extract features from the drawing feature tensor to obtain a feature map sequence.

[0099] 32) Based on the convolutional recurrent neural network, the feature map sequence is modeled to obtain the hidden state sequence;

[0100] A bidirectional LSTM based on a convolutional recurrent neural network models the feature graph sequence to obtain a hidden state sequence. The bidirectional LSTM learns temporal dependencies by processing the feature graph sequence and captures the contextual associations of the feature graph sequence.

[0101] 33) Based on the Transformer network, attention decoding is performed on the hidden state sequence and the power terminology dictionary to obtain the secondary circuit text recognition results.

[0102] The Transformer network uses a multi-head attention mechanism to combine the hidden state sequence with semantic information from the power terminology dictionary to generate secondary circuit text recognition results. These secondary circuit text recognition results include but are not limited to terminal numbers and device names.

[0103] S204. Based on the graph neural network, the secondary component coordinate matrix and the secondary circuit text recognition results are associated with reasoning to obtain a secondary circuit connection relationship diagram.

[0104] Specifically, step S204 includes:

[0105] 1) Node definition: Nodes include terminal nodes and device nodes. The attributes of each terminal node are defined, including coordinates, number, and type. The attributes of each device node are defined, including name and type.

[0106] 2) Edge Construction Rules: Edge construction rules include spatial proximity rules, text association rules, and domain rules. Spatial proximity rules include establishing an edge when the Euclidean distance between the starting or ending coordinates of a cable outlet and the terminal coordinates is ≤ 2 pixels. Text association rules include establishing an edge when the cable label text matches the line identifier in the terminal number. Domain rules include supplementing implicit connections based on the "Design Specifications for Relay Protection Secondary Circuits," for example, requiring that the tripping outlet of a protection device must be connected to the corresponding line terminal.

[0107] 3) Model Inference: A graph attention network is trained on a supervised dataset of manually labeled quadratic loop connections to produce a trained graph attention network. The cross-entropy loss is used as the loss function. In one specific embodiment, the trained graph attention network achieved a 96.7% accuracy rate in inferring quadratic loop connections, compared to traditional geometric matching algorithms, demonstrating the effectiveness of graph neural network inference.

[0108] S3. Based on the pre-built relay protection safety measure rule library, the safety measure rule reasoning is performed on the structured analysis results of the drawing to obtain the safety measure ticket.

[0109] Specifically, step S3 includes:

[0110] 1) Conduct structured modeling of power industry safety standard documents to obtain a relay protection safety measure rule base;

[0111] The power industry safety standard documents are stored in the form of "scenario-rule-measure" triples to obtain the relay protection safety measure rule base. The triple form is shown in Table 1.

[0112] Table 1 “Scenario-Rule-Measure” triplet form

[0113]

[0114] 2) Based on the relay protection safety measures rule base, the structural analysis results of the drawings are used to identify scenarios and determine the current maintenance scenario;

[0115] Determine the current maintenance scenario corresponding to the structured analysis results of the drawings based on the relay protection safety measures rule base.

[0116] 3) Based on the relay protection safety measure rule library, the current maintenance scenario is matched with rules to obtain a safety measure ticket.

[0117] Based on the current maintenance scenario, matching rules are retrieved from the relay protection safety measure rule library to determine the corresponding safety measures, thereby generating a safety measure ticket. In order to ensure that the safety measure ticket is consistent with the actual situation on site, it supports re-retrieval of matching rules, correction of corresponding safety measures, and thus updating of the safety measure ticket. The final generated safety measure ticket includes basic information, a list of safety measures, logic verification results, and an electronic signature area. Among them, basic information includes maintenance time, equipment name, and person in charge. The safety measures list includes disconnect terminals, isolation cables, and pressure plate insertion and withdrawal status. The logic verification results include that all trip outlet terminals have been disconnected and there are no duplicate terminal numbers. The electronic signature area supports handwritten signatures or digital signatures.

[0118] like Figure 3 As shown in FIG, it is a flow chart of a method for generating a secondary drawing safety ticket for relay protection according to an embodiment of the present invention. Figure 3, after step S3, further comprising the steps of:

[0119] S4. Perform consistency check on the actual execution results of the security ticket to obtain the verification results, and optimize the drawing structured analysis results based on the verification results.

[0120] Specifically, step S4 includes:

[0121] 1) Compare the theoretical execution results of the Ancuo ticket with the actual execution results to obtain the verification results;

[0122] If the comparison is consistent, the verification result is correct, otherwise the verification result is wrong. At the same time, when the verification result is wrong, an abnormal alarm is triggered so that maintenance personnel can find the problem in time.

