Working system, method and equipment of intelligent distribution network hot-line work robot with body and storage medium

Abstract

The invention discloses an operation system, method and equipment of a live working robot with an intelligent distribution network, and a storage medium, and relates to the technical field of intelligent distribution networks of robots, and the system comprises a first man-machine interaction module which supports bidirectional interaction with an operator through voice, characters, images and the like; the second multi-mode sensing module senses the equipment state, the obstacle and the environment change in the working environment in real time through a visual sensor, an auditory sensor, a force sensor and the like; the big language model processes and analyzes the instruction through a natural language and generates a task execution step, and adjusts a task strategy according to the perception data; the decision planning module formulates a task path and resource scheduling in combination with a multi-objective optimization algorithm; the safety monitoring module monitors the safety state in the operation process in real time and triggers early warning; the anomaly detection and processing module receives the early warning information and starts an autonomous navigation and path planning technology; according to the system, autonomous and safe operation of the robot is achieved, and task execution efficiency, adaptability and flexibility are improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent power distribution network technology, specifically to an intelligent power distribution network live-line working robot system, method, equipment, and storage medium. Background Technology

[0002] With the rapid development of robotics, artificial intelligence, and automated control, robots have gradually become important tools in the power industry. Live-line working robots can not only replace humans in high-voltage and hazardous working environments, but also improve work efficiency and reduce human error. However, current live-line working robots still face several challenges, particularly in their ability to perceive and respond to the environment, their flexibility in task execution, and their interaction with operators. Existing robots have a low level of intelligence in human-robot interaction, typically relying on manual remote control or fixed commands, unable to engage in natural two-way communication with operators, and struggling to adjust in real time according to environmental changes. In complex and dynamic working environments, robots often lack sufficient environmental adaptability and generalization capabilities, making it difficult to handle multimodal information and multi-objective tasks.

[0003] Currently, the adaptability of robots to live-line working environments is a particularly prominent issue. Especially when facing different environmental conditions and task requirements, existing robots mostly rely on preset environmental models, making it difficult to adjust work strategies in real time and lacking sufficient adaptability. Although robots in existing technologies perform well in simulation environments, in actual applications, due to hardware limitations, insufficient information processing capabilities, and the complexity of environmental changes, they often cannot ensure the efficiency and safety of operations.

[0004] In terms of multimodal data processing, existing robot systems typically can only process information from a single sensor, lacking the ability to comprehensively process multidimensional sensory data from vision, hearing, force, and other senses, thus failing to fully perceive dynamic changes in the working environment. Furthermore, when performing complex tasks, the execution sequence and path planning of robots still lack sufficient flexibility, resulting in low operational efficiency and difficulty in making immediate responses in emergency situations. Summary of the Invention

[0005] In view of the above-mentioned problems, the present invention is proposed.

[0006] Therefore, the technical problem solved by this invention is that existing technologies in live-line work of power distribution networks have problems such as poor environmental adaptability, weak human-machine interaction capabilities, low task execution flexibility, and untimely response to complex environmental changes. Existing robots often rely on single sensor data and lack the ability to fuse and process multimodal information. The invention also addresses how to improve the robot's intelligent analysis, execution capabilities, and safety assurance in complex and ever-changing working environments.

[0007] To address the aforementioned technical problems, this invention provides the following technical solution: an embodied intelligent power distribution network live-line working robot system, comprising a first human-machine interaction module and a second multimodal perception module for bidirectional interaction between the user and the robot by issuing work instructions; including visual, auditory, and force sensors to perceive the status of target equipment, obstacles, and environmental changes in the work environment in real time; a large language model that uses natural language processing to parse the semantic information in the work instructions issued by the first human-machine interaction module, generates specific execution steps of the task based on context, and adjusts the task execution strategy according to the environmental perception data of the second multimodal perception module; a decision planning module that, based on the task execution strategy generated by the large language model and combined with a multi-objective optimization algorithm, formulates the robot's motion trajectory, work sequence, and collaborative process in real time, and performs task path resource scheduling; a safety monitoring module that monitors different types of safety states during the robot's operation in real time, detects potential dangers and triggers warnings based on safety standards and anomaly detection mechanisms; and an anomaly detection and processing module that receives warning information from the safety monitoring module, combines the second multimodal perception module and the large language model, and uses autonomous navigation and path planning technology to initiate relevant processing procedures for the warning information, allowing the robot to autonomously locate and avoid obstacles in complex environments.

