Nuclear industrial robot body intelligent system based on large model

Abstract

The invention discloses a nuclear industrial robot body intelligent system based on a large model, which realizes autonomous sensing, decision making and execution of a robot in a nuclear industrial environment by integrating a sensing module, a learning model module, an execution module, a communication module and a security module. The sensing module adopts a multi-sensor fusion technology to collect environmental data such as temperature, radiation intensity and gas concentration in real time; the decision-making module dynamically generates a task execution strategy based on an open source large model base in combination with deep learning and reinforcement learning algorithms; the execution module completes complex tasks such as equipment maintenance, fault processing and nuclear waste carrying through a high-precision mechanical arm and a force feedback control technology; the communication module supports data interaction between the robot and a control center, and remote monitoring and emergency intervention are achieved; the safety module monitors the state of the robot in real time, multiple safety protection mechanisms are provided, and stable operation of the robot in the high-radiation environment is ensured. According to the invention, the operation efficiency and safety of nuclear industry facilities are obviously improved.

Description

Technical Field

[0001] The present invention relates to the cross - field of artificial intelligence and nuclear industry applications, and specifically relates to a nuclear industry application system and method based on the embodied intelligence model of nuclear industry robots with large models, which are used to perform complex tasks in nuclear industry scenarios and improve the operation efficiency and safety of nuclear facilities. Background Art

[0002] The nuclear industry is an important field involving nuclear energy development and utilization, covering multiple branches such as nuclear power plants, nuclear fuel processing, nuclear waste treatment, and nuclear medicine. Nuclear industry facilities are characterized by high danger, high complexity, and high precision. Traditional manual operations and ordinary robots are difficult to meet their requirements.

[0003] In the nuclear industry, workers often face extreme environments such as high radiation, high temperature, and high pressure. Long - term exposure may cause serious harm to health. In addition, tasks such as equipment maintenance, fault handling, and nuclear waste handling in nuclear industry facilities require extremely high precision and reliability, and manual operations pose great risks.

[0004] In the prior art, most robots used in the nuclear industry adopt preset programs or remote control methods, lacking the ability of autonomous decision - making and environmental adaptability. For example, in the equipment maintenance of nuclear power plants, robots usually need to rely on manual remote control, with low efficiency and difficulty in dealing with emergencies. In nuclear waste treatment, existing robots lack the ability to perceive complex environments and are difficult to achieve precise operations.

[0005] The embodied intelligence model of nuclear industry robots with large models is an advanced technology that combines deep learning, reinforcement learning, and embodied intelligence, enabling robots to perform autonomous perception, decision - making, and execution in complex environments. However, there is currently no specific solution to apply this technology to the nuclear industry, especially the systematic application in high - risk scenarios such as nuclear power plants and nuclear waste treatment. Summary of the Invention

[0006] The purpose of the present invention is to provide a nuclear industry application system and method based on the embodied intelligence model of nuclear industry robots with large models, which realizes the efficient and safe operation of robots in the nuclear industry environment by combining artificial intelligence large models with nuclear industry facilities.

[0007] The present invention proposes an embodied intelligence model of nuclear industry robots with large models, and this model realizes its functions through the following core modules:

[0008] The perception module is used to collect environmental data of nuclear industry facilities in real - time, including temperature, radiation intensity, gas concentration, equipment status, etc., and provides the robot with comprehensive multi - modal environmental perception capabilities by means of an open - source large - model base.

[0009] The decision-making module conducts in-depth reasoning based on the open-source large model base, combines task requirements and perceived multi-modal information, and makes intelligent decisions autonomously. This module can dynamically adjust task strategies according to environmental data to adapt to the complex requirements of nuclear industrial facilities;

[0010] The action module, according to the decision-making information given by the decision-making module, gradually calls the instructions generated by the skill learning small model module to achieve the basic motion control function of the robot, and thus gradually completes tasks such as robot equipment maintenance, fault handling, and nuclear waste handling. The action module drives a high-precision robotic arm and multi-degree-of-freedom joints to ensure the accuracy and reliability of operations;

[0011] The communication module realizes data interaction between the robot and the control center, and supports remote monitoring and emergency intervention. The communication module adopts high-speed and low-latency communication technologies to ensure the real-time and security of data transmission.

