Electric power operation safety violation identification method based on intelligent device and related device

Abstract

The invention belongs to the field of electric power operation safety supervision, and discloses an intelligent electric power operation safety violation identification method and related device, and the method comprises the steps: collecting the posture, position and physiological data of an operator, a field image, and environment parameters to construct a dynamic scene model; and then a deep neural network model is called as a basis, a violation behavior recognition model with historical and real-time data continuous optimization is combined for recognition, and finally an execution instruction is output to the robot according to a result. A dynamic model is constructed through multi-source data fusion, and an operation scene is comprehensively reflected; the feature extraction and learning ability of the deep neural network is utilized, and a continuously optimized model is combined, so that violation behaviors are accurately identified; and quick response and control of the robot are realized. According to the method, the problems of low efficiency, insufficient identification precision, risk response lag and the like of a traditional supervision mode are effectively solved, safety violation behaviors can be accurately identified by integrating multi-aspect information, the supervision efficiency and precision are improved, potential risks are timely intervened, and safe development of electric power operation is guaranteed.

Description

Technical Field

[0001] The present invention belongs to the technical field of power operation safety supervision, and specifically relates to a power operation safety violation identification method based on embodied intelligence and related devices. Background Art

[0002] Safety supervision is crucial in today's power operations. Traditional safety supervision methods for power operations rely primarily on manual inspections and fixed camera surveillance. Manual inspections, as a relatively basic supervision method, have long been widely used; while fixed camera surveillance, with technological advancements, has gradually become a key auxiliary supervision tool. Together, they ensure the safety of power operations to a certain extent.

[0003] However, traditional approaches to safety supervision of power operations present numerous challenges that urgently need to be addressed. For one thing, supervision efficiency is extremely low. Manual inspections are limited by manpower and time constraints, and their coverage is limited. They struggle to provide real-time, comprehensive monitoring in complex and dangerous operating environments, such as those at height and under live electricity. Existing fixed camera surveillance, due to its fixed location, inevitably has numerous blind spots, making it impossible to dynamically and continuously track workers' behavior. Furthermore, the accuracy of identifying violations is severely insufficient. Existing systems have a significant difficulty identifying relatively hidden violations, such as operating without a ticket or improperly wearing protective equipment. Furthermore, these systems rely heavily on manual experience and lack precise judgment criteria, resulting in a high rate of false positives and failing to meet the growing demand for high-precision supervision. Furthermore, a significant lag in risk response is a prominent issue. Existing systems mostly focus on post-event analysis, failing to promptly identify and intervene in potential risks during the operation, making it difficult to effectively reduce the accident rate. Furthermore, the lack of dynamic risk assessment and graded early warning mechanisms prevents effective risk prevention, making it difficult to ensure the safe conduct of power operations. With the continuous advancement of science and technology, intelligent technology has gradually been introduced into the field of power operation safety supervision. However, most existing intelligent monitoring systems are single-function and can only monitor a specific type of safety risk. For example, they only monitor the wearing of protective equipment or rely solely on fixed cameras to monitor the work area. These systems are unable to adapt to the complex and changing power operation scenarios and struggle to comprehensively consider multiple aspects of information such as the operator's posture, physiological state, environmental factors, and actual operating behavior to accurately identify safety violations.

[0004] It can be seen from this that traditional methods of supervising safety violations in power operations can mostly only monitor a certain type of specific safety risk, cannot adapt to the complex and changeable power operation scenarios, and find it difficult to accurately identify safety violations by integrating multiple aspects of information such as the operator's posture, physiological state, environmental factors, and actual operating behavior. Summary of the Invention

[0005] The present invention provides a method and related device for identifying safety violations in power operations based on embodied intelligence. This method can achieve active adaptation to dynamic scenarios through real-time interaction between the embodied intelligent body and the environment, combined with autonomous perception and decision-making capabilities. It uses intelligent robots, multimodal perception equipment and deep learning algorithms to collaboratively achieve real-time monitoring of operator behavior, dynamic identification of violations and risk warnings, thereby improving the intelligence level of power operation safety management.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions: A method for identifying safety violations in power operations based on embodied intelligence, comprising: Collect workers' posture, position and physiological data, power operation site images and environmental parameters; Build a dynamic power operation scene model based on the operator's posture, position and physiological data, power operation site images and environmental parameters; The newly optimized violation behavior recognition model is used to identify the dynamic power operation scenario model to obtain the power operation safety violation identification results. The basic model of the violation behavior recognition model is a deep neural network model, which is constructed based on historical violation data and real-time operation scenario data. The violation behavior recognition model is continuously optimized based on each dynamic power operation scenario model to be identified and the corresponding power operation safety violation identification results. Based on the results of the identification of safety violations in power operations, identification result execution instructions are output to the robot, so that the robot can perform safety management and feedback operations according to the result execution instructions.

[0007] Furthermore, after calling the latest optimized violation behavior recognition model to identify the dynamic power operation scenario model and obtaining the power operation safety violation identification result, the method further includes: Based on the results of the identification of safety violations in power operations, a risk coefficient is evaluated; the risk coefficient is used to trigger a graded warning.

[0008] Furthermore, the risk factor is evaluated based on the results of the power operation safety violation identification, including: The risk coefficient is calculated based on the comprehensive analysis of the severity of the power operation safety violation identification results and the environmental parameters.

[0009] Furthermore, after the risk coefficient is obtained through evaluation based on the results of the safety violation identification of the power operation, the method further includes: Trigger graded warnings based on the current risk factor, including voice reminders, remote braking, and emergency intervention. When the risk factor exceeds the preset threshold, an emergency intervention instruction is output for the robot to perform emergency intervention operations.

[0010] Furthermore, the power operation site image includes an operator's behavior image; the environmental parameters include wind speed, temperature, humidity and air pressure; and the physiological data includes the operator's heart rate data.