[0008] As a preferred embodiment of the intelligent live-line working robot system for power distribution networks described in this invention, the first human-machine interaction module includes an instruction input processing unit and a multimodal signal conversion unit. The instruction input processing unit supports users to input work instructions through various means such as voice, text, and images. The multimodal signal conversion unit converts different forms of input work instructions into a unified machine instruction format for multimodal input.

[0009] As a preferred embodiment of the intelligent live-line working robot system for power distribution networks described in this invention, the second multimodal perception module includes an environmental perception unit and a data fusion module. The environmental perception unit includes visual, auditory, and force sensors. The data fusion module integrates the data from the sensors in the environmental perception unit to generate a comprehensive environmental information map, and constructs a complete dynamic picture of the environment through real-time feedback from the multi-sensor system.

[0010] As a preferred embodiment of the intelligent live-line working robot system for power distribution networks described in this invention, the large language model includes: receiving voice and text commands transmitted by the operator through a first human-machine interaction module; converting the voice into text through a voice recognition unit; analyzing the grammatical structure of the commands through text parsing; and extracting the task objectives and operation objects from the commands. A semantic understanding unit, combining contextual information, uses a deep learning algorithm to perform semantic analysis on the commands, identifies key task information in the commands, and, based on the semantic parsing results, decomposes complex tasks into executable sub-tasks, generating operational steps that the robot can execute. The deep learning algorithm is expressed as: ; in, Conditional probability refers to the probability given the input text. Visual information and audio information ,Task The probability of being executed. This represents the weight matrix, which is multiplied by the LSTM output. Assume the size of the LSTM output vector is... The size of the weight matrix should be , It is the dimension of the output task space. This represents a feature fusion function that fuses visual and audio information into a joint feature vector. This represents the bias term, which shifts and adjusts the output of the large language model; its dimension is 1.

[0011] As a preferred embodiment of the intelligent live-line working robot system for power distribution networks described in this invention, the decision-making and planning module includes a path generation unit and a task coordination unit. The path generation unit includes real-time planning of the robot's motion path and work sequence based on a task execution strategy generated from a large language model. The task coordination unit includes coordinating the work of different modules during task execution using a multi-objective optimization algorithm. The multi-objective optimization algorithm is expressed as follows: ; in, This represents the total cost of the target task after optimization. Indicates task The weighting coefficients, This represents the weighting coefficient of execution time cost in the total cost. This represents the weighting factor of resource consumption in the total cost. This represents the weighting coefficient of environmental adaptability in the total cost. Indicates task Execution time, Indicates task resource consumption, Indicates task Environmental adaptability This indicates the total number of tasks currently in progress.

[0012] As a preferred embodiment of the intelligent live-line working robot system for power distribution networks described in this invention, the safety monitoring module includes a real-time monitoring unit and an anomaly warning unit. The real-time monitoring unit monitors different safety states during the operation process in real time, and, in conjunction with the anomaly warning unit, promptly detects safety hazards and issues alarms. The anomaly warning unit includes triggering a soft restart, automatic shutdown, or fault recording program when the system detects sensor failure, motor failure, or power system anomaly; attempting to switch to a backup communication channel when the robot cannot communicate with the remote monitoring center or other equipment; pausing operation and awaiting manual intervention if communication fails to recover; and responding to unidentified obstacles or environmental conditions that do not meet operational requirements when the robot detects them in the working environment. The system will automatically avoid obstacles and adjust operating parameters. If it cannot adapt to changes in the environment, it will pause the task. When the distance between the robot and electrical equipment or obstacles is less than the safety threshold set based on the average historical threshold, the system will immediately stop the operation, trigger a safety alarm, and adjust its position to restore a safe operating state. When the robot experiences material loss or operational failure during task execution, the system will reschedule the task, pause the operation, and wait for manual intervention. When the robot's computing resources and storage space usage exceed the maximum memory capacity of the task, the system will adjust the task priority, clean up useless data, and reallocate resources to restore a normal operating state. When the algorithm or system malfunctions, the system will restart the relevant modules, record error logs, and switch to a backup control program to continue executing the task if necessary.

