Substation work order generation method, device, electronic equipment and program product

Abstract

Embodiments of the present application provide a power substation work order generation method and device, electronic equipment and program product, by acquiring multi-modal input data, the multi-modal input data at least includes the first input data with the first data format and the second input data with the second data format, the first input data is used to describe the task content of the target work task of the target power substation, and the second input data is used to describe the device space relationship and line connection relationship of the electrical equipment in the target power substation;The multi-modal input data is processed, and structured task data is generated, the structured task data is used to describe the operation process and corresponding safety measures of the target work task for the target power substation;Based on the structured task data, the work order corresponding to the target work task of the target power substation is generated, and the generation efficiency and accuracy of the work order are improved.

Description

Technical Field

[0001] This application relates to the field of intelligent operation and maintenance technology for power grids, and in particular to a method, device, electronic equipment and program product for generating substation work orders. Background Technology

[0002] In the operation of power systems, substations, as core hubs, undertake the critical functions of power transmission and distribution. Their equipment inspection and maintenance operations must strictly adhere to safety regulations. Among these, work orders (also known as work permits) serve as the legal basis for operational safety, requiring detailed records of equipment status, operating procedures, and safety measures. Workers must strictly follow these work orders to ensure the safety of personnel and equipment.

[0003] In traditional work order creation scenarios, operations and maintenance personnel need to manually fill out and generate work orders after consulting relevant materials. However, the existing work order creation solutions described above suffer from low efficiency and poor accuracy in work order generation. Summary of the Invention

[0004] This application provides a method, apparatus, electronic device, and program product for generating substation work orders, which aims to improve the efficiency and accuracy of work order generation.

[0005] In a first aspect, embodiments of this application provide a method for generating substation work orders, including:

[0006] Acquire multimodal input data, wherein the multimodal input data includes at least first input data having a first data format and second input data having a second data format, wherein the first input data is used to describe the task content of the target work task of the target substation, and the second input data is used to describe the equipment spatial relationship and line connection relationship of the electrical equipment in the target substation;

[0007] The multimodal input data is processed to generate structured task data, which is used to describe the work process and corresponding safety measures for the target work task of the target substation.

[0008] Based on the structured task data, a work order corresponding to the target work task of the target substation is generated.

[0009] In one possible implementation, processing the multimodal input data to generate structured task data includes:

[0010] By extracting the text content from the first input data, structured text is obtained;

[0011] The device layout information is obtained by extracting the symbol identifiers from the second input data;

[0012] Semantic fusion is performed on the structured text and the equipment layout information to generate a structured task description, which includes the task type of the target work task, the equipment status of the target electrical equipment, and the operation requirements.

[0013] Based on the structured task description, structured task data is generated.

[0014] In one possible implementation, obtaining device layout information by extracting symbol identifiers from the second input data includes:

[0015] Image recognition is performed on the second input data to obtain a first type of symbol and a second type of symbol. The first type of symbol indicates electrical equipment, and the second type of symbol indicates the connection relationship between the electrical equipment.

[0016] Based on the first type of symbols and the second type of symbols, the device layout information is obtained.

[0017] In one possible implementation, it also includes:

[0018] The second input data is subjected to image recognition to obtain a third type of symbol, which is used to indicate the operating requirements of electrical equipment;

[0019] The process of obtaining device layout information based on the first type of symbols and the second type of symbols includes:

[0020] Based on the first type of symbols, the second type of symbols, and the third type of symbols, the device layout information is obtained.

[0021] In one possible implementation, the semantic fusion of the structured text and the device layout information to generate a structured task description includes:

[0022] The semantics of the structured text and the equipment layout information are extracted and semantically fused to obtain a first description text. The first description text is used to describe the first task step of the target work task and the electrical equipment involved in the first task step.

[0023] Based on the first description text, the dynamic security rule engine is invoked to obtain the corresponding security measure description text, which is used to generate the operation requirements corresponding to the first task step;

[0024] The structured task description is generated based on the first description text and the security measure description text.

[0025] In one possible implementation, it also includes:

[0026] The historical operation data and historical maintenance data of the target substation are obtained, wherein the historical operation data is used to characterize the operation characteristics of the target substation under different historical operation periods; and the historical maintenance data characterizes the feedback of work results of performing work tasks under at least one of the historical operation periods.

[0027] The dynamic security rule engine is updated based on the historical operation data and historical maintenance data.

[0028] In one possible implementation, acquiring the multimodal input data includes:

[0029] Receive user voice commands;

[0030] The first input data is obtained by recognizing the user's voice command;

[0031] Based on the first input data, content retrieval is performed to obtain the original data corresponding to the target substation;

[0032] The raw data is processed based on a pre-configured large language model to generate the second input data;

[0033] Based on the first input data and the second input data, multimodal input data is obtained.

[0034] Secondly, embodiments of this application provide a substation work order generation device, comprising:

[0035] The acquisition module is used to acquire multimodal input data, which includes at least first input data with a first data format and second input data with a second data format. The first input data is used to describe the task content of the target work task of the target substation, and the second input data is used to describe the equipment spatial relationship and line connection relationship of the electrical equipment in the target substation.

[0036] The processing module is used to process the multimodal input data to generate structured task data, which describes the work process and corresponding safety measures for the target work task of the target substation.

[0037] The generation module is used to generate a work order corresponding to the target work task of the target substation based on the structured task data.

[0038] In one possible implementation, the processing module is specifically configured to: obtain structured text by extracting text content from the first input data; obtain equipment layout information by extracting symbol identifiers from the second input data; perform semantic fusion on the structured text and the equipment layout information to generate a structured task description, the structured task description including the task type of the target work task, the equipment status of the target electrical equipment, and the operation requirements; and generate structured task data based on the structured task description.