[0012] The safety module monitors the robot's status in real time, including battery power, mechanical component status, radiation exposure, etc., and provides multiple safety protection mechanisms to ensure the stable operation of the robot in a high-radiation environment. Description of the Drawings

[0013] Figure 1 It is a flowchart of the robot inspection task method according to an embodiment of the present application; Detailed Embodiments

[0014] The present invention will be described in detail below with reference to the drawings and embodiments.

[0015] As Figure 1 shown, the embodied intelligent system of the nuclear industry robot based on a large model, which system includes a perception module, a decision-making module, an action module, an action module, a communication module, and a safety module;

[0016] The perception module collects information in the nuclear industry environment through a variety of sensors, including equipment status, radiation level, and obstacle position. The information is processed and then transmitted to the decision-making module.

[0017] After receiving the information transmitted by the perception module, the decision-making module analyzes and reasons in combination with the task objectives, and generates specific action strategies, including: according to the perceived equipment fault location and the surrounding environment, the decision-making module plans the best path for the robot to reach the fault point.

[0018] The action strategy generated by the decision-making module will be sent to the action module in the form of instructions. The action module controls the robotic arm and the mobile chassis of the robot to complete specific actions according to these instructions, including: after the decision-making module plans the path to grasp a certain component, the action module controls the robotic arm to move along this path and grasp the component. During the execution of the task, the action module will feedback the execution result to the decision-making module. The decision-making module evaluates the effect of the task execution according to the feedback information or adjusts the decision. Including: if the action module finds that the position of the component is deviated during the grasping process, it will feedback to the decision-making module, and the decision-making module will readjust the grasping strategy.

[0019] The communication module is responsible for information interaction between various modules within the embodied intelligent system and between the robot and the external environment. It is responsible for uploading the data collected by the perception module to the cloud for model training and optimization, and at the same time transmitting the instructions generated by the decision-making module to the action module. Through the communication module, information can be transmitted in real time between modules, ensuring the coordinated operation of the entire system. When the perception module detects an abnormal situation, it can timely transmit the information to the decision-making module through the communication module, and the decision-making module makes a response and commands the action module to take corresponding measures.

[0020] The security module runs throughout the operation of the embodied intelligent system, and conducts real-time security monitoring on the perception module, decision-making module, and action module; the security module checks whether the working status of each module meets the security requirements, including whether the perception module normally collects data, whether the instructions of the decision-making module are reasonable, and whether the actions of the action module are safe.

[0021] Furthermore, the decision-making module is based on the open-source large model module and is used to perform in-depth reasoning by combining task requirements and perceived multi-modal information, making intelligent decisions autonomously, and being able to dynamically adjust task strategies according to environmental data to adapt to the complex requirements of nuclear industrial facilities; the decision-making module receives task instructions from the outside, and these instructions include equipment maintenance tasks and nuclear waste handling tasks. The decision-making module analyzes the task instructions through natural language processing (NLP) technology to clarify key information such as task objectives, task priorities, and task scopes. For the task of "checking the leakage of the reactor cooling system pipeline", the decision-making module analyzes that the task objective is to check for leakage and the task scope is the reactor cooling system pipeline. The knowledge base unit in the decision-making module stores knowledge related to the nuclear industry, including equipment structure, failure modes, and operation procedures. According to the current task requirements and environmental information, relevant knowledge entries are retrieved from the knowledge base. For example, for the task of "checking for leakage", the leakage location and detection methods related to pipeline leakage detection are retrieved. The decision-making module is based on the DeepSeek open-source large model base and calls the in-depth thinking decision-making model therein. This model is a pre-trained deep learning model with powerful reasoning and decision-making capabilities. The decision-making module inputs the task requirement analysis results and the three-dimensional environment model generated by environmental modeling into the in-depth thinking decision-making model. Inside the in-depth thinking decision-making model, through a multi-modal information fusion mechanism, comprehensive analysis and reasoning are carried out on the task requirements and visual information, radiation information, and temperature information in the environmental model. For example, according to the task objective of checking for leakage, the model combines the geometric structure and radiation distribution of the pipeline in the environmental model to infer possible leakage locations and paths. At the same time, it also considers the location of obstacles and changes in radiation intensity in the environment to evaluate the risks and feasibility of task execution. After in-depth reasoning, the in-depth thinking decision-making model generates specific decision instructions. These decision instructions include the task execution path (such as the optimal path for the robot to reach the target location), operation steps (such as the detection sequence and methods when checking for leakage), safety measures (such as precautions for operations in high-radiation areas), etc. For example, the decision instructions generated by the model may include detailed information such as "the robot travels along path A to the cooling system pipeline, first checks connection 1 of the pipeline, then checks connection 2, and pay attention to pausing operations and returning to the safe area when the radiation intensity exceeds threshold X". The reinforcement learning unit in the decision-making module optimizes the generated decision instructions according to historical task execution data and environmental feedback information. The reinforcement learning algorithm learns to make optimal decisions in different environmental states through a reward mechanism. For example, if the robot finds that there are obstacles in the originally planned path during task execution, the reinforcement learning algorithm will adjust the path planning strategy according to the feedback information and optimize the decision instructions for subsequent task execution.