[0013] As a preferred embodiment of the intelligent distribution network live-line working robot system described in this invention, the anomaly detection and processing module includes an autonomous navigation unit. The autonomous navigation unit combines real-time early warning information from the safety monitoring module with environmental data from the second multimodal perception module, and adjusts the robot's working path in real time through a dynamic path adjustment algorithm. In response to emergencies, it automatically avoids obstacles and safely completes the task.

[0014] Another objective of this invention is to provide an embodied intelligent live-line working robot method for power distribution networks, which can address the shortcomings of current live-line working robots in terms of environmental adaptability, human-computer interaction capabilities, and task execution flexibility by introducing a scheme that combines multimodal perception and large language models.

[0015] As a preferred embodiment of the live-line working method of the embodied intelligent distribution network robot described in this invention, the method includes: receiving multimodal input commands from the operator through a first human-machine interaction module; parsing the semantic information in the commands through a large language model; generating task execution steps based on the context; and adjusting the task execution strategy based on environmental perception data from a second multimodal perception module. Based on the task execution strategy generated by the large language model, a decision planning module plans the robot's motion path and work sequence in real time, coordinates the work of each module using a multi-objective optimization algorithm, and optimizes task path resource scheduling. A safety monitoring module monitors the safety status of the robot during operation in real time, and, in conjunction with an anomaly detection and handling module, initiates corresponding processing procedures upon receiving early warning information, adjusts the robot's work path, and autonomously avoids obstacles.

[0016] Another object of the present invention is to provide a live-line working robot for a power distribution network, comprising a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program as a step to realize the live-line working robot system for a power distribution network.

[0017] Another object of the present invention is to provide a working storage medium for a live-line working robot for a power distribution network, wherein a computer program is stored thereon, and when the computer program is executed by a processor, the steps of the working system of the live-line working robot for a power distribution network are implemented.

[0018] The beneficial effects of the present invention are as follows: The embodied intelligent distribution network live-line working robot system provided by the present invention achieves natural language interaction and real-time environmental perception by combining large language models and multimodal perception technology, intelligent decision-making and path planning by combining multi-objective optimization algorithms, and ensures the safety of operation through safety monitoring and anomaly handling mechanisms. The present invention achieves better results in improving the intelligence of robot task understanding and execution, adaptability to the working environment, resource scheduling optimization and safety assurance capabilities. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is an overall connection diagram of an intelligent distribution network live-line working robot system provided in Embodiment 1 of the present invention.

[0021] Figure 2This is an overall flowchart of a live-line working method for a smart distribution network robot provided in Embodiment 2 of the present invention. Detailed Implementation

[0022] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0023] Example 1, referring to Figure 1 As one embodiment of the present invention, a live-line working robot system for power distribution networks is provided, comprising: The system comprises: a first human-computer interaction module S, a second multimodal perception module 200, a large language model 300, a decision planning module 400, a security monitoring module 500, an anomaly detection and handling module 600, a refined operation execution module 700, and an autonomous operation module 800.

[0024] It should be noted that the first human-machine interaction module S includes a command input processing unit S1, which receives operation commands input by operators in various ways such as voice, text, and images, and a multimodal signal conversion unit S2, which converts different forms of input operation commands into a unified machine command format to support complex multimodal input tasks.

[0025] The second multimodal perception module 200 includes an environmental perception unit 200a for collecting data from sensors such as vision, hearing, and force, perceiving the status of target equipment, obstacles, and environmental changes in the work environment, and fusing the multi-sensor data through the data fusion module 200b to generate a comprehensive environmental information map, thereby constructing a complete dynamic picture of the work environment.

[0026] The large language model 300 converts the voice commands transmitted by the operator through the first human-computer interaction module S into text through the speech recognition unit 300a. After text parsing, the grammatical structure of the commands is analyzed, and the task objectives and operation objects in the commands are extracted. The semantic understanding unit 300b combines contextual information and uses deep learning algorithms to perform semantic analysis on the commands, identify key task information in the commands, and generate executable operation steps. All tasks will be adjusted according to feedback information from the actual working environment.