[0039] In one possible implementation, when the processing module obtains device layout information by extracting symbol identifiers from the second input data, it is specifically used to: perform image recognition on the second input data to obtain a first type of symbol and a second type of symbol, wherein the first type of symbol indicates electrical equipment and the second type of symbol indicates the connection relationship between the electrical equipment; and obtain device layout information based on the first type of symbol and the second type of symbol.

[0040] In one possible implementation, the processing module is further configured to: perform image recognition on the second input data to obtain a third type of symbol, the third type of symbol being used to indicate the operating requirements of the electrical equipment; when the processing module obtains equipment layout information based on the first type of symbol and the second type of symbol, it is specifically configured to: obtain equipment layout information based on the first type of symbol, the second type of symbol, and the third type of symbol.

[0041] In one possible implementation, when the processing module performs semantic fusion on the structured text and the device layout information to generate a structured task description, it specifically performs the following steps: extracting the semantics of the structured text and the device layout information for semantic fusion to obtain a first description text, which describes the first task step of the target work task and the electrical equipment involved in the first task step; based on the first description text, calling a dynamic safety rule engine to obtain a corresponding safety measure description text, which generates the operation requirements corresponding to the first task step; and generating the structured task description based on the first description text and the safety measure description text.

[0042] In one possible implementation, the processing module is further configured to: acquire historical operating data and historical maintenance data of the target substation, wherein the historical operating data is used to characterize the operating characteristics of the target substation under different historical operating periods; the historical maintenance data characterizes the feedback of work results of performing work tasks under at least one of the historical operating periods; and update the dynamic safety rule engine based on the historical operating data and historical maintenance data.

[0043] In one possible implementation, the acquisition module is specifically configured to: receive user voice commands; obtain first input data by recognizing the user voice commands; perform content retrieval based on the first input data to obtain raw data corresponding to the target substation; process the raw data based on a pre-configured large language model to generate second input data; and obtain multimodal input data based on the first input data and the second input data.

[0044] Thirdly, embodiments of this application provide an electronic device, including: a memory and a processor;

[0045] The memory stores computer-executed instructions;

[0046] The processor executes computer execution instructions stored in the memory, causing the processor to perform the first aspect and / or various possible implementations of the first aspect as described above.

[0047] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the first aspect and / or various possible implementations of the first aspect.

[0048] Fifthly, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the first aspect and / or various possible implementations of the first aspect.

[0049] The substation work order generation method, apparatus, electronic device, and program product provided in this application embodiment acquire multimodal input data, which includes at least first input data with a first data format and second input data with a second data format. The first input data describes the task content of the target work task at the target substation, and the second input data describes the spatial relationship of the electrical equipment and the line connection relationship within the target substation. The multimodal input data is processed to generate structured task data, which describes the work process and corresponding safety measures for the target work task at the target substation. Based on the structured task data, a work order corresponding to the target work task at the target substation is generated. By receiving multimodal input data, converting it into corresponding structured task data, and then automatically generating work orders corresponding to the target work task based on the structured task data, the efficiency and accuracy of work order generation are improved. Attached Figure Description

[0050] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0051] Figure 1 This application provides a schematic diagram of a substation equipment maintenance operation scenario.

[0052] Figure 2 A flowchart illustrating the substation work order generation method provided in this application. Figure 1 ;

[0053] Figure 3 A flowchart illustrating the specific implementation of step S102;

[0054] Figure 4 A flowchart illustrating the specific implementation of step S1022;

[0055] Figure 5 A flowchart illustrating the specific implementation of step S1023;

[0056] Figure 6 A flowchart illustrating the substation work order generation method provided in this application. Figure 2 ;

[0057] Figure 7 This application provides a schematic diagram illustrating a process for generating multimodal input data in an embodiment of the present application.

[0058] Figure 8 A schematic diagram of the substation work order generation device provided in this application;

[0059] Figure 9 A schematic diagram of the structure of the electronic device provided in this application.

[0060] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0061] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0062] The specific application scenario for this application is the maintenance and repair of substation equipment. Figure 1This application provides a schematic diagram of a substation equipment maintenance operation scenario, for reference. Figure 1 As shown, in the maintenance of substation equipment in a power system, maintenance personnel need to quickly generate compliant work orders based on on-site task requirements, primary wiring diagrams, and safety procedures. Subsequent maintenance work is then carried out based on these work orders.

[0063] In different work scenarios involving different equipment and tasks, the detailed steps and safety procedures for the maintenance tasks are different. The work order is determined by combining the task content, image files, and the contents of the safety manual. Then, the operators carry out maintenance work on the substation equipment under the guidance of the work order.

[0064] Based on the above scenario description, work orders are an important basis for ensuring operational safety. However, since the design of work orders involves a variety of information and has strong professional and personalized attributes, the existing technology usually relies on human experience to design and review work orders. This process is not only time-consuming and labor-intensive, but also prone to errors due to human factors, which can lead to safety hazards. In other words, the existing technology has the problems of low efficiency and poor accuracy in generating work orders.

[0065] The substation work order generation method provided in this application receives multimodal input data, converts the multimodal input data into corresponding structured task data, and then automatically generates work orders corresponding to the target work tasks based on the structured task data, thereby improving the efficiency and accuracy of work order generation.

[0066] The technical solution of this application and how it solves the above-mentioned technical problems will be described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will be described below with reference to the accompanying drawings.