[0022] Furthermore, the action module gradually calls the instructions generated by the skill learning sub-model module according to the decision information given by the decision module to achieve the basic motion control function of the robot, thereby gradually completing tasks such as nuclear industry robot equipment maintenance, fault handling, and nuclear waste handling. The action module executes tasks according to the decision instructions generated by the decision module and feeds back the execution results to the decision module in real time. The decision module evaluates the effect of task execution based on the feedback information. If it is found that there is a deviation between the execution result and the expected goal (such as the detected leakage location does not match the inference result), the decision module will combine the new feedback information and re-call the deep thinking decision model for inference to adjust the decision instructions. For example, if the action module feedbacks that no leakage is found when checking pipe joint 1, but there is an abnormality at joint 2, the decision module will re-infer and generate new instructions to guide the action module to adjust the inspection order or increase the inspection frequency, etc.

[0023] Furthermore, the multiple sensors of the perception module include visual sensors, lidar, and radiation sensors; the visual sensors acquire visual information of the equipment appearance and obstacle positions; the radiation sensors detect radiation intensity; the temperature sensors monitor the ambient temperature; the gas sensors detect the concentration of harmful gases; the collected raw data may have problems such as noise and inconsistent formats. The perception module preprocesses this raw data, including data cleaning, data format conversion, and data fusion. In terms of vision, the visual information is combined with the radiation intensity data through a data fusion algorithm to generate a comprehensive environmental map containing the equipment appearance and radiation distribution.

[0024] The action module drives a high-precision robotic arm and multi-degree-of-freedom joints to ensure the accuracy and reliability of operations;

[0025] The communication module realizes data interaction between the nuclear industry robot and the control center, supporting remote monitoring and emergency intervention.

[0026] The safety module monitors the status of the nuclear industry robot in real time, including battery power, mechanical component status, and radiation exposure, and provides multiple safety protection mechanisms to ensure the stable operation of the nuclear industry robot in a high-radiation environment.

[0027] Furthermore, the perception module includes:

[0028] A multi-sensor fusion unit that integrates temperature sensors, radiation sensors, gas sensors, and visual sensors for comprehensively collecting environmental data;

[0029] An environmental modeling unit that constructs a three-dimensional environmental model of the nuclear industry facility based on the collected data to provide spatial positioning and path planning support for the nuclear industry robot.

[0030] The decision module includes:

[0031] A deep learning unit for analyzing perceptual data, identifying environmental states and potential risks;

[0032] A reinforcement learning unit for optimizing task execution strategies and enhancing the autonomous decision-making ability of nuclear industry robots;

[0033] A knowledge base unit that constructs a knowledge base for nuclear industry tasks through continuous learning and accumulation, enabling in-depth reasoning and enhancing the adaptability and efficiency of nuclear industry robots in similar scenarios.