[0011] Furthermore, the dynamic power operation scenario model is identified by calling the latest optimized violation behavior identification model to obtain the power operation safety violation identification result, including: The latest optimized violation behavior recognition model is used to identify the dynamic power operation scenario model. The recognition process includes: Use the YOLOv7 algorithm to identify whether workers are wearing protective equipment; Use the Transformer algorithm to identify whether workers accidentally touch live equipment or perform other illegal operations; After the identification is completed, the results of the power operation safety violation identification are obtained.

[0012] Furthermore, the identification process further includes: Determine whether the operator's physiological state is abnormal based on the operator's physiological data; According to the operator's posture, it is judged whether the operator's posture is abnormal. If the posture is judged to be abnormal, the locking function is automatically activated to prevent falling from a height.

[0013] Furthermore, before collecting the posture, position and physiological data of the operator, the power operation site image and environmental parameters, the following steps are also included: Verify the relevance between work orders and plan orders to identify planned work risks; Based on the collected environmental parameters, determine whether the working environment meets the working safety conditions.

[0014] A power operation safety violation identification system based on embodied intelligence, comprising: The acquisition module is used to collect the posture, position and physiological data of workers, images of the power operation site and environmental parameters; A scenario construction module is used to construct a dynamic power operation scenario model based on the operator's posture, position and physiological data, power operation site images and environmental parameters; An identification module is configured to call a newly optimized violation behavior identification model to identify a dynamic power operation scenario model and obtain a power operation safety violation identification result; wherein the violation behavior identification model is based on a deep neural network model constructed based on historical violation data and real-time operation scenario data, and the violation behavior identification model is continuously optimized based on each dynamic power operation scenario model to be identified and the corresponding power operation safety violation identification result; The instruction output module is used to output the identification result execution instruction to the robot based on the identification result of the power operation safety violation, so that the robot can perform safety control and feedback operations according to the result execution instruction.

[0015] A robot, comprising a robot body; the robot body being equipped with a computer program; the computer program, when executed by a processor, being used to implement the steps of the above-mentioned method for identifying safety violations in power operations based on embodied intelligence; The robot body is provided with a mobile chassis unit for autonomous navigation at the power operation site and supports multi-terrain movement; The robot body is provided with a voice interaction unit for voice communication with the operator and providing real-time violation reminders.

[0016] Compared with the prior art, the present invention has the following beneficial effects: The present invention provides a method for identifying safety violations in power operations based on embodied intelligence. First, the posture, position, physiological data of the operator, as well as on-site images and environmental parameters, are collected to construct a dynamic scene model. Then, a violation behavior recognition model based on a deep neural network model and continuously optimized in combination with historical and real-time data is called for identification. Finally, execution instructions are output to the robot based on the results. This method constructs a dynamic model through multi-source data fusion to comprehensively reflect the operation scene; utilizes the powerful feature extraction and learning capabilities of the deep neural network, combined with the continuously optimized model, to accurately identify violations; the robot executes instructions to achieve rapid response and control. This method effectively solves the problems of low efficiency, insufficient recognition accuracy, and delayed risk response of traditional supervision methods. It can accurately identify safety violations by integrating information from multiple aspects, improve supervision efficiency and accuracy, intervene in potential risks in a timely manner, and ensure the safe conduct of power operations.

[0017] Preferably, in the present invention, the risk factor is assessed based on the violation identification results, and a graded warning is triggered according to the risk factor, which helps to timely discover and intervene in potential risks, reduce the accident rate, and ensure the safe implementation of power operations.

[0018] Preferably, in the present invention, the risk coefficient is calculated by comprehensively analyzing the severity of the violation identification results and environmental parameters, so that the risk assessment is more accurate and comprehensive, providing strong support for subsequent risk warning and intervention.

[0019] Preferably, in the present invention, graded warnings are triggered according to the current risk factor, including voice reminders, remote braking, and emergency intervention measures. When the risk factor exceeds a preset threshold, emergency intervention operations are performed, which effectively reduces the possibility of accidents and ensures the safety of operators.

[0020] Preferably, in the present invention, the power operation site images include behavioral images of the operators, environmental parameters include wind speed, temperature, humidity and air pressure, and physiological data include heart rate data of the operators, which provide a clear direction and basis for subsequent data collection and processing.

[0021] Preferably, in the present invention, the YOLOv7 algorithm is used to identify whether the operator is wearing protective equipment, and the Transformer algorithm is used to identify whether the operator accidentally touches live equipment or performs other illegal operations, thereby improving the accuracy and efficiency of illegal behavior identification.

[0022] Preferably, in the present invention, whether the operator's state is abnormal is determined based on the operator's physiological data and posture. If the posture is determined to be abnormal, the locking function is automatically activated to prevent accidents such as falling from a height, thereby further ensuring the safety of the operator.

[0023] Preferably, in the present invention, the correlation between the work order and the plan order is verified before data is collected to identify the risks of the planned operation; at the same time, whether the working environment meets the working safety conditions is judged based on the collected environmental parameters, which helps to discover and solve potential safety problems in advance and ensure the smooth progress of the operation.