[0067] The execution subject of the substation work order generation method provided in this embodiment can be an electronic device, such as a personal computer or mobile phone. The electronic device can execute the substation work order generation method provided in this embodiment by executing a locally running program. During execution, it can call services, models, programs, and data deployed on external servers or in the cloud to implement one or more processing steps in the substation work order generation method.

[0068] Figure 2 A flowchart illustrating the substation work order generation method provided in this application. Figure 1 ,like Figure 2 As shown, the method includes:

[0069] Step S101: Obtain multimodal input data. The multimodal input data includes at least first input data with a first data format and second input data with a second data format. The first input data is used to describe the task content of the target work task of the target substation, and the second input data is used to describe the equipment spatial relationship and line connection relationship of the electrical equipment in the target substation.

[0070] For example, multimodal input data refers to a set of input information from different data sources with different data formats, including at least first input data and second input data. The first input data describes the content of the target work task at the target substation, and the corresponding first data format includes directly input text format, speech-to-text format, etc. The second input data describes the spatial relationships of electrical equipment and the connection relationships of lines within the target substation, and the corresponding second data format includes image format, computer-aided design document format, etc. The target substation is the substation currently undergoing maintenance work, and the target work task is the specific work task for which a work order is to be issued, such as equipment power outage maintenance, routine maintenance, etc.

[0071] In practice, maintenance personnel can submit relevant information through the system's web front-end interactive interface by manually entering task text, uploading voice recordings, uploading wiring diagrams or electrical wiring files in specific formats, etc. The system backend receives and summarizes the above-mentioned input information in different formats to complete the acquisition of multimodal input data.

[0072] In another possible implementation, the system establishes a data interface with the substation's SCADA system and equipment ledger database. Based on the task keywords entered by the user, it automatically retrieves the corresponding equipment wiring files and ledger information as the second input data, and combines them with the task description entered by the user as the first input data to form multimodal input data.

[0073] In a more specific application scenario, the text command "Q1 circuit breaker power outage maintenance" submitted by the maintenance personnel and the JPG file of the primary wiring diagram of the 10kV bay of the target substation are processed by the server and output as multimodal input data containing the above text and image. The first input data is the text information "Q1 circuit breaker power outage maintenance" and the second input data is the primary wiring diagram image data.

[0074] Step S102: Process the multimodal input data to generate structured task data. The structured task data is used to describe the work process and corresponding safety measures for the target work task of the target substation.

[0075] For example, structured task data refers to data that has undergone standardization and has fixed fields and logical structures, used to fully represent the workflow and corresponding safety requirements of the target task. After obtaining multimodal input data, the system parses and extracts the first and second input data respectively, converting unstructured text, images, and other information into standardized key-value pairs and logical relationship data. Then, through semantic fusion and security rule matching, corresponding security measures information is supplemented, ultimately generating complete structured task data.

[0076] The processing of multimodal input data mainly includes two implementation methods: step-by-step parsing and fusion, and end-to-end multimodal model processing. In one possible implementation, the system independently parses the two types of input data through text processing and image processing modules, respectively, and then integrates the parsing results through a semantic fusion module, matching security rules to generate structured task data.

[0077] In another possible implementation, the system uses a pre-trained multimodal large model, directly inputs multimodal input data, and outputs structured task data that meets the preset format requirements from the model end to end.

[0078] More specifically, the multimodal input data includes the text "Q1 circuit breaker power outage maintenance" and a primary wiring diagram. After processing by the steps in this embodiment, the output is structured task data, which includes fields such as task type power outage maintenance, target equipment circuit breaker Q1, work process disconnecting the disconnecting switch - testing for voltage - hanging the grounding wire, and corresponding safety measures such as setting up fences and hanging signs.

[0079] In the above implementation method, in order to solve the technical problems of manual work order filling relying on experience, low efficiency and easy error, structured task data is generated by intelligently parsing and rule matching of multimodal input data, which achieves the effect of automatically sorting out the work process and safety requirements, and improving data processing efficiency and accuracy.

[0080] Step S103: Based on the structured task data, generate the work order corresponding to the target work task of the target substation.

[0081] For example, a work order, or substation work permit, is a legally mandated operational basis for standardizing on-site operations and clarifying safety measures. It includes fixed fields such as equipment information, operating procedures, safety measures, and the person in charge of the work. In practice, the system automatically fills the corresponding fields from the structured task data into a preset work order template. After logical validation, it generates a previewable electronic work order. Users can directly view the content on the front-end interface, export it as a PDF or Word document, or submit it for online approval.

[0082] There are several ways to generate work orders, such as template mapping and filling, or natural language generation followed by formatting. Specifically, in one possible implementation, the system pre-configures a work order template that conforms to power industry standards, maps each field in the structured task data to the corresponding column in the template, and automatically fills in the template to generate a standardized work order.

[0083] In another possible implementation, the system uses natural language generation technology to write complete work order text based on structured task data, and then typesets and outputs it according to the standard format to adapt to work order scenarios with different format requirements.

[0084] More specifically, the work order includes: filling in "Circuit Breaker Q1" in the equipment information column, filling in "disconnecting the two disconnect switches, testing for voltage, and hanging the grounding wire" in the operation procedure column, and filling in "setting up a safety fence and hanging a "Do Not Close the Switch, People Working" sign in the safety measures column.

[0085] In the above implementation method, in order to solve the technical problems of low efficiency and easy error in format when manually filling out work orders, standardized work orders are automatically generated by structured task data, which achieves the effect of improving the efficiency of work order issuance and ensuring the compliance of content format.