[0034] The action module includes:

[0035] A high-precision robotic arm for performing equipment maintenance and nuclear waste handling;

[0036] A force feedback control unit that achieves precise control through force feedback technology to avoid damaging equipment or the environment during operation;

[0037] A task feedback unit that real-time feeds back the operation status to the learning model module to ensure the accuracy and safety of task execution.

[0038] The communication module includes: A high-speed data transmission unit that uses 5G or fiber optic communication technology to ensure the real-time and reliability of data transmission;

[0039] A remote monitoring and intervention unit that supports the control center to remotely monitor and emergently intervene in nuclear industry robots;

[0040] A data encryption unit that uses encryption transmission technology to prevent data leakage and malicious attacks;

[0041] Furthermore, the security module includes:

[0042] A status monitoring unit that real-time monitors the status of nuclear industry robots, including battery power, mechanical component status, and radiation exposure;

[0043] A safety protection mechanism unit that immediately activates a safety protection mechanism, such as stopping operation and returning to a safe area, when an abnormal situation is detected;

[0044] A radiation protection unit that uses special materials and designs to reduce the degree of radiation exposure of nuclear industry robots and extend their service life.

[0045] A method for learning the embodied intelligence model of nuclear industry robots based on large models, including the following steps:

[0046] The nuclear industry robot collects environmental data of nuclear industry facilities through the perception module;

[0047] The learning model module based on the DeepSeek interface analyzes and learns the perceptual data to generate task execution strategies;

[0048] The action module controls the nuclear industry robot to complete specific operations according to the instructions of the learning model module;

[0049] The safety module monitors the robot's status in real time to ensure its safe operation in the nuclear industry environment;

[0050] The communication module transmits the operation data and task execution status of the nuclear industry robot to the control center to support remote monitoring and emergency intervention.

[0051] The environmental perception and data acquisition steps include:

[0052] Using multi-sensor fusion technology to collect temperature, radiation intensity, gas concentration, and equipment status data;

[0053] Transmit the collected data to the learning model module.

[0054] The data learning and task planning steps include:

[0055] Analyze the perception data based on deep learning algorithms to identify the environmental status and potential risks;

[0056] Optimize the task execution strategy based on reinforcement learning algorithms, including task priorities, execution paths, and operation steps.

[0057] The task execution and operation control steps include:

[0058] Control the nuclear industry robot to complete equipment maintenance, fault handling, and nuclear waste handling tasks;

[0059] Real-time feedback the operation status to the learning model module to ensure the accuracy and safety of task execution.

[0060] Furthermore, the real-time monitoring and safety protection steps include:

[0061] Real-time monitor the robot's status, including battery power, mechanical component status, and radiation exposure;

[0062] When an abnormal situation is detected, immediately activate the safety protection mechanism, including stopping operations and returning to a safe area.

[0063] Furthermore, the data interaction and remote intervention steps include:

[0064] Transmit the robot's operation data and task execution status to the control center;

[0065] Support the control center to remotely monitor and emergently intervene in the nuclear industry robot.

[0066] An application system of an embodied intelligence model for a nuclear industry robot based on a large model, comprising:

[0067] Nuclear power plant equipment maintenance system, used to realize the equipment inspection, fault diagnosis and maintenance tasks of nuclear industry robots in nuclear power plants;

[0068] Nuclear waste treatment system, used to realize the precise operation of nuclear industry robots in nuclear waste handling and treatment tasks;

[0069] Nuclear reactor maintenance system, used to realize the replacement and maintenance tasks of reactor components by nuclear industry robots in high-radiation environments;

[0070] Nuclear accident emergency handling system, used to realize the emergency handling tasks of nuclear industry robots entering dangerous areas after nuclear accidents occur.

[0071] The nuclear power plant equipment maintenance system includes:

[0072] Equipment inspection unit, used to regularly inspect the key equipment of the nuclear power plant;

[0073] Fault diagnosis unit, used to identify equipment anomalies and generate maintenance plans;

[0074] Maintenance execution unit, used to control nuclear industry robots to complete equipment maintenance tasks.