[0024] The present invention also provides a robot, including a robot body, which autonomously navigates at the power operation site through a mobile chassis unit to expand the scope of supervision; a computer program executes a violation identification method based on embodied intelligence, integrates the operator's posture, position, physiological data, on-site images and environmental parameters to construct a dynamic operation scene model, and uses a deep neural network model for identification. The robot's autonomous movement overcomes the limitations of manual inspections and fixed cameras, and realizes real-time and comprehensive monitoring of complex scenes; the deep neural network model improves the accuracy of violation identification and reduces the misjudgment rate. This robot can accurately identify safety violations in power operations in real time, promptly remind operators, and promptly discover and intervene in potential risks during the operation process. It prevents risks in advance through dynamic risk assessment and graded early warning mechanisms, effectively ensuring the safe conduct of power operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 This is a diagram showing the overall architecture of a power operation safety violation identification system based on embodied intelligence provided by an embodiment of the present invention; Figure 2 A schematic diagram of an embodied intelligent robot module of an embodied intelligence-based power operation safety violation identification system provided by an embodiment of the present invention; Figure 3 A schematic diagram of the smart wearable device module of the power operation safety violation identification system based on embodied intelligence provided by an embodiment of the present invention; Figure 4Schematic diagram and workflow diagram of a multimodal data fusion module of an embodied intelligence-based power operation safety violation identification system provided by an embodiment of the present invention; wherein (a) is a structural diagram; (b) is a flowchart; Figure 5 Schematic diagram of the violation behavior identification module and workflow diagram of the power operation safety violation identification system based on embodied intelligence provided by an embodiment of the present invention; wherein (a) is a structural diagram; (b) is a flowchart; Figure 6 Schematic diagram and workflow diagram of a dynamic risk assessment module of a power operation safety violation identification system based on embodied intelligence provided by an embodiment of the present invention; wherein (a) is a structural diagram; (b) is a flowchart; Figure 7 Schematic diagram and workflow diagram of the self-learning optimization module of the power operation safety violation identification system based on embodied intelligence provided by an embodiment of the present invention; wherein (a) is a structural diagram; (b) is a flowchart; Figure 8 A flowchart of a method for identifying safety violations in power operations based on embodied intelligence provided by an embodiment of the present invention; Figure 9 A flow chart of the method for identifying safety violations in power operations based on embodied intelligence provided by the present invention; Figure 10 This is a structural diagram of a power operation safety violation identification system based on embodied intelligence provided by the present invention.

[0026] Reference numerals: 1. Visual sensor; 2. Voice interaction unit; 3. Environmental sensor; 4. Robotic arm; 5. Mobile chassis unit. DETAILED DESCRIPTION

[0027] In order to further understand the content of the present invention, the present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the embodiments are only for explaining the present invention and are not intended to limit it.

[0028] This embodiment provides a method for identifying safety violations in power operations based on embodied intelligence, which is applied to a control terminal and includes: Collect workers' posture, position and physiological data, power operation site images and environmental parameters; Build a dynamic power operation scene model based on the operator's posture, position and physiological data, power operation site images and environmental parameters; The newly optimized violation behavior recognition model is used to identify the dynamic power operation scenario model to obtain the power operation safety violation identification results. The basic model of the violation behavior recognition model is a deep neural network model, which is constructed based on historical violation data and real-time operation scenario data. The violation behavior recognition model is continuously optimized based on each dynamic power operation scenario model to be identified and the corresponding power operation safety violation identification results. Based on the results of the power operation safety violation identification, the robot is given an execution instruction to execute the identification result, so that the robot can perform safety management and feedback operations according to the result execution instruction. Among them, safety management and feedback operations specifically include early warning, emergency intervention, and data receipt.

[0029] In this embodiment, the method for identifying safety violations in power operations based on embodied intelligence is mainly applied to the control end, and the entire system architecture also includes a collection terminal and a robot; preferably, the control end can be integrated on the robot or be an independent processor.

[0030] The operator's posture, position and physiological data are collected by the collection terminal, which is preferably a smart wearable device; the power operation site images and environmental parameters are collected by the visual sensor and environmental sensor integrated on the robot; The acquisition terminal and robot transmit the collected data to the control terminal, which builds a dynamic power operation scenario model based on the real-time data. The dynamic power operation scenario model is then transmitted to the newly optimized violation behavior recognition model for recognition. After each recognition, the control terminal updates the current violation behavior recognition model based on the dynamic power operation scenario model to be recognized and the output power operation safety violation identification results, achieving continuous optimization and ensuring higher recognition accuracy. Finally, the control terminal generates an execution instruction based on the identification results of power operation safety violations, and controls the robot to perform corresponding actions, such as early warning, emergency intervention, data receipt and other safety control and feedback operations; of course, if the control terminal is integrated with the robot, the control terminal directly transmits the information to the execution terminal of the robot, such as controlling the robot's voice interaction unit to issue an early warning instruction; controlling the robot's mobile chassis unit to achieve displacement, and controlling the robot's robotic arm to perform on-site forced intervention (operating the emergency brake button or cutting off the power supply, etc.); controlling the robot's data receipt unit to return the collected relevant data and the output power operation safety violation identification results and other data to the back-end scheduling management platform.

[0031] For example, the method for identifying safety violations in power operations based on embodied intelligence provided in this embodiment may be implemented in more specific steps as follows: S1. Data collection steps Collect workers' posture, position, and physiological data, as well as images and environmental parameters of the power operation site. Specifically, smart wearable devices are used to collect workers' posture, position, and physiological data in real time. The smart wearable device module includes a seat belt status detection unit, a helmet posture detection unit, and a heart rate monitoring unit, which can accurately obtain real-time status information of workers. At the same time, the visual sensors in the embodied intelligent robot module collect images of workers' behavior, and environmental sensors collect working environment parameters, including wind speed, temperature, humidity, and air pressure, providing a comprehensive data foundation for subsequent accurate identification.

[0032] S2. Steps for building a dynamic operation scenario model A dynamic power operation scenario model is constructed based on the operator's posture, position, and physiological data, as well as images and environmental parameters of the power operation site. The multimodal data fusion module integrates data collected from the smart wearable device module and the embodied intelligent robot module, unifying data from different modalities into a single model framework. This model fully considers the dynamic changes in factors such as personnel, equipment, and the environment at the operation site, ensuring that the constructed dynamic operation scenario model truly reflects the real-time conditions of the operation site.

[0033] S3. Identification steps for illegal behavior The dynamic power operation scenario model is identified by calling the latest optimized violation behavior recognition model to obtain the power operation safety violation identification results. The basic model of the violation behavior recognition model is a deep neural network model, which is trained based on historical violation data and real-time operation scenario data. During the recognition process, a combination of multiple advanced algorithms is used: S31. Use the YOLOv7 algorithm to identify whether workers are wearing protective equipment. This algorithm has the characteristics of high precision and rapid detection. It can accurately locate and determine whether workers are correctly wearing protective equipment such as insulating gloves and safety belts in complex work scene images.

[0034] S32. The Transformer algorithm is used to identify whether the operator accidentally touches live equipment or performs other illegal operations. The Transformer algorithm is good at processing sequence data and can effectively capture the dynamic changes in the operator's operating behavior and accurately determine whether there is any illegal operation behavior.