[0027] Deep learning algorithms are represented as: ; in, Conditional probability refers to the probability given the input text. Visual information and audio information ,Task The probability of being executed. This represents the weight matrix, which is multiplied by the LSTM output. Assume the size of the LSTM output vector is... The size of the weight matrix should be , It is the dimension of the output task space. This represents a feature fusion function that fuses visual and audio information into a joint feature vector. This represents the bias term, which shifts and adjusts the output of the large language model; its dimension is 1.

[0028] The decision planning module 400 plans the robot's motion path and work sequence in real time through the path generation unit 400a based on the task execution strategy generated by the large language model 300. At the same time, the task coordination unit 400b, combined with the multi-objective optimization algorithm 400c, schedules task path resources and optimizes the execution strategy.

[0029] The multi-objective optimization algorithm 400c is represented as: ; in, This represents the total cost of the target task after optimization. Indicates task The weighting coefficients, This represents the weighting coefficient of execution time cost in the total cost. This represents the weighting factor of resource consumption in the total cost. This represents the weighting coefficient of environmental adaptability in the total cost. Indicates task Execution time, Indicates task resource consumption, Indicates task Environmental adaptability This indicates the total number of tasks currently in progress.

[0030] The safety monitoring module 500 includes a real-time monitoring unit 500a, which monitors the safety status during the operation in real time, and an abnormal early warning unit 500b, which triggers a safety warning and takes timely emergency measures when a potential hazard is detected.

[0031] When the robot is unable to communicate with the remote monitoring center or other devices, it attempts to switch to a backup communication channel. If communication fails to be restored, the operation is suspended and manual intervention is required. When the robot detects unidentified obstacles or environmental conditions that do not meet the operational requirements in the working environment, the system will automatically avoid obstacles and adjust operational parameters. If it cannot adapt to changes in the environment, it will pause the task. When the distance between the robot and electrical equipment or obstacles is less than the safety threshold set based on the average historical threshold, the system will immediately stop operation, trigger a safety alarm, and adjust its position to restore a safe operating state. When the robot encounters material shortages or operational failures during task execution, the system will reschedule the task, suspend operations, and await human intervention. When the robot's computing resources and storage space usage exceed the maximum memory capacity of the task, the system will adjust the task priority, clean up useless data, and reallocate resources to restore normal operation. When an algorithm or system malfunctions, the system will restart the relevant modules, record error logs, and switch to a backup control program to continue executing the task if necessary.

[0032] The anomaly detection and handling module 600, through the autonomous navigation unit 600a, combined with the real-time early warning information from the safety monitoring module 500 and the environmental data from the second multimodal perception module 200, uses a dynamic path adjustment algorithm to adjust the robot's working path in real time, ensuring that the robot can autonomously avoid obstacles and continue to perform its tasks when dealing with emergencies.

[0033] The precision operation execution module 700 includes a high-precision robotic arm and motion control system. Based on the task path resource scheduling of the decision planning module 400, it performs precision operation tasks such as live wire installation and fault indicator location.

[0034] The autonomous operation module 800 integrates all the above modules and dynamically adjusts the execution strategy through an intelligent collaboration mechanism to complete the operation instructions.

[0035] It should also be noted that the person in charge of the work first organizes a site survey, comprehensively considering the surrounding environment, conductor specifications, terrain factors, and possible operational risks. Taking into account factors such as the surrounding environment of the tower, conductor specifications, and terrain, the person in charge determines whether live-line work is possible and determines the work method and safety technical measures. A live-line work permit is issued. Workers must meet the requirements for health, qualifications, knowledge and skills. Various tools and materials such as robots, insulated shoes, and smart safety helmets are prepared. The bucket truck driver parks the vehicle in a suitable position and completes the outrigger setup. At the same time, safety fences and warning signs are set up on site.

[0036] Determining the necessary safety technology additions for power operation tasks requires assessing potential risks in the work environment, such as aging power facilities and external weather effects, to determine relevant safety protection measures and select a suitable robot to perform the task. The person in charge of the work develops a detailed work plan based on the survey results. After confirming that all conditions meet safety requirements on site, the person in charge of the work assigns the robot to start performing the task based on the task requirements and the robot's capabilities.