[0075] The nuclear waste treatment system includes:

[0076] Nuclear waste identification unit, used to accurately identify the location and status of nuclear waste;

[0077] Handling and encapsulation unit, used to control nuclear industry robots to complete nuclear waste handling and encapsulation tasks;

[0078] Safety inspection unit, used to detect radiation in the encapsulation area to ensure no leakage.

[0079] Adopt high-precision sensors, such as temperature sensors, radiation sensors, gas sensors, etc., to ensure the accuracy of data collection.

[0080] Through multi-sensor fusion technology, improve the comprehensiveness and reliability of environmental perception.

[0081] Based on open-source large model base interfaces (such as DeepSeek, Tongyi Qianwen, etc.), combined with the characteristics of the nuclear industry scenario, train special industry small models required for the decision-making of nuclear industry robots.

[0082] Utilize the multi-expert algorithm (MoE) to optimize the task execution strategy and continuously improve the robot's autonomous decision-making ability.

[0083] Adopt high-precision robotic arms and multi-degree-of-freedom joints to ensure that the robot can complete complex operations.

[0084] Through force feedback control technology, precise control of the robot during operation is achieved.

[0085] Multiple safety mechanisms are set up, including hardware redundancy, software fault tolerance, emergency braking, etc., to ensure the safe operation of the robot in the nuclear industry environment.

[0086] The radiation exposure of the robot is monitored in real time, and when it exceeds the safety threshold, protective measures are immediately initiated.

[0087] High-speed and low-latency communication technology, 5G communication, is adopted to ensure the real-time and reliability of data transmission.

[0088] Encryption transmission is supported to prevent data leakage and malicious attacks.

[0089] Intelligent inspection:

[0090] Temperature sensors, radiation sensors, vision sensors, etc. are integrated to collect environmental data in real time;

[0091] Based on the trained dedicated small model for the nuclear industry, the inspection path is dynamically planned to avoid obstacles;

[0092] The sensor data is analyzed through deep learning algorithms to identify equipment anomalies or potential risks;

[0093] The robot starts the inspection task and goes to the target area according to the preset path;

[0094] Through multi-sensor fusion technology, data such as equipment temperature and radiation intensity are collected in real time;

[0095] When the temperature of a certain device is detected to be abnormal (exceeding the threshold), the robot immediately marks the abnormal position and sends an alarm to the control center through the communication module;

[0096] After the control center confirms, the robot continues to perform other inspection tasks.

[0097] It should be noted that in the specification provided here, a large number of specific details are described. However, it can be understood that the embodiments of the present application can be practiced without these specific details. In some instances, well-known structures and technologies are not shown in detail so as not to obscure the understanding of this specification.

[0098] It should also be noted that the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, such that a process, method, commodity or device comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, commodity or device. Without further limitation, an element defined by the statement "comprising an..." does not exclude the presence of additional identical elements in the process, method, commodity or device comprising said element.

[0099] The above are only examples of the present application and are not intended to limit the present application. For those skilled in the art, various changes and modifications can be made to the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present application shall be included within the scope of the claims of the present application.

Claims

1. A nuclear industry robot embodied intelligent system based on a large model, characterized by: The system includes a perception module, a decision module, an action module, an action module, a communication module, and a security module; The perception module collects information in the nuclear industry environment through a variety of sensors, including equipment status, radiation levels, and obstacle locations; the information is processed and transmitted to the decision module; After receiving the information from the perception module, the decision module analyzes and infers the information based on the task objectives and generates a specific action strategy, including: based on the perceived equipment fault location and surrounding environment, the decision module plans the best path for the robot to reach the fault point; The action strategy generated by the decision module will be sent to the action module in the form of instructions. The action module controls the robot's mechanical arm and mobile chassis to complete specific actions according to these instructions, including: after the decision module plans the path to grab a certain part, the action module controls the mechanical arm to move along the path and grab the part; during the execution of the task, the action module will feedback the execution result to the decision module, and the decision module will evaluate the effect of the task execution based on the feedback information, or adjust the decision; including: if the action module finds that the position of the part is deviated during the grasping process, it will feedback to the decision module, and the decision module will readjust the grasping strategy; The communication module is responsible for information exchange between modules within the embodied intelligent system and between the robot and the external environment. It is responsible for uploading the data collected by the perception module to the cloud for model training and optimization, and transmitting the instructions generated by the decision module to the action module. Through the communication module, information can be transmitted between modules in real time to ensure the coordinated work of the entire system. When the perception module detects an abnormal situation, it can promptly transmit the information to the decision module through the communication module, and the decision module responds and instructs the action module to take corresponding measures. The security module runs through the entire operation process of the embodied intelligent system, and performs real-time security monitoring of the perception module, decision module, and action module; the security module will check whether the working status of each module meets the safety requirements, including whether the perception module collects data normally, whether the instructions of the decision module are reasonable, and whether the actions of the action module are safe.