[0035] S33. The system also determines whether the worker's physiological state is abnormal based on the worker's physiological data and the worker's posture. If the posture is abnormal, the system automatically activates the locking function to prevent a fall. For example, if the heart rate monitoring unit detects that the worker's heart rate exceeds 120 beats per minute for a sustained period, it is marked as "abnormal physiological state." If the smart seat belt detects that the worker's tilt angle exceeds 30 degrees, the anti-fall locking unit automatically activates.

[0036] S4. Risk assessment and early warning steps S41. Based on the results of the identification of safety violations in power operations, a risk coefficient is evaluated; the risk coefficient is used to trigger a graded warning. Specifically, the risk coefficient is calculated based on the comprehensive analysis of the severity of the results of the identification of safety violations in power operations and the environmental parameters. For example, for relatively low-risk violations such as not wearing insulating gloves, a lower risk coefficient is given in combination with the environmental parameters at the time, such as normal wind speed, suitable temperature and humidity; while for high-risk violations such as accidentally touching live equipment, especially in harsh environments such as wind speed exceeding the threshold, high humidity and easy conductivity, an extremely high risk coefficient is given.

[0037] S42. Trigger a tiered warning based on the current risk factor, including voice reminders, remote braking, and emergency intervention. When the risk factor exceeds a preset threshold, emergency intervention is performed. For low-risk violations, the system sends a voice reminder through the helmet's built-in headphones, requiring the operator to immediately correct the problem. For high-risk violations, the embodied intelligent robot activates its robotic arm to forcibly isolate the source of danger and notifies the supervisor through the management platform.

[0038] Exemplarily, this embodiment provides a method for identifying safety violations in power operations based on embodied intelligence, further comprising: S5. Pre-operation verification steps Before collecting the posture, position and physiological data of the workers, images of the power operation site and environmental parameters, the system also includes: verifying the correlation between the work order and the plan order to identify the risks of the planned operation; judging whether the working environment meets the safety conditions for the operation based on the collected environmental parameters. The system automatically verifies the correlation between the work order held by the worker and the plan order. If the work order is missing or inconsistent with the plan order, the system marks it as "no planned operation risk" and sends an early warning message to the supervisor through the management platform. At the same time, the environmental risk assessment unit determines whether high-altitude operation or live operation is allowed based on data such as wind speed, temperature and humidity. If the environmental parameters exceed the safe range, the system triggers the "environmental risk warning" and reminds the worker to stop the operation through the voice interaction unit.

[0039] S6, self-learning optimization steps This method uses a self-learning optimization module to update the algorithm parameters of the violation identification module based on reinforcement learning to adapt to different power operation scenarios. Specifically, this includes optimizing the algorithm parameters through reinforcement learning based on historical violation data and real-time operation scenario data, and regularly updating the model to adapt to new operation scenarios and violation patterns. For example, after multiple transmission line maintenance operations and substation equipment maintenance operations, the system accumulates violation data and normal operation data from different scenarios. The self-learning optimization module uses this data to optimize the violation identification model, continuously improving the model's recognition accuracy in subsequent similar operation scenarios.

[0040] like Figure 1 As shown, this embodiment also provides an electric power operation safety violation identification system based on embodied intelligence, which specifically includes: a perception and execution layer, a data processing layer, and a decision and feedback layer; wherein, the perception and execution layer includes an embodied intelligent robot module and an intelligent wearable device module, the data processing layer includes a multimodal data fusion module and a violation behavior identification module, and the decision and feedback layer includes a dynamic risk assessment module and a self-learning optimization module.

[0041] like Figure 2 As shown, an embodied intelligent robot module is used to inspect and monitor power operation sites. The embodied intelligent robot module includes a visual sensor 1, an environmental sensor 3, and a robotic arm 4. The visual sensor 1 is used to collect images of operator behavior, the environmental sensor 3 is used to collect operating environment parameters, and the robotic arm 4 is used to perform emergency intervention operations. The module also includes: Voice interaction unit 2, used for voice communication with operators and providing real-time violation reminders; The mobile chassis unit 5 is used for autonomous navigation at the power operation site and supports multi-terrain movement.

[0042] like Figure 3 As shown, the smart wearable device module is used to collect the posture, position and physiological data of the operator in real time. The smart wearable device module includes a seat belt status detection unit and a helmet posture detection unit; and also includes: Heart rate monitoring unit, used to collect the operator's heart rate data in real time to determine whether their physiological state is abnormal; The anti-fall locking unit is used to automatically activate the locking function when an abnormal posture of the operator is detected to prevent falling from a height.

[0043] The data processing layer includes a multimodal data fusion module and a violation behavior recognition module.

[0044] like Figure 4 As shown, the multimodal data fusion module is used to integrate the data collected by the embodied intelligent robot module and the intelligent wearable device module to build a dynamic operation scene model; it also includes: The work order compliance analysis unit is used to verify the relevance between work orders and planned orders and identify the risk of unplanned operations. The environmental risk assessment unit is used to determine whether the working environment meets safety conditions based on environmental parameters (such as wind speed, temperature and humidity).

[0045] like Figure 5 As shown, the violation behavior recognition module analyzes the dynamic operation scene model based on a deep learning algorithm to identify the operator's violation behavior; and also includes: Ticketless operation identification unit, used to identify whether the operator holds a valid work order; Protective equipment detection unit, used to identify whether workers are wearing protective equipment such as insulating gloves and safety belts in accordance with regulations; The illegal operation identification unit is used to identify whether the operator accidentally touches the live equipment or performs other illegal operations.

[0046] The decision-making and feedback layer includes a dynamic risk assessment module and a self-learning optimization module.

[0047] like Figure 6 As shown, the dynamic risk assessment module is used to calculate the risk coefficient based on the output results of the violation behavior identification module and trigger a graded warning; it also includes: A graded warning unit, which triggers different levels of warnings based on risk factors, including voice reminders, remote braking, and emergency intervention; The event recording unit is used to automatically generate violation event reports and upload them to the management platform.