[0037] The person in charge of the work tracks the progress of the operation in real time through the remote monitoring center and adjusts the execution order of the tasks according to the on-site tasks. The system adjusts the working status in real time by identifying the line (such as identifying the voltage level and operating status of the power line), the status of the power equipment (such as the equipment operation data monitored by the sensor), and sensor signals (such as environmental data such as temperature, humidity, and pressure). By integrating environmental data and task instructions, the robot can make intelligent decisions in real time and adjust the operation path according to the task requirements to ensure that the task can be executed smoothly in the predetermined priority order and that safety is guaranteed.

[0038] When the robot begins to perform a task, it first identifies the power lines and target area of ​​the equipment through path planning technology and generates the optimal operation path. The robot then completes the installation and debugging of specific devices according to the task instructions. For example, when installing a power fault indicator, the robot can accurately identify the installation location and adjust its actions to ensure that the equipment status meets the preset standards.

[0039] During operation, the safety monitoring module 500 monitors every step the robot takes in real time. The robot obtains real-time data from the environment through sensors and the environmental perception module 200, monitors and adjusts the robot's behavior path. When a physical collision or other safety hazard is detected, the automatic system will immediately trigger an early warning mechanism to avoid risks by adjusting the operation strategy or suspending the task.

[0040] After receiving the audio data, the speech recognition network in the second multimodal perception module 200 of the robot's embodied intelligence is activated, translating human speech into text and sending this text to the large language model. For example, "Move to the work position, identify the line, install the fault indicator, confirm the installation status, return to the initial position" translates to the robot's instruction set. The large language model 300 uses thought chain analysis technology to break the task down into a set of robot instructions, such as: (1) go_to("job_location"), (2)recognize_line("power_line"), (3)install_fault_indicator("fault_indicator"), (4)check_installation_status("fault_indicator"), (5) go_back("initial_position").

[0041] During task execution, the robot continuously monitors its own working status. Through a built-in self-diagnostic mechanism, the robot can detect any system malfunctions or performance degradation in real time. Once an anomaly occurs, the robot automatically switches to standby mode. The system performs necessary adjustments according to the preset fault handling procedures, enabling it to avoid work interruption through self-repair capabilities when a fault occurs. Furthermore, it can adjust according to environmental and task requirements, as shown below: (1) When the robot executes the go_to(“job_location”) instruction, it needs to perceive the surrounding information in real time, plan according to the map information in the original memory module, find the job location, and execute navigation planning. During the process of moving along the navigation path, it needs to avoid dynamic or static obstacles. (2) Upon reaching the work location, the robot executes recognize_line (“power_line”) according to the task plan. First, it finds the line and then obtains the location knowledge of the line, which can be obtained through an online knowledge base or user instruction. (3) The robot executes the install_fault_indicator (“fault_indicator”) instruction to control the robot arm to install the fault indicator. During the installation process, it needs to avoid other obstacles. (4) After the robot successfully installs the fault indicator, execute check_installation_status(“fault_indicator”) to confirm whether the installation status is correct; (5) After confirming that the installation status is correct, the robot executes go_back (“initial_position”), plans according to the map information in the original memory module, finds the initial position, and executes navigation planning. During the navigation process, it needs to avoid dynamic or static obstacles.

[0042] After the robot completes its task, the system will automatically summarize the work results through a feedback mechanism and perform a self-check on the work area. The robot will return to its initial working position after completing the task and wait for further task instructions. The introduction of a self-checking step after the task is completed enables real-time feedback on the quality and completion status of the task.

[0043] During task execution, the system collects and records the work status in real time, generates detailed reports, and provides feedback to the work supervisor on work progress, completion status, and environmental safety data. The reports include not only the task execution status but also environmental conditions and equipment health status.

[0044] Operators can remotely access the control system to view the robot's working status in real time. The system provides a control interface for the robot's autonomous operation, enabling operators to adjust the task flow or perform emergency operations when needed. Operators can flexibly control the robot in different working environments, adjust the operation strategy in a timely manner, and ensure the smooth completion of tasks.