2. The large model-based nuclear industry robot embodied intelligent system according to claim 1 is characterized in that: The decision module is based on the open source large model module, which is used to combine task requirements and perceived multimodal information for deep reasoning, make intelligent decisions autonomously, and can dynamically adjust task strategies according to environmental data to adapt to the complex needs of nuclear industry facilities; the decision module receives task instructions from the outside, which include equipment maintenance tasks and nuclear waste handling tasks; the decision module parses task instructions through natural language processing (NLP) technology to clarify task objectives, task priorities, and key information on task scope; for the task of "checking the leakage of reactor cooling system pipelines", the decision module parses out that the task objective is to check leakage, and the task scope is reactor cooling system pipelines; the knowledge base unit in the decision module stores knowledge related to the nuclear industry, including equipment structure, failure mode, and operating procedures. According to the current task requirements and environmental information, relevant knowledge items are retrieved from the knowledge base. For the task of "checking leakage", the leakage location and detection method related to pipeline leakage detection are retrieved; the decision module The block is based on the DeepSeek open source large model base, and calls the deep thinking decision model in it; the decision module inputs the task requirement analysis results and the three-dimensional environmental model generated by environmental modeling into the deep thinking decision model; the deep thinking decision model uses a multimodal information fusion mechanism to comprehensively analyze and reason the visual information, radiation information, and temperature information in the task requirements and environmental model, including that the model will infer the possible leakage location and path based on the task goal of checking for leaks, combined with the geometric structure and radiation distribution of the pipeline in the environmental model; at the same time, the location of obstacles and radiation intensity changes in the environment are considered to evaluate the risk and feasibility of task execution; after deep reasoning, the deep thinking decision model generates specific decision instructions; the reinforcement learning unit in the decision module optimizes the generated decision instructions based on historical task execution data and environmental feedback information; the reinforcement learning algorithm learns to make optimal decisions under different environmental conditions through a reward mechanism.

3. The large-model-based nuclear industry robot embodied intelligent system according to claim 1 is characterized in that: Based on the decision information given by the decision module, the action module gradually calls the instructions generated by the skill learning small model module to realize the basic motion control function of the robot, thereby gradually completing the nuclear industry robot equipment maintenance, troubleshooting, and nuclear waste handling tasks; the action module executes tasks according to the decision instructions generated by the decision module, and feeds back the execution results to the decision module in real time; the decision module evaluates the effect of task execution based on the feedback information; if it is found that the execution result deviates from the expected goal, the decision module will re-call the deep thinking decision model for reasoning based on the new feedback information and adjust the decision instructions.

4. The large-model-based nuclear industry robot embodied intelligent system according to claim 1 is characterized in that: The various sensors of the perception module include visual sensors, laser radars and radiation sensors; the visual sensor obtains visual information about the appearance of the equipment and the location of obstacles; the radiation sensor detects the radiation intensity; the temperature sensor monitors the ambient temperature; the gas sensor detects the concentration of harmful gases; the collected raw data may have noise and inconsistent formats; the perception module pre-processes these raw data, including data cleaning, data format conversion, and data fusion; In terms of vision, the visual information is combined with the radiation intensity data through a data fusion algorithm to generate a comprehensive environmental map including the equipment appearance and radiation distribution; The action module drives high-precision robotic arms and multi-degree-of-freedom joints to ensure the accuracy and reliability of operations; The communication module enables data exchange between nuclear industrial robots and the control center, supporting remote monitoring and emergency intervention; The safety module monitors the status of nuclear industrial robots in real time, including battery power, status of mechanical components, and radiation exposure, and provides multiple safety protection mechanisms to ensure the stable operation of nuclear industrial robots in high radiation environments.