[0048] like Figure 7 As shown, the self-learning optimization module is used to update the algorithm parameters of the violation behavior recognition module through reinforcement learning to adapt to different power operation scenarios; and also includes: Scenario adaptation unit, used to optimize algorithm parameters according to different power operation scenarios (such as substations and transmission lines); The historical data learning unit is used to optimize the model generalization capability based on historical violation data.

[0049] like Figure 8 As shown, based on the above-mentioned embodied intelligence-based power operation safety violation identification system, the specific implementation steps are as follows: a. Using the embodied intelligent robot module to collect images and environmental parameters of the power operation site, specifically using visual sensors to collect images of worker behavior and environmental sensors to collect operating environment parameters, including wind speed, temperature, humidity, and air pressure; b. Collect the operator's posture, position and physiological data through the smart wearable device module; c. Integrate the image, environmental parameters, posture, position and physiological data through a multimodal data fusion module to build a dynamic operation scene model; d. Analyze the dynamic operation scenario model using a deep learning algorithm through the violation behavior recognition module to identify the operator's violation behavior, including: using the YOLOv7 algorithm to identify whether the operator is wearing insulating gloves; using the Transformer algorithm to identify whether the operator accidentally touches live equipment; e. Calculate the risk factor based on the output of the violation identification module through the dynamic risk assessment module and trigger a graded warning. Specifically, the risk factor is calculated based on a comprehensive analysis of the severity of the violation and environmental parameters; when the risk factor exceeds a preset threshold, the robotic arm is triggered to perform an emergency intervention operation; f. Update the algorithm parameters of the violation behavior identification module based on reinforcement learning through the self-learning optimization module to adapt to different power operation scenarios, specifically including: optimizing the algorithm parameters through reinforcement learning based on historical violation data and real-time operation scenario data; and regularly updating the model to adapt to new operation scenarios and violation behavior patterns.

[0050] Exemplarily, this embodiment also provides a robot, including a robot body, which is equipped with a computer program, and when the computer program is executed by a processor, it is used to implement the steps of the above-mentioned method for identifying safety violations in power operations based on embodied intelligence; wherein, a mobile chassis unit is provided at the bottom of the robot body, which is used for autonomous navigation at the power operation site and supports multi-terrain movement; and a voice interaction unit is also provided on the robot body, which is used for voice communication with the operator and providing real-time violation reminders.

[0051] The method for identifying safety violations in power operations based on embodied intelligence provided in this embodiment is specifically implemented as follows in combination with actual scenarios: Example 1: Safety Violation Identification and Intervention in Transmission Line Maintenance Scenario Scenario Description: During power transmission line maintenance operations, workers must perform equipment maintenance at high altitudes, exposing themselves to risks such as falling from heights and accidentally touching live equipment. This embodiment utilizes embodied intelligent robots, smart wearable devices, and multimodal data fusion technology to achieve real-time monitoring of worker behavior and dynamic intervention in violations.

[0052] The specific operation methods and steps are as follows: Pre-operation preparation stage Step 1.1: Work Order Compliance Verification The system automatically verifies the relevance of the operator's work order to the planned order. If a work order is missing or inconsistent with the planned order, the system marks it as "unplanned operation risk" and sends an early warning to the supervisor via the management platform.

[0053] Step 1.2: Protective equipment testing Workers wear smart safety belts and helmets. The system uses the safety belt status detection unit to check whether the buckle is secure, and the helmet posture detection unit to check whether the helmet is worn correctly. If the system detects that the protective equipment is not worn correctly, the system prohibits the worker from entering and reminds them to correct the problem through voice prompts.

[0054] Monitoring phase during operation Step 2.1: Embodied Intelligent Robot Inspection The embodied intelligent robot moves along the transmission line, collecting images of workers' behavior through visual sensors and environmental parameters such as wind speed, temperature and humidity through environmental sensors.

[0055] The robot uploads data to the multimodal data fusion module in real time to build a dynamic operation scene model.

[0056] Step 2.2: Identify traffic violations The violation behavior recognition module analyzes image data based on the YOLOv7 algorithm to identify whether the operator is wearing insulating gloves or accidentally touching live equipment.

[0057] If it is detected that the person is not wearing insulating gloves or accidentally touches live equipment, the system will mark it as a "high-risk violation."

[0058] Step 2.3: Environmental Risk Assessment The environmental risk assessment unit uses wind speed data to determine whether high-altitude work is permitted. If the wind speed exceeds a safety threshold (e.g., 10 m / s), the system triggers an "environmental risk warning" and uses the voice interaction unit to remind the operator to stop the work.

[0059] Step 2.4: Physiological status monitoring The heart rate monitoring unit of the smart wearable device collects the operator's heart rate data in real time. If an abnormal heart rate is detected (such as a sustained rate exceeding 120 beats per minute), the system marks it as "abnormal physiological state" and recommends that the operator rest.

[0060] Risk management stage Step 3.1: Graded warning For low-risk violations (such as not wearing insulating gloves), the system sends voice reminders through the helmet's built-in headphones, requiring the operator to make immediate corrections.

[0061] For high-risk violations (such as accidentally touching live equipment), the system triggers an "emergency intervention" command, the embodied intelligent robot activates the robotic arm to forcibly isolate the source of danger, and notifies the supervisor through the management platform.

[0062] Step 3.2: Anti-fall lock If the smart safety belt detects that the operator's posture is abnormal (such as a tilt angle exceeding 30 degrees), the anti-fall locking unit will automatically activate to prevent falling from a height.

[0063] Step 3.3: Event Recording and Report Generation The system automatically records the time, location, type and handling results of the violation, generates a violation incident report and uploads it to the management platform for subsequent analysis and training.

[0064] Post-job optimization phase Step 4.1: Data upload and analysis The system uploads the monitoring data of this operation to the cloud and optimizes the parameters of the violation behavior identification model through big data analysis.

[0065] Step 4.2: Model update and adaptation The self-learning optimization module is based on a reinforcement learning algorithm. It combines historical violation data with real-time operation scenario data to update model parameters and improve its adaptability to different operation scenarios.