[0045] Example 2, refer to Figure 2 As an embodiment of the present invention, a method for operating a live-line working robot in a power distribution network is provided, comprising: S1: Receive multimodal input instructions from the operator through the first human-computer interaction module S, parse the semantic information in the instructions through the large language model 300, generate task execution steps based on the context, and adjust the task execution strategy according to the environmental perception data of the second multimodal perception module 200. S2: Based on the task execution strategy generated by the large language model 300, the decision planning module 400 plans the robot's motion path and work sequence in real time, and coordinates the work of each module with the multi-objective optimization algorithm 400c to optimize the task path resource scheduling. S3: The safety monitoring module 500 monitors the safety status of the robot in real time during operation. Combined with the anomaly detection and processing module 600, after receiving the early warning information, the corresponding processing program is initiated to adjust the robot's operation path and perform autonomous obstacle avoidance.

[0046] This embodiment also provides a computer device, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements a personnel positioning safety management visualization analysis system as proposed in the above embodiment.

[0047] This embodiment also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements a personnel positioning safety management visualization analysis system as proposed in the above embodiment.

[0048] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0049] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0050] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0051] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0052] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A live-line working robot system for a power distribution network, characterized in that, include: A first human-computer interaction module (S) used for two-way interaction between the user and the robot by issuing work instructions. and, The second multimodal sensing module (200) includes visual, auditory, and force sensors to perceive the status of target equipment, obstacles, and environmental changes in the working environment in real time. The large language model (300) uses natural language processing to parse the semantic information in the work instructions issued by the first human-computer interaction module (S), generates specific execution steps of the task based on the context, and adjusts the task execution strategy according to the environmental perception data of the second multimodal perception module (200). The decision planning module (400) formulates the robot's motion trajectory, work sequence and collaboration process in real time based on the task execution strategy generated by the large language model (300) and combined with the multi-objective optimization algorithm, and performs task path resource scheduling. The safety monitoring module (500) monitors different safety states during robot operation in real time, and detects potential dangers and triggers warnings based on safety standards and anomaly detection mechanisms. The anomaly detection and processing module (600) receives early warning information from the safety monitoring module (500), and, in conjunction with the second multimodal perception module (200) and the large language model (300), adopts autonomous navigation and path planning technology to initiate relevant processing procedures for the early warning information, allowing the robot to perform autonomous positioning and obstacle avoidance in complex environments.

2. The embodied intelligent power distribution network live-line working robot system as described in claim 1, characterized in that: The first human-computer interaction module (S) includes an instruction input processing unit (S1) and a multi-modal signal conversion unit (S2). The instruction input processing unit (S1) includes features that support users to input work instructions through multiple methods such as voice, text, and images; The multimodal signal conversion unit (S2) includes converting different forms of input operation instructions into a unified machine instruction format for multimodal input.

3. The embodied intelligent power distribution network live-line working robot system as described in claim 1 or 2, characterized in that: The second multimodal perception module (200) includes an environment perception unit (200a) and a data fusion module (200b). The environmental sensing unit (200a) includes visual, auditory, and force sensors; The data fusion module (200b) includes fusing data from sensors in the environmental sensing unit (200a) to generate a comprehensive environmental information map, and constructing a complete dynamic picture of the environment through real-time feedback from the multi-sensor system.

4. The embodied intelligent power distribution network live-line working robot system as described in claim 3, characterized in that: The large language model (300) includes, After receiving voice and text commands transmitted by the operator through the first human-computer interaction module (S), the voice is converted into text by the voice recognition unit (300a), and the grammatical structure of the command is analyzed through text parsing to extract the task target and operation object in the command; The semantic understanding unit (300b) combines contextual information and uses deep learning algorithms to perform semantic analysis on instructions, identify key task information in instructions, and decompose complex tasks into executable sub-tasks based on the semantic parsing results, generating operation steps that the robot can execute. The deep learning algorithm is represented as follows: ， in, Conditional probability refers to the probability given the input text. Visual information and audio information ,Task The probability of being executed. This represents the weight matrix, which is multiplied by the LSTM output. Assume the size of the LSTM output vector is... The size of the weight matrix should be , It is the dimension of the output task space. This represents a feature fusion function that fuses visual and audio information into a joint feature vector. This represents the bias term, which shifts and adjusts the output of the large language model; its dimension is 1.