[0086] Furthermore, in one possible implementation, such as Figure 3 As shown, the specific implementation of step S102 includes:

[0087] Step S1021: Obtain structured text by extracting the text content from the first input data.

[0088] Step S1022: Obtain device layout information by extracting symbol identifiers from the second input data.

[0089] For example, in this step, structured text refers to text information with standardized fields extracted from the first input data, including core content such as task type, equipment number, and operational requirements; equipment layout information refers to structured information representing the location, type, and connection relationship of electrical equipment extracted from the second input data. In the specific implementation process, firstly, for the first input data, the system uses natural language processing technology to perform entity recognition and semantic extraction, extracting the core elements related to the task and organizing them into structured text; then, for the second input data, the system uses image recognition and document parsing technology to identify electrical symbols and connection identifiers, extracting the location, type, and connection relationship of the equipment to form equipment layout information.

[0090] There are several ways to extract symbol identifiers from the second input data to obtain equipment layout information. For example, for image-type second input data in formats such as JPG and PNG, the system uses OCR technology and a target detection model to recognize text and electrical symbols in the image and extract equipment layout information. Another example is for computer-aided design file-type second input data in a specific format, where the system directly reads vector data and component attributes from the file through a file parsing library and extracts equipment layout information.

[0091] In a more specific application scenario, firstly, the second input data is obtained. The second input data includes the first input data describing "power outage maintenance of circuit breaker Q1" and the second input data describing the primary wiring diagram. After processing in step S1021, the output structured text contains "Equipment number: Q1, Equipment type: Circuit breaker, Task type: Power outage maintenance". After processing in step S1022, the output equipment layout information contains "Circuit breaker Q1 is connected to disconnecting switches on both sides and transformer T2 is connected at the rear end".

[0092] In the above implementation method, in order to solve the technical problem of inconsistent multimodal data formats and inability to be directly fused and processed, the core information of the two types of input data is extracted and structured respectively, which achieves the effect of unifying the data format and providing a standardized data foundation for subsequent semantic fusion.

[0093] Furthermore, in one possible implementation, such as Figure 4 As shown, the specific implementation steps of step S1022 include:

[0094] Step S1022-1: Perform image recognition on the second input data to obtain a first type of symbol and a second type of symbol. The first type of symbol indicates electrical equipment, and the second type of symbol indicates the connection relationship between electrical equipment.

[0095] Step S1022-2: Obtain device layout information based on the first type of symbols and the second type of symbols.

[0096] For example, in the steps of this embodiment, the first type of symbols are graphic symbols representing various electrical devices in the second input data, such as symbols for circuit breakers, disconnect switches, transformers, and instrument transformers; the second type of symbols are graphic symbols representing the connection relationship between electrical devices, such as symbols for wires, busbars, and terminals.

[0097] In the specific execution process, the system first uses a trained target detection model to detect the image of the second input data, identify all first-class device symbols and second-class connection symbols and mark their positions and categories. Then, based on the positional relationship of the symbols, it sorts out the connection topology between devices and generates complete device layout information.

[0098] Furthermore, there are several ways to perform image recognition on the second input data to obtain the two types of symbols. For example, object detection based on convolutional neural networks, such as using convolutional neural network object detection models like YOLO and Faster R-CNN, can identify various electrical and connection symbols in the image. Another example is visual model detection based on the Transformer architecture. This involves using visual Transformer models such as VisionTransformer, which capture the global dependencies of symbols through a self-attention mechanism, improving the symbol recognition accuracy in scenes with dense symbols and complex layouts.

[0099] More specifically, for example, the input to the steps of the above embodiment is a wiring diagram image (second input data), and the output first type of symbols includes the circuit breaker Q1 symbol, the two disconnecting switches symbols, and the transformer T2 symbol, and the second type of symbols includes the wire symbols connecting the above-mentioned equipment; the equipment layout information generated based on the two types of symbols clarifies the connection relationship between the circuit breaker Q1 and the two disconnecting switches and the downstream transformer.

[0100] In the above implementation method, in order to solve the technical problems of low efficiency and easy omission of connection relationships when manually consulting wiring diagrams, the device and connection symbols are extracted by image recognition classification and layout information is generated, which achieves the effect of automatically sorting out the device topology relationship and improving the accuracy of information extraction.

[0101] Further, optionally, in the above embodiments, it also includes:

[0102] Step S1022-3: Perform image recognition on the second input data to obtain a third type of symbol, which is used to indicate the operating requirements of electrical equipment.

[0103] Accordingly, the specific implementation of step S1022-2 includes:

[0104] Step S1022-2A: Obtain device layout information based on the first type of symbols, the second type of symbols, and the third type of symbols.

[0105] For example, the third type of symbols are graphic or textual symbols in the second input data that identify the operating attributes and status requirements of electrical equipment, such as circuit breaker symbols, grounding symbols, rated parameter symbols, etc. In this embodiment, during the image recognition process of the second input data, based on the recognition of equipment symbols and connection symbols, the operation category symbols associated with the equipment in the image are further identified, and the third type of symbols are associated and bound with the first type of equipment symbols at the corresponding positions, ultimately generating complete equipment layout information containing equipment attributes, connection relationships, and operating requirements.

[0106] The identification method for the third type of symbols is similar to that for the first and second types of symbols. For example, text labels in the image are extracted using OCR technology as the third type of symbols, and they are bound and associated with nearby device symbols according to their position coordinates. Alternatively, a semantic segmentation model can be used to identify operational graphic symbols with specific meanings in the image, such as grounding symbols and energy storage symbols, and associate them with the corresponding devices.