[0066] Example 2: Safety Violation Identification and Intervention in Substation Equipment Maintenance Scenario Scenario Description: During substation equipment maintenance, operators must operate high-voltage equipment in a live environment, posing risks such as electric shock and equipment misoperation. This embodiment utilizes embodied intelligent robots, smart wearable devices, and multimodal data fusion technology to achieve real-time monitoring of operator behavior and dynamic intervention in violations.

[0067] The specific operation methods and steps are as follows: Pre-operation preparation stage Step 1.1: Work Order Compliance Verification The system automatically verifies the relevance of the operator's work order to the planned order. If a work order is missing or inconsistent with the planned order, the system marks it as "unplanned operation risk" and sends an early warning to the supervisor via the management platform.

[0068] Step 1.2: Protective equipment testing Workers wear smart safety belts and helmets. The system uses the safety belt status detection unit to check whether the buckle is secure, and the helmet posture detection unit to check whether the helmet is worn correctly. If the system detects that the protective equipment is not worn correctly, the system prohibits the worker from entering and reminds them to correct the problem through voice prompts.

[0069] Monitoring phase during operation Step 2.1: Embodied Intelligent Robot Inspection The embodied intelligent robot moves within the substation, collecting images of workers' behavior through visual sensors and environmental parameters such as temperature, humidity, and air pressure through environmental sensors.

[0070] The robot uploads data to the multimodal data fusion module in real time to build a dynamic operation scene model.

[0071] Step 2.2: Identify traffic violations The violation behavior recognition module analyzes image data based on the Transformer algorithm to identify whether the operator accidentally touches live equipment or performs operations without a ticket.

[0072] If the system detects accidental contact with live equipment or operation without a ticket, it will be marked as a "high-risk violation."

[0073] Step 2.3: Environmental Risk Assessment The environmental risk assessment unit uses temperature and humidity data to determine whether live work is permitted. If environmental parameters exceed safe ranges, the system triggers an "environmental risk warning" and uses the voice interaction unit to remind operators to stop work.

[0074] Risk management stage Step 3.1: Graded warning For low-risk violations (such as not wearing insulating gloves), the system sends voice reminders through the helmet's built-in headphones, requiring the operator to make immediate corrections.

[0075] For high-risk violations (such as accidentally touching live equipment), the system triggers an "emergency intervention" command, the embodied intelligent robot activates the robotic arm to forcibly isolate the source of danger, and notifies the supervisor through the management platform.

[0076] Step 3.2: Event Recording and Report Generation The system automatically records the time, location, type and handling results of the violation, generates a violation incident report and uploads it to the management platform for subsequent analysis and training.

[0077] Post-job optimization phase Step 4.1: Data upload and analysis The system uploads the monitoring data of this operation to the cloud and optimizes the parameters of the violation behavior identification model through big data analysis.

[0078] Step 4.2: Model update and adaptation The self-learning optimization module is based on a reinforcement learning algorithm. It combines historical violation data with real-time operation scenario data to update model parameters and improve its adaptability to different operation scenarios.

[0079] For example, Figure 9 As shown, this embodiment provides a method for identifying safety violations in power operations based on embodied intelligence, including: Collect workers' posture, position and physiological data, power operation site images and environmental parameters; Build a dynamic power operation scene model based on the operator's posture, position and physiological data, power operation site images and environmental parameters; The newly optimized violation behavior recognition model is used to identify the dynamic power operation scenario model to obtain the power operation safety violation identification results. The basic model of the violation behavior recognition model is a deep neural network model, which is constructed based on historical violation data and real-time operation scenario data. The violation behavior recognition model is continuously optimized based on each dynamic power operation scenario model to be identified and the corresponding power operation safety violation identification results. Based on the results of the identification of safety violations in power operations, identification result execution instructions are output to the robot, so that the robot can perform safety management and feedback operations according to the result execution instructions.

[0080] In this embodiment, after calling the latest optimized violation behavior recognition model to identify the dynamic power operation scenario model and obtaining the power operation safety violation identification result, the method further includes: Based on the results of the identification of safety violations in power operations, a risk coefficient is evaluated; the risk coefficient is used to trigger a graded warning.

[0081] In this embodiment, the risk factor is evaluated based on the results of the power operation safety violation identification, including: The risk coefficient is calculated based on the comprehensive analysis of the severity of the power operation safety violation identification results and the environmental parameters.

[0082] In this embodiment, after the risk coefficient is obtained through evaluation based on the results of the safety violation identification of the power operation, the following steps are further included: Trigger graded warnings based on the current risk factor, including voice reminders, remote braking, and emergency intervention. When the risk factor exceeds the preset threshold, an emergency intervention instruction is output for the robot to perform emergency intervention operations.

[0083] In this embodiment, the power operation site image includes the operator's behavior image; the environmental parameters include wind speed, temperature, humidity and air pressure; and the physiological data includes the operator's heart rate data.

[0084] In this embodiment, the dynamic power operation scenario model is identified by calling the latest optimized violation behavior identification model to obtain the power operation safety violation identification result, including: The latest optimized violation behavior recognition model is used to identify the dynamic power operation scenario model. The recognition process includes: Use the YOLOv7 algorithm to identify whether workers are wearing protective equipment; Use the Transformer algorithm to identify whether workers accidentally touch live equipment or perform other illegal operations; After the identification is completed, the results of the power operation safety violation identification are obtained.

[0085] In this embodiment, the identification process further includes: Determine whether the operator's physiological state is abnormal based on the operator's physiological data; According to the operator's posture, it is judged whether the operator's posture is abnormal. If the posture is judged to be abnormal, the locking function is automatically activated to prevent falling from a height.

[0086] In this embodiment, before collecting the posture, position and physiological data of the operator, the power operation site image and environmental parameters, the following steps are also included: Verify the relevance between work orders and plan orders to identify planned work risks; Based on the collected environmental parameters, determine whether the working environment meets the working safety conditions.