[0123] 2) Reversely iterate and optimize the structural analysis results of the drawing based on the verification results.

[0124] The verification results (correct or incorrect) are added to the training set as new samples, and the structural analysis results of the drawings are reversely optimized to form an iterative closed loop of "execution-verification-optimization".

[0125] The embodiment of the present invention provides a method for generating an installation ticket for a relay protection secondary drawing. Through multi-source drawing preprocessing, feature tensor extraction and multi-task header parsing, the method achieves precise positioning of secondary component coordinates, high-accuracy recognition of secondary circuit text and topological modeling of connection relationships. Combined with the reasoning of a pre-built installation rule base, the method can automatically generate an installation ticket that meets industry standards, significantly reducing the labor cost and risk of human error in the preparation of the installation ticket. Especially in the processing of complex protection panel cabinet drawings, the method can quickly complete the logical verification and step generation of the installation steps, providing technical support for the standardized and intelligent execution of relay protection maintenance operations.

[0126] Based on the above-mentioned method for generating a secondary protection drawing installation ticket, Figure 4 As shown in FIG, it is a structural diagram of a system for generating a secondary drawing safety ticket for relay protection according to an embodiment of the present invention. Figure 4 The embodiment of the present invention provides a system for generating a secondary drawing safety ticket for relay protection, including:

[0127] The data acquisition and preprocessing module 1 is used to acquire original relay protection secondary drawings in multiple source formats and preprocess the original relay protection secondary drawings to obtain target relay protection secondary drawings;

[0128] Intelligent parsing module 2 is used to extract features from the target relay protection secondary drawing to obtain a drawing feature tensor, and perform multi-task header parsing on the drawing feature tensor to obtain a drawing structured parsing result;

[0129] The structural analysis results of the drawings include the secondary component coordinate matrix, secondary circuit text recognition results, and secondary circuit connection diagram;

[0130] The safety measure ticket generation module 3 is used to perform safety measure rule reasoning on the structured analysis results of the drawing based on the pre-built relay protection safety measure rule library to obtain the safety measure ticket.

[0131] It should be noted that each module in the above-mentioned relay protection secondary drawing safety ticket generation system can be fully or partially implemented by software, hardware and their combination. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules. For the specific definition of a relay protection secondary drawing safety ticket generation system, please refer to the definition of a relay protection secondary drawing safety ticket generation method above. The two have the same functions and effects and will not be repeated here.

[0132] To sum up, the embodiment of the present invention provides a method and system for generating safety precautions tickets for secondary drawings of relay protection. Through multi-source drawing preprocessing, feature tensor extraction and multi-task header parsing, it realizes the precise positioning of secondary component coordinates, high-accuracy recognition of secondary circuit text and topological modeling of connection relationships. Combined with the reasoning of the pre-built safety precautions rule base, it can automatically generate safety precautions tickets that meet industry standards, significantly reducing the labor cost and risk of human error in the preparation of safety precautions tickets. Especially in the processing of complex protection panel cabinet drawings, it can quickly complete the logical verification and step generation of safety precautions steps, providing technical support for the standardized and intelligent execution of relay protection maintenance operations.

[0133] Each embodiment in this specification is described in a progressive manner, and the same or similar parts of each embodiment can be directly referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. It should be noted that the various technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the various technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0134] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and substitutions can be made without departing from the technical principles of the present invention. These improvements and substitutions should also be regarded as the scope of protection of the present invention.