[0107] Specifically, the third type of symbols includes, for example, "normally closed" status indicators and "do not touch" operation instruction indicators. In the above implementation method, in order to solve the technical problems of incomplete equipment information extraction and lack of operation attributes, the recognition and association of the third type of operation symbols are added, thereby enriching the dimensions of equipment layout information and improving the accuracy of subsequent task parsing.

[0108] Step S1023: Perform semantic fusion on the structured text and equipment layout information to generate a structured task description. The structured task description includes the task type of the target work task, the equipment status of the target electrical equipment, and the operation requirements.

[0109] Step S1024: Generate structured task data based on the structured task description.

[0110] For example, in the above steps, semantic fusion refers to the process of semantically aligning and associating task information in structured text with topological information in equipment layout information to form a unified task description (i.e., structured task description) used to standardize the complete process of the job task; the structured task description is a complete standardized description of the job task, including core elements such as task type, target equipment, equipment status, and operation requirements.

[0111] In the specific implementation process, the system uses key information such as device number as alignment anchors to associate the task instructions in the structured text with the corresponding devices in the device layout information, and combines the device connection relationship to complete the operation scope of the task, forming a complete structured task description, and then converts the task description into field-based structured task data.

[0112] Furthermore, regarding the semantic fusion process, in one possible implementation, keywords such as device number and device type are used as anchor points to match the task description in the structured text with the corresponding device in the device layout information, and the operation range is completed by combining topological relationships to achieve semantic fusion. In another possible implementation, a cross-modal attention mechanism is used to jointly model the structured text features and device layout features, automatically align the semantic associations of the two types of information, and generate the fused task description.

[0113] The above steps are explained below using a high-voltage circuit breaker power outage maintenance application scenario. For example, the input to the above steps is the structured text "Q1 circuit breaker power outage maintenance" and equipment layout information containing the connection relationships of Q1. After processing in step S1023, the output structured task description is "Perform power outage maintenance work on circuit breaker Q1, requiring the disconnecting switches on both sides of Q1 to be disconnected." Subsequently, the above structured task description is further categorized and organized to generate structured task data. The structured task data includes standardized fields such as task type, target equipment, and operation scope.

[0114] In the above implementation method, in order to solve the technical problems of the separation between task information and equipment topology information and the inaccurate judgment of the operation range, the task instructions and equipment layout information are associated through semantic fusion, which achieves the effect of accurately sorting out the operation range and operation requirements and improving the completeness of task description.

[0115] Furthermore, in one possible implementation, such as Figure 5 As shown, the specific implementation steps of step S1023 include:

[0116] Step S1023-1: Extract the semantics of the structured text and equipment layout information and perform semantic fusion to obtain the first description text. The first description text is used to describe the first task step of the target work task and the electrical equipment involved in the first task step.

[0117] Step S1023-2: Based on the first description text, call the dynamic security rule engine to obtain the corresponding security measure description text. The security measure description text is used to generate the operation requirements corresponding to the first task step.

[0118] Step S1023-3: Generate a structured task description based on the first description text and the security measure description text.

[0119] For example, the first description text is a basic task description that only includes the work steps and the equipment involved. This first description text can be the result of semantic fusion of structured text and equipment layout information in step S1023 of the previous embodiment. In the implementation process of the structured task description, the dynamic safety rule engine is a rule processing module that stores power safety regulations, can dynamically update rules, and automatically match safety measures; the safety measure description text is the matched safety requirement text corresponding to the work steps.

[0120] First, the system integrates the two types of information to generate a first descriptive text containing only basic operation steps, such as a structured task description text. Then, the first descriptive text is input into the dynamic security rule engine. The engine matches the corresponding security measures from the rule base according to the device type and operation type and generates a security measure description text. Finally, the basic steps and security measures are integrated to generate a complete structured task description.

[0121] The matching process of the dynamic safety rule engine can be implemented in two ways: keyword rule mapping matching and knowledge base reasoning supplementation. Specifically, in one possible implementation, the rule engine extracts the equipment type and operation type keywords from the first description text, retrieves the corresponding preset safety measures in the rule base, and generates a safety measure description text; in another possible implementation, the rule engine combines the power operation knowledge base to perform logical reasoning, supplements derived safety measures based on the rule base matching results, and generates a more comprehensive safety measure description text.

[0122] In a more specific application scenario, for example, the initial description text is "Disconnect the isolating switches on both sides of circuit breaker Q1"; after calling the dynamic safety rule engine, the resulting safety measure description text is "Test for voltage, connect grounding wire, set up safety fence"; the final generated structured task description includes the above basic operation steps and corresponding safety measures. In this implementation method, to address the technical problems of safety measures relying on manual judgment, being prone to omissions, and being unable to adapt to new regulations, the dynamic safety rule engine is used to automatically match safety measures, achieving the effect of ensuring the compliance of safety measures and improving the efficiency of measure matching.

[0123] Furthermore, prior to this embodiment, it also includes:

[0124] Step S100A: Obtain historical operation data and historical maintenance data of the target substation. The historical operation data is used to characterize the operation characteristics of the target substation under different historical operation periods. The historical maintenance data characterizes the feedback of work results of performing work tasks under at least one historical operation period.

[0125] Step S100B: Update the dynamic security rule engine based on historical operation data and historical maintenance data.