[0087] like Figure 10 As shown, this embodiment also provides an electric power operation safety violation identification system based on embodied intelligence, including: an acquisition module for collecting the posture, position and physiological data of the operator, the electric power operation site image and environmental parameters; a scene construction module for constructing a dynamic electric power operation scene model based on the posture, position and physiological data of the operator, the electric power operation site image and environmental parameters; an identification module for calling the latest optimized violation behavior identification model to identify the dynamic electric power operation scene model and obtain the electric power operation safety violation identification result; wherein, the basic model of the violation behavior identification model is a deep neural network model, which is constructed based on historical violation data and real-time operation scene data, and the violation behavior identification model is continuously optimized according to each dynamic electric power operation scene model to be identified and the corresponding electric power operation safety violation identification result; an instruction output module for outputting the identification result execution instruction to the robot based on the electric power operation safety violation identification result, so that the robot can execute the instruction according to the result to perform safety control and feedback operations.

[0088] The present invention also provides an electronic device, comprising: a memory for storing a computer program; and a processor for implementing the steps of the method for identifying safety violations in power operations based on embodied intelligence when executing the computer program.

[0089] When the processor executes the computer program, it implements the above-mentioned steps of identifying power operation safety violations based on embodied intelligence, for example: collecting the posture, position and physiological data of the workers, images of the power operation site and environmental parameters; constructing a dynamic power operation scene model based on the posture, position and physiological data of the workers, images of the power operation site and environmental parameters; calling the latest optimized violation behavior recognition model to identify the dynamic power operation scene model to obtain the power operation safety violation identification result; wherein, the basic model of the violation behavior recognition model is a deep neural network model, which is constructed based on historical violation data and real-time operation scene data, and the violation behavior recognition model is continuously optimized according to each dynamic power operation scene model to be identified and the corresponding power operation safety violation identification result; based on the power operation safety violation identification result, an identification result execution instruction is output to the robot, so that the robot performs safety control and feedback operations according to the result execution instruction.

[0090] For example, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to implement the present invention. The one or more modules / units may be a series of computer program instruction segments capable of performing preset functions, and the instruction segments are used to describe the execution process of the computer program in the embodied intelligence-based power operation safety violation identification device. For example, the computer program can be divided into an acquisition module, a scene construction module, an identification module and an instruction output module; the specific functions of each module are as follows: an acquisition module, used to collect the posture, position and physiological data of the workers, the images of the power operation site and the environmental parameters; a scene construction module, used to construct a dynamic power operation scene model based on the posture, position and physiological data of the workers, the images of the power operation site and the environmental parameters; an identification module, used to call the latest optimized violation behavior recognition model to identify the dynamic power operation scene model and obtain the power operation safety violation identification result; wherein, the basic model of the violation behavior recognition model is a deep neural network model, which is constructed based on historical violation data and real-time operation scene data, and the violation behavior recognition model is continuously optimized according to each dynamic power operation scene model to be identified and the corresponding power operation safety violation identification result; an instruction output module, used to output the identification result execution instruction to the robot based on the power operation safety violation identification result, so that the robot can execute the instruction according to the result to perform safety control and feedback operations.

[0091] The power operation safety violation identification device based on embodied intelligence can be a computing device such as a desktop computer, a notebook, a PDA, and a cloud server. The power operation safety violation identification device based on embodied intelligence can include, but is not limited to, a processor and a memory. Those skilled in the art will understand that the above examples of power operation safety violation identification devices based on embodied intelligence do not constitute a limitation on power operation safety violation identification devices based on embodied intelligence, and can include more components than the above, or a combination of certain components, or different components. For example, the power operation safety violation identification device based on embodied intelligence can also include input and output devices, network access devices, buses, etc.

[0092] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor, or the processor may be any conventional processor, etc. The processor serves as the control center of the embodied intelligence-based power operation safety violation identification system, and utilizes various interfaces and lines to connect the various components of the embodied intelligence-based power operation safety violation identification system.

[0093] The memory can be used to store the computer program and / or module, and the processor realizes the various functions of the power operation safety violation identification device based on embodied intelligence by running or executing the computer program and / or module stored in the memory, and calling the data stored in the memory.

[0094] The memory may primarily include a program storage area and a data storage area. The program storage area may store an operating system and at least one application required for a function (such as sound playback or image playback); the data storage area may store data generated based on the use of the mobile phone (such as audio data and a phone book). Furthermore, the memory may include high-speed random access memory (RAM) and non-volatile memory, such as a hard disk, internal memory, a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) card, a flash card, at least one disk storage device, a flash memory device, or other volatile solid-state storage device.

[0095] The present invention also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the steps of the method for identifying safety violations in power operations based on embodied intelligence.

[0096] If the modules / units integrated in the power operation safety violation identification system based on embodied intelligence are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium.

[0097] Based on this understanding, the present invention implements all or part of the process of the above-mentioned method for identifying safety violations in power operations based on embodied intelligence, and can also be accomplished by using a computer program to instruct relevant hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, the computer program can implement the steps of the above-mentioned method for identifying safety violations in power operations based on embodied intelligence. The computer program includes computer program code, which can be in source code form, object code form, executable file, or a pre-defined intermediate form.

[0098] The computer-readable storage medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc.

[0099] It should be noted that the content contained in the computer-readable storage medium can be appropriately increased or decreased according to the requirements of legislation and patent practices in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practices, computer-readable storage media do not include electrical carrier signals and telecommunication signals.

[0100] In summary, the present invention provides a method and related device for identifying safety violations in power operations based on embodied intelligence. The core innovations are: First, collaborative operation of embodied intelligent bodies: robots not only serve as monitoring terminals, but can also perform emergency braking (such as cutting off dangerous power supplies) and assist personnel in evacuating, thus realizing a closed loop of "perception-decision-execution".

[0101] Second, dynamic fusion of multimodal data: Combining image recognition, work order compliance analysis (such as unplanned work detection) and environmental parameters, the accuracy of violation identification is improved to over 98%.

[0102] Third, edge-cloud collaborative computing: low-latency preliminary judgment is achieved on the wearable device side, and big data analysis and model iteration are performed on the cloud.