5. The embodied intelligent power distribution network live-line working robot system as described in any one of claims 1, 2, and 4, characterized in that: The decision planning module (400) includes a path generation unit (400a) and a task coordination unit (400b). The path generation unit (400a) includes planning the robot's motion path and operation sequence in real time based on the task execution strategy generated by the large language model (300); The task coordination unit (400b) includes a multi-objective optimization algorithm (400c) that coordinates the work of different modules in task execution; The multi-objective optimization algorithm (400c) is expressed as follows: ， in, This represents the total cost of the target task after optimization. Indicates task The weighting coefficients, This represents the weighting coefficient of execution time cost in the total cost. This represents the weighting factor of resource consumption in the total cost. This represents the weighting coefficient of environmental adaptability in the total cost. Indicates task Execution time, Indicates task resource consumption, Indicates task Environmental adaptability This indicates the total number of tasks currently in progress.

6. The embodied intelligent power distribution network live-line working robot system as described in claim 5, characterized in that: The security monitoring module (500) includes a real-time monitoring unit (500a) and an anomaly warning unit (500b). The real-time monitoring unit (500a) monitors different safety conditions during the operation process in real time, and, in conjunction with the abnormal early warning unit (500b), promptly detects safety hazards and issues alarms. The abnormality warning unit (500b) includes a program that triggers a soft restart, automatic shutdown, or fault recording procedure when the system detects a sensor fault, motor fault, or power system abnormality. When the robot is unable to communicate with the remote monitoring center or other devices, it attempts to switch to a backup communication channel. If communication fails to be restored, the operation is suspended and manual intervention is required. When the robot detects unidentified obstacles or environmental conditions that do not meet the operational requirements in the working environment, the system will automatically avoid obstacles and adjust operational parameters. If it cannot adapt to changes in the environment, it will pause the task. When the distance between the robot and electrical equipment or obstacles is less than the safety threshold set based on the average historical threshold, the system will immediately stop operation, trigger a safety alarm, and adjust its position to restore a safe operating state. When the robot encounters material shortages or operational failures during task execution, the system will reschedule the task, suspend operations, and await human intervention. When the robot's computing resources and storage space usage exceed the maximum memory capacity of the task, the system will adjust the task priority, clean up useless data, and reallocate resources to restore normal operation. When an algorithm or system malfunctions, the system will restart the relevant modules, record error logs, and switch to a backup control program to continue executing the task if necessary.

7. The embodied intelligent power distribution network live-line working robot system as described in any one of claims 1, 2, 4, and 6, characterized in that: The anomaly detection and processing module (600) includes an autonomous navigation unit (600a). The autonomous navigation unit (600a) combines the real-time early warning information from the safety monitoring module (500) with the environmental data from the second multimodal perception module (200) to adjust the robot's working path in real time through a dynamic path adjustment algorithm. When dealing with emergencies, it can automatically avoid obstacles and safely complete the task.

8. A method for operating a live-line working robot for a power distribution network, comprising the live-line working robot system for a power distribution network as described in any one of claims 1 to 7, characterized in that, include: The first human-computer interaction module (S) receives multimodal input instructions from the operator, parses the semantic information in the instructions through the large language model (300), generates task execution steps based on the context, and adjusts the task execution strategy according to the environmental perception data of the second multimodal perception module (200). Based on the task execution strategy generated by the large language model (300), the decision planning module (400) plans the robot's motion path and work sequence in real time, and coordinates the work of each module in conjunction with the multi-objective optimization algorithm (400c) to optimize the task path resource scheduling. The safety monitoring module (500) monitors the safety status of the robot in real time during operation. Combined with the anomaly detection and processing module (600), after receiving the early warning information, the corresponding processing program is initiated to adjust the robot's operation path and perform autonomous obstacle avoidance.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the embodied intelligent distribution network live-line working robot operating system according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the live-line working robot system for a personal intelligent power distribution network as described in any one of claims 1 to 7.

Images