[0126] For example, in this step, historical operational data refers to data such as past equipment operating parameters, load status, and operating environment of the target substation, used to characterize the operational characteristics under different historical operating periods; historical maintenance data refers to data such as past work order records, execution feedback, safety event records, and acceptance results, used to characterize the execution result feedback of historical work tasks. Before calling the dynamic safety rule engine, the system periodically collects historical operation and maintenance data of the substation, analyzes the operational risks and effectiveness of measures under different operating scenarios, and achieves dynamic updates of the rule engine by adjusting the matching weights and rule entries of the rule engine, ensuring that the safety rules are adapted to the actual needs on site.

[0127] Furthermore, updating the dynamic safety rule engine mainly includes two implementation methods: manual import of new regulations for synchronous updates and incremental learning for automatic iteration. In one possible implementation, when the power safety regulations are updated, the management personnel import the new regulations into the system, and the rule engine synchronously updates the entries and priorities in the rule base. In another possible implementation, the system uses an online incremental learning algorithm to train on historical operation and maintenance data, automatically optimizes the rule matching strategy, and supplements rules to adapt to the field scenario.

[0128] For example, by using equipment operation data and maintenance work order feedback data from the past year, an updated dynamic safety rule engine can be generated. More specifically, for example, a rule entry for "outdoor maintenance requires additional windproof measures" can be added based on historical feedback, and the priority of safety measures for maintenance during high-load periods can be adjusted.

[0129] In the above implementation method, in order to solve the technical problem of the rule base being statically updated lagging behind and unable to adapt to the actual on-site scenario, the security rule engine is dynamically updated through historical data, which achieves the effect of improving rule adaptability and ensuring that security measures are synchronized with on-site needs and the latest procedures.

[0130] The substation work order generation method provided in this embodiment acquires multimodal input data containing task descriptions and equipment wiring relationships. It then parses and extracts heterogeneous data from multiple sources, performs semantic fusion to generate structured task descriptions, and combines this with a dynamic safety rule engine to match safety measures, generating structured task data. Finally, it automatically generates standardized work orders, forming a closed-loop processing flow of "input-parsing-matching-generation." This embodiment breaks through the information fragmentation limitations of traditional manual invoicing, significantly improves work order issuance efficiency, reduces the risk of human error, ensures the compliance and consistency of work order content with safety procedures, and adapts to the complex and ever-changing operational needs of substations.

[0131] The substation work order generation method provided in this application embodiment acquires multimodal input data, which includes at least first input data with a first data format and second input data with a second data format. The first input data describes the task content of the target work task at the target substation, and the second input data describes the spatial relationship of electrical equipment and the connection relationship of lines within the target substation. The multimodal input data is processed to generate structured task data, which describes the work process and corresponding safety measures for the target work task at the target substation. Based on the structured task data, a work order corresponding to the target work task at the target substation is generated. By receiving multimodal input data, converting it into corresponding structured task data, and then automatically generating the work order corresponding to the target work task based on the structured task data, the efficiency and accuracy of work order generation are improved.

[0132] Figure 6 A flowchart illustrating the substation work order generation method provided in this application. Figure 2 ,like Figure 6 As shown, in this embodiment... Figure 2 Based on the embodiments, the human-computer interaction process is described in detail, and the method includes:

[0133] Step S201: Receive user voice commands.

[0134] Step S202: Obtain the first input data by recognizing the user's voice command.

[0135] Step S203: Perform content retrieval based on the first input data to obtain the original data corresponding to the target substation.

[0136] For example, in this step, the user's voice command is the task requirement information input by the maintenance personnel via voice; the first input data is the text-formatted task description obtained after voice command recognition; and the raw data is the basic information related to the target substation, such as equipment ledgers, wiring diagrams, and historical records, retrieved by the system. During operation, maintenance personnel can speak their task requirements through the system's front-end voice input function. The system automatically recognizes the voice content and converts it into text, then automatically retrieves the relevant raw data of the corresponding substation based on keywords in the text.

[0137] Furthermore, in the process of acquiring raw data, in one possible implementation, the system extracts keywords such as substation name and equipment number from the first input data and performs precise matching and retrieval in the equipment ledger database and drawing library to obtain the corresponding raw data; in another possible implementation, natural language understanding technology is used to analyze the semantics of the first input data and expand the retrieval of relevant equipment intervals, associated equipment and other information to obtain more comprehensive raw data.

[0138] In a more specific application scenario, the input is first the user's voice command "Q1 circuit breaker power outage maintenance." This voice command is then recognized, and the first input data output is the text "Q1 circuit breaker power outage maintenance." Next, based on this first input data, a retrieval is performed to obtain the raw data, which includes the equipment ledger information and the original primary wiring diagram of the 10kV bay of the target substation. In this implementation, to solve the technical problems of tedious manual input and low efficiency in manually searching for data, voice recognition and automatic retrieval of task information and basic data are used, thus simplifying user operations and improving information acquisition efficiency.

[0139] Step S204: Process the original data based on the pre-configured large language model to generate the second input data.

[0140] Step S205: Based on the first input data and the second input data, obtain multimodal input data.

[0141] For example, in this step, the pre-configured large language model is a large language model that has been fine-tuned with data from the power sector and has the ability to process and generate structured data of power equipment; the second input data is structured data representing the spatial and connection relationships of equipment generated by the large language model after processing the original data.

[0142] After obtaining the raw data through the above steps, the system inputs the retrieved raw data into a pre-configured large language model. The model extracts and organizes the layout and connection information of the devices to generate standardized second input data, which is then integrated with the first input data to form complete multimodal input data.

[0143] Specifically, in one possible implementation, the large language model extracts existing equipment connection relationship information from the original ledgers and drawings, and organizes it into structured second input data; or, the large language model combines knowledge of the power field to infer and complete the equipment topology relationship based on the scattered information in the original data, and generates more complete second input data.