[0103] Compared with existing operation safety supervision methods, it has the following advantages: First, real-time: Through the collaborative operation of embodied intelligent robots and smart wearable devices, real-time monitoring and intervention of violations can be achieved.

[0104] Second, safety: Through real-time intervention, the incidence of accidents such as falling from heights and electric shock can be reduced (experimental data shows a 60% reduction).

[0105] Third, precision: Based on multimodal data fusion and deep learning algorithms, the accuracy of identifying violations is improved.

[0106] Fourth, optimize management efficiency: automatically generate violation reports and provide targeted suggestions for personnel training based on historical data.

[0107] Fifth, adaptability to complex scenarios: Through self-learning optimization modules, the system can adapt to the needs of different power operation scenarios, can be expanded to scenarios such as substations and transmission line inspections, and supports multi-robot collaborative networking.

[0108] Sixth, reduce human resources: In actual work operations, two people must cooperate, one is responsible for operation, and the other is responsible for supervising the operator and the work process. The use of this method and the robot equipped with this method is equivalent to replacing the guardian, which greatly saves human resources.

[0109] The above embodiment is only one of the implementation methods that can realize the technical solution of the present invention. The scope of protection claimed by the present invention is not limited only to this embodiment, but also includes changes, replacements and other implementation methods that can be easily thought of by any technician familiar with this technical field within the technical scope disclosed by the present invention.

[0110] Finally, 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 the above embodiments, ordinary technicians in the field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered by the scope of protection of the claims of the present invention.

Claims

1. A method for identifying safety violations in power operations based on embodied intelligence, characterized in that: include: Collect workers' posture, position and physiological data, power operation site images and environmental parameters; Build a dynamic power operation scene model based on the operator's posture, position and physiological data, power operation site images and environmental parameters; The newly optimized violation behavior recognition model is used to identify the dynamic power operation scenario model to obtain the power operation safety violation identification results. The basic model of the violation behavior recognition model is a deep neural network model, which is constructed based on historical violation data and real-time operation scenario data. The violation behavior recognition model is continuously optimized based on each dynamic power operation scenario model to be identified and the corresponding power operation safety violation identification results. Based on the results of the identification of safety violations in power operations, identification result execution instructions are output to the robot, so that the robot can perform safety management and feedback operations according to the result execution instructions.

2. The method for identifying safety violations in power operations based on embodied intelligence according to claim 1, characterized in that: After calling the latest optimized violation behavior recognition model to identify the dynamic power operation scenario model and obtaining the power operation safety violation identification result, the method further includes: Based on the results of the identification of safety violations in power operations, a risk coefficient is evaluated; the risk coefficient is used to trigger a graded warning.

3. The method for identifying safety violations in power operations based on embodied intelligence according to claim 2, characterized in that: The risk factor obtained by evaluating the safety violation identification results of power operations includes: The risk coefficient is calculated based on the comprehensive analysis of the severity of the power operation safety violation identification results and the environmental parameters.

4. The method for identifying safety violations in power operations based on embodied intelligence according to claim 2, characterized in that: After the risk factor is obtained based on the results of the safety violation identification of the power operation, the following steps are also performed: Trigger graded warnings based on the current risk factor, including voice reminders, remote braking, and emergency intervention. When the risk factor exceeds the preset threshold, an emergency intervention instruction is output for the robot to perform emergency intervention operations.

5. The method for identifying safety violations in power operations based on embodied intelligence according to claim 1, characterized in that: The power operation site image includes the behavior image of the operator; the environmental parameters include wind speed, temperature, humidity and air pressure; and the physiological data includes the heart rate data of the operator.

6. The method for identifying safety violations in power operations based on embodied intelligence according to claim 1, characterized in that: The method uses the latest optimized violation behavior recognition model to identify the dynamic power operation scenario model, and obtains the power operation safety violation identification results, including: The latest optimized violation behavior recognition model is used to identify the dynamic power operation scenario model. The recognition process includes: Use the YOLOv7 algorithm to identify whether workers are wearing protective equipment; Use the Transformer algorithm to identify whether workers accidentally touch live equipment or perform other illegal operations; After the identification is completed, the results of the power operation safety violation identification are obtained.

7. The method for identifying safety violations in power operations based on embodied intelligence according to claim 6, characterized in that: The identification process further includes: Determine whether the operator's physiological state is abnormal based on the operator's physiological data; According to the operator's posture, it is judged whether the operator's posture is abnormal. If the posture is judged to be abnormal, the locking function is automatically activated to prevent falling from a height.

8. The method for identifying safety violations in power operations based on embodied intelligence according to claim 1, characterized in that: Before collecting the posture, position and physiological data of the operator, the power operation site image and environmental parameters, the following steps are also included: Verify the relevance between work orders and plan orders to identify planned work risks; Based on the collected environmental parameters, determine whether the working environment meets the working safety conditions.

9. A power operation safety violation identification system based on embodied intelligence, characterized by: include: The acquisition module is used to collect the posture, position and physiological data of workers, images of the power operation site and environmental parameters; A scenario construction module is used to construct a dynamic power operation scenario model based on the operator's posture, position and physiological data, power operation site images and environmental parameters; An identification module is configured to call a newly optimized violation behavior identification model to identify a dynamic power operation scenario model and obtain a power operation safety violation identification result; wherein the violation behavior identification model is based on a deep neural network model constructed based on historical violation data and real-time operation scenario data, and the violation behavior identification model is continuously optimized based on each dynamic power operation scenario model to be identified and the corresponding power operation safety violation identification result; The instruction output module is used to output the identification result execution instruction to the robot based on the identification result of the power operation safety violation, so that the robot can perform safety control and feedback operations according to the result execution instruction.

10. A robot, characterized in that: The robot comprises a robot body; the robot body is equipped with a computer program; when the computer program is executed by a processor, the computer program is used to implement the steps of the method for identifying safety violations in power operations based on embodied intelligence according to any one of claims 1 to 8; The robot body is provided with a mobile chassis unit for autonomous navigation at the power operation site and supports multi-terrain movement; The robot body is provided with a voice interaction unit for voice communication with the operator and providing real-time violation reminders.

Images