Claims

1. A method for generating a secondary drawing installation ticket for relay protection, characterized in that: include: Collecting original relay protection secondary drawings in multiple source formats, and preprocessing the original relay protection secondary drawings to obtain target relay protection secondary drawings; Performing feature extraction on the target relay protection secondary drawing to obtain a drawing feature tensor, and performing multi-task header parsing on the drawing feature tensor to obtain a drawing structured parsing result, wherein the drawing structured parsing result includes a secondary element coordinate matrix, a secondary circuit text recognition result, and a secondary circuit connection relationship diagram; Based on the pre-built relay protection safety measure rule library, the safety measure rule reasoning is performed on the structured analysis result of the drawing to obtain the safety measure ticket; The feature extraction of the target relay protection secondary drawing is performed to obtain a drawing feature tensor, and the multi-task header parsing of the drawing feature tensor is performed to obtain a drawing structured parsing result, including: Extracting features of the target relay protection secondary drawing based on the first residual network to obtain a drawing feature tensor; Performing secondary component detection on the drawing feature tensor based on a pre-built target detection model to obtain a secondary component coordinate matrix; Performing secondary loop recognition on the drawing feature tensor based on a pre-built text recognition model to obtain a secondary loop text recognition result; Performing associative reasoning on the secondary component coordinate matrix and the secondary circuit text recognition result based on a graph neural network to obtain a secondary circuit connection relationship diagram; The pre-built text recognition model is used to perform secondary loop recognition on the drawing feature tensor to obtain a secondary loop text recognition result, including: A hybrid OCR model is constructed based on the second residual network, convolutional recurrent neural network and Transformer network; Supervised training of the hybrid OCR model is performed based on a pre-built electric power drawing text sample set to obtain a text recognition model; Inputting the drawing feature tensor into the text recognition model to perform secondary circuit recognition, and obtaining a secondary circuit text recognition result, including: Performing feature extraction on the drawing feature tensor based on the second residual network to obtain a feature map sequence; Performing sequence modeling on the feature graph sequence based on the convolutional recurrent neural network to obtain a hidden state sequence; Based on the Transformer network, attention decoding is performed on the hidden state sequence and the electric power terminology dictionary to obtain a secondary circuit text recognition result.

2. The method for generating a secondary drawing installation ticket for relay protection according to claim 1, characterized in that: After the safety measure rule reasoning is performed on the structured analysis result of the drawing based on the pre-built relay protection safety measure rule library to obtain the safety measure ticket, the method further includes: A consistency check is performed on the actual execution result of the security ticket to obtain a check result, and the drawing structured analysis result is optimized based on the check result.

3. The method for generating a secondary drawing installation ticket for relay protection according to claim 1, characterized in that: The collecting of original relay protection secondary drawings in multiple source formats and preprocessing the original relay protection secondary drawings to obtain target relay protection secondary drawings include: Collect original relay protection secondary drawings in multiple source formats, including paper scans, on-site copies, and electronic documents; The original relay protection secondary drawing is preprocessed to obtain a target relay protection secondary drawing, wherein the preprocessing includes illumination correction, denoising, binarization and image normalization.

4. The method for generating a secondary drawing installation ticket for relay protection according to claim 1, characterized in that: The method of performing secondary component detection on the drawing feature tensor based on the pre-built target detection model to obtain a secondary component coordinate matrix includes: Performing structural optimization on the original YOLOv8 model to obtain an improved YOLOv8 model, wherein the structural optimization includes embedding channel attention and spatial attention in the backbone network of the original YOLOv8 model; The improved YOLOv8 model is supervised and trained based on a pre-built power secondary circuit drawing dataset to obtain a target detection model; The drawing feature tensor is input into the target detection model to perform secondary component detection to obtain a secondary component coordinate matrix.

5. The method for generating a secondary drawing installation ticket for relay protection according to claim 1, characterized in that: The method of performing safety measure rule reasoning on the structured analysis result of the drawing based on the pre-built relay protection safety measure rule library to obtain a safety measure ticket includes: Conduct structured modeling of power industry safety standard documents to obtain a relay protection safety rule base; Based on the relay protection safety measure rule base, scenario recognition is performed on the structured analysis results of the drawing to determine the current maintenance scenario; Based on the relay protection safety measure rule base, rule matching is performed on the current maintenance scenario to obtain a safety measure ticket.

6. The method for generating a secondary drawing installation ticket for relay protection according to claim 2, characterized in that: The consistency check of the actual execution result of the security ticket is performed to obtain a check result, and the optimization of the drawing structured analysis result based on the check result includes: Comparing the theoretical execution result and the actual execution result of the security ticket to obtain a verification result; The drawing structured parsing result is optimized by reverse iteration based on the verification result.

7. A system for generating secondary drawings and measures for relay protection, characterized in that: The method for generating a secondary drawing safety measure ticket for relay protection according to any one of claims 1 to 6 comprises: A data acquisition and preprocessing module is used to acquire original relay protection secondary drawings in multiple source formats and preprocess the original relay protection secondary drawings to obtain target relay protection secondary drawings; an intelligent parsing module, configured to extract features from the target relay protection secondary drawing to obtain a drawing feature tensor, and perform multi-task header parsing on the drawing feature tensor to obtain a drawing structured parsing result, wherein the drawing structured parsing result includes a secondary element coordinate matrix, a secondary circuit text recognition result, and a secondary circuit connection relationship diagram; The safety measure ticket generation module is used to perform safety measure rule reasoning on the structured analysis results of the drawing based on a pre-built relay protection safety measure rule library to obtain a safety measure ticket.

Images