[0144] Based on the application scenario of high-voltage circuit breaker power outage maintenance, the input for this step is the original data of the equipment ledger and primary wiring diagram of the 10kV bay. The output of the second input data is the structured equipment layout data containing circuit breaker Q1, disconnector, transformer T2 and connection relationships. After integration with the first input data, complete multimodal input data is obtained.

[0145] In the above implementation method, in order to solve the technical problem that the original data is not uniform in format and cannot be directly used as the second input data, the original data is processed by a large language model to generate standardized second input data, thereby improving data availability and ensuring the quality of multimodal data.

[0146] Figure 7 This application provides a schematic diagram of a process for generating multimodal input data, as illustrated in the embodiments of this application. Figure 7 As illustrated, exemplarily, firstly, the user inputs speech via a personal computer (electronic device). This speech is then converted to generate text describing the target work tasks of the target substation—this is the first input data. Next, a data query is performed within a database or knowledge base based on the first input data to obtain the raw data. Then, the raw data is processed using a large language model, combined with prior knowledge of the target substation, to generate second input data describing the spatial relationships of electrical equipment and the connections between electrical lines within the target substation. Finally, the first and second input data are merged to generate multimodal input data.

[0147] Step S206: Obtain multimodal input data. The multimodal input data includes at least first input data with a first data format and second input data with a second data format. The first input data is used to describe the task content of the target work task of the target substation, and the second input data is used to describe the equipment spatial relationship and line connection relationship of the electrical equipment in the target substation.

[0148] Step S207: Process the multimodal input data to generate structured task data. The structured task data is used to describe the work process and corresponding safety measures for the target work task of the target substation.

[0149] Step S208: Based on the structured task data, generate the work order corresponding to the target work task of the target substation.

[0150] The implementation methods of steps S206 to S208 are the same as those of steps S101 to S103 in the first embodiment, and will not be repeated here.

[0151] This embodiment, based on the first embodiment, optimizes the input process to meet on-site human-computer interaction needs. It receives user voice commands to obtain the first input data, automatically retrieves and processes it using a large language model to generate the second input data, and integrates it into multimodal input data to complete the subsequent parsing, matching, and work order generation process. This embodiment further lowers the operational threshold for maintenance personnel, adapts to scenarios such as noisy substation sites and situations where workers' hands are occupied, and improves the speed and convenience of invoicing response in emergency repairs and temporary operations.

[0152] Figure 8 This is a schematic diagram of the substation work order generation device provided in this application, as shown below. Figure 8 As shown, the substation work order generation device 40 provided in this embodiment includes:

[0153] The acquisition module 401 is used to acquire multimodal input data. The multimodal input data includes at least first input data with a first data format and second input data with a second data format. The first input data is used to describe the task content of the target work task of the target substation, and the second input data is used to describe the equipment spatial relationship and line connection relationship of the electrical equipment in the target substation.

[0154] The processing module 402 is used to process the multimodal input data and generate structured task data. The structured task data is used to describe the work process and corresponding safety measures for the target work task of the target substation.

[0155] The generation module 403 is used to generate work orders corresponding to the target work tasks of the target substation based on structured task data.

[0156] In one possible implementation, the processing module 402 is specifically used for: obtaining structured text by extracting text content from the first input data; obtaining equipment layout information by extracting symbol identifiers from the second input data; performing semantic fusion on the structured text and equipment layout information to generate a structured task description, the structured task description including the task type of the target work task, the equipment status of the target electrical equipment, and the operation requirements; and generating structured task data based on the structured task description.

[0157] In one possible implementation, when the processing module 402 obtains the device layout information by extracting symbol identifiers from the second input data, it is specifically used to: perform image recognition on the second input data to obtain a first type of symbol and a second type of symbol, wherein the first type of symbol indicates electrical equipment and the second type of symbol indicates the connection relationship between electrical equipment; and obtain the device layout information based on the first type of symbol and the second type of symbol.

[0158] In one possible implementation, the processing module 402 is further configured to: perform image recognition on the second input data to obtain a third type of symbol, the third type of symbol being used to indicate the operating requirements of the electrical equipment; when the processing module obtains the equipment layout information based on the first type of symbol and the second type of symbol, it is specifically configured to: obtain the equipment layout information based on the first type of symbol, the second type of symbol and the third type of symbol.

[0159] In one possible implementation, when the processing module 402 performs semantic fusion on the structured text and equipment layout information to generate a structured task description, it specifically performs the following steps: extracting the semantics of the structured text and equipment layout information and performing semantic fusion to obtain a first description text, which describes the first task step of the target work task and the electrical equipment involved in the first task step; based on the first description text, calling the dynamic safety rule engine to obtain a corresponding safety measure description text, which is used to generate the operation requirements corresponding to the first task step; and generating a structured task description based on the first description text and the safety measure description text.

[0160] In one possible implementation, the processing module 402 is further configured to: acquire historical operating data and historical maintenance data of the target substation, wherein the historical operating data is used to characterize the operating characteristics of the target substation under different historical operating periods; the historical maintenance data characterizes the feedback of work results of performing work tasks under at least one historical operating period; and update the dynamic safety rule engine based on the historical operating data and historical maintenance data.

[0161] In one possible implementation, the acquisition module 401 is specifically used for: receiving user voice commands; obtaining first input data by recognizing user voice commands; performing content retrieval based on the first input data to obtain the original data corresponding to the target substation; processing the original data based on a pre-configured large language model to generate second input data; and obtaining multimodal input data based on the first input data and the second input data.

[0162] The substation work order generation device provided in this embodiment can execute the method provided in the above method embodiment. Its implementation principle and technical effect are similar, and will not be described in detail here.

[0163] Figure 9 A schematic diagram of the structure of the electronic device provided in this application. Figure 9 As shown, the electronic device 50 provided in this embodiment includes at least one processor 501 and a memory 502. Optionally, the device 50 further includes a communication component 503. The processor 501, memory 502, and communication component 503 are connected via a bus 504.

[0164] In a specific implementation, at least one processor 501 executes computer execution instructions stored in memory 502, causing at least one processor 501 to perform the above-described method.

[0165] The specific implementation process of processor 501 can be found in the above method embodiments, and its implementation principle and technical effect are similar. It will not be repeated here.

[0166] In the above embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor.

[0167] The memory may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage device.

[0168] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0169] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0170] This application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the above-described method.

[0171] The aforementioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium can be any available medium accessible to a general-purpose or special-purpose computer.

[0172] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium.

[0173] Of course, readable storage media can also be a component of the processor. The processor and readable storage media can reside in an Application Specific Integrated Circuit (ASIC). Alternatively, the processor and readable storage media can exist as discrete components within the device.

[0174] The division of units is merely a logical functional division. In actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed.

[0175] Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interface, device or unit, and can be electrical, mechanical or other forms.

[0176] The unit described as a separate component may or may not be physically separate. The component shown as a unit may or may not be a physical unit, that is, it may be located in one place or it may be distributed across multiple network units.

[0177] Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0178] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0179] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium.

[0180] Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention.

[0181] The aforementioned storage media include: USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, optical disks, and other media that can store program code.

[0182] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions.

[0183] The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disk, or optical disk.

[0184] Finally, it should be noted that other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein, and is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope.

[0185] The scope of this invention is limited only by the appended claims.

Claims

1. A method for generating substation work orders, characterized in that, include: Acquire multimodal input data, wherein the multimodal input data includes at least first input data having a first data format and second input data having a second data format, wherein the first input data is used to describe the task content of the target work task of the target substation, and the second input data is used to describe the equipment spatial relationship and line connection relationship of the electrical equipment in the target substation; The multimodal input data is processed to generate structured task data, which is used to describe the work process and corresponding safety measures for the target work task of the target substation. Based on the structured task data, a work order corresponding to the target work task of the target substation is generated.

2. The method according to claim 1, characterized in that, The process of processing the multimodal input data to generate structured task data includes: By extracting the text content from the first input data, structured text is obtained; The device layout information is obtained by extracting the symbol identifiers from the second input data; Semantic fusion is performed on the structured text and the equipment layout information to generate a structured task description, which includes the task type of the target work task, the equipment status of the target electrical equipment, and the operation requirements. Based on the structured task description, structured task data is generated.

3. The method according to claim 2, characterized in that, The step of extracting symbol identifiers from the second input data to obtain device layout information includes: Image recognition is performed on the second input data to obtain a first type of symbol and a second type of symbol. The first type of symbol indicates electrical equipment, and the second type of symbol indicates the connection relationship between the electrical equipment. Based on the first type of symbols and the second type of symbols, the device layout information is obtained.

4. The method according to claim 3, characterized in that, Also includes: The second input data is subjected to image recognition to obtain a third type of symbol, which is used to indicate the operating requirements of electrical equipment; The process of obtaining device layout information based on the first type of symbols and the second type of symbols includes: Based on the first type of symbols, the second type of symbols, and the third type of symbols, the device layout information is obtained.

5. The method according to claim 2, characterized in that, The step of semantically fusing the structured text and the device layout information to generate a structured task description includes: The semantics of the structured text and the equipment layout information are extracted and semantically fused to obtain a first description text. The first description text is used to describe the first task step of the target work task and the electrical equipment involved in the first task step. Based on the first description text, the dynamic security rule engine is invoked to obtain the corresponding security measure description text, which is used to generate the operation requirements corresponding to the first task step; The structured task description is generated based on the first description text and the security measure description text.

6. The method according to claim 5, characterized in that, Also includes: The historical operation data and historical maintenance data of the target substation are obtained, wherein the historical operation data is used to characterize the operation characteristics of the target substation under different historical operation periods; and the historical maintenance data characterizes the feedback of work results of performing work tasks under at least one of the historical operation periods. The dynamic security rule engine is updated based on the historical operation data and historical maintenance data.

7. The method according to claim 1, characterized in that, The acquisition of multimodal input data includes: Receive user voice commands; The first input data is obtained by recognizing the user's voice command; Based on the first input data, content retrieval is performed to obtain the original data corresponding to the target substation; The raw data is processed based on a pre-configured large language model to generate the second input data; Based on the first input data and the second input data, multimodal input data is obtained.

8. A substation work order generation device, characterized in that, include: The acquisition module is used to acquire multimodal input data, which includes at least first input data with a first data format and second input data with a second data format. The first input data is used to describe the task content of the target work task of the target substation, and the second input data is used to describe the equipment spatial relationship and line connection relationship of the electrical equipment in the target substation. The processing module is used to process the multimodal input data to generate structured task data, which describes the work process and corresponding safety measures for the target work task of the target substation. The generation module is used to generate a work order corresponding to the target work task of the target substation based on the structured task data.

9. An electronic device, characterized in that, include: Memory, processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory, causing the processor to perform the method as described in any one of claims 1-8.

10. A computer program product, characterized in that, Includes a computer program that, when executed by a processor, implements the method described in any one of claims 1-8.

Images