5. The embodied intelligent model of nuclear industry robots based on a large model according to claim 1 is characterized in that: The perception module comprises: Multi-sensor fusion unit, integrating temperature sensor, radiation sensor, gas sensor, and visual sensor, for comprehensive collection of environmental data; The environmental modeling unit builds a three-dimensional environmental model of nuclear industrial facilities based on the collected data, providing spatial positioning and path planning support for nuclear industrial robots; The decision module comprises: A deep learning unit that analyzes sensory data to identify environmental conditions and potential risks; Reinforcement learning unit, used to optimize task execution strategies and enhance the autonomous decision-making capabilities of nuclear industry robots; The knowledge base unit builds a nuclear industry task knowledge base through continuous learning and accumulation, which can perform deep reasoning and improve the adaptability and efficiency of nuclear industry robots in similar scenarios; The action module includes: High-precision robotic arms, used to complete equipment maintenance and nuclear waste handling; Force feedback control unit, which achieves precise control through force feedback technology to avoid damage to equipment or environment during operation; Task feedback unit, which provides real-time feedback of operation status to the learning model module to ensure the accuracy and safety of task execution; The communication module includes: a high-speed data transmission unit, which uses 5G or optical fiber communication technology to ensure the real-time and reliability of data transmission; Remote monitoring and intervention unit, which supports the control center to remotely monitor and conduct emergency intervention on nuclear industrial robots; The data encryption unit uses encrypted transmission technology to prevent data leakage and malicious attacks.

6. The large-scale nuclear industry robot embodied intelligent model according to claim 1, characterized in that: The security module comprises: The status monitoring unit monitors the status of nuclear industrial robots in real time, including battery power, mechanical component status, and radiation exposure; The safety protection mechanism unit immediately activates the safety protection mechanism when an abnormal situation is detected, such as stopping operation and returning to a safe area; The radiation protection unit uses special materials and designs to reduce the degree to which nuclear industrial robots are affected by radiation and extend their service life.

7. A method for learning a nuclear industry robot embodied intelligent model for a large model of a system as described in any one of claims 1 to 6, characterized in that: The following steps are involved: Nuclear industry robots collect environmental data of nuclear industry facilities through perception modules; The learning model module based on the DeepSeek interface analyzes and learns the perception data and generates task execution strategies; The action module controls the nuclear industrial robot to complete specific operations according to the instructions of the learning model module; The safety module monitors the robot status in real time to ensure its safe operation in the nuclear industry environment; The communication module transmits the operation data and task execution status of the nuclear industrial robot to the control center, supporting remote monitoring and emergency intervention; The environmental perception and data collection steps include: Use multi-sensor fusion technology to collect temperature, radiation intensity, gas concentration, and equipment status data; Transmit the collected data to the learning model module; The data learning and task planning steps include: Analyze perception data based on deep learning algorithms to identify environmental conditions and potential risks; Optimize task execution strategies based on reinforcement learning algorithms, including task priority, execution path, and operation steps; The task execution and operation control steps include: Control nuclear industrial robots to complete equipment maintenance, troubleshooting, and nuclear waste handling tasks; Real-time feedback of operation status to the learning model module ensures the accuracy and safety of task execution.

8. The large-scale nuclear industry robot embodied intelligent model learning method according to claim 7 is characterized in that: The real-time monitoring and safety protection steps include: Real-time monitoring of robot status, including battery charge, mechanical component status, and radiation exposure; When an abnormal situation is detected, the safety protection mechanism is immediately activated, including stopping operation and returning to a safe area.

9. The large-scale nuclear industry robot embodied intelligent model learning method according to claim 7 is characterized in that: The data interaction and remote intervention steps include: Transmit the robot's operating data and task execution status to the control center; Support the control center to conduct remote monitoring and emergency intervention of nuclear industry robots.

10. An application system of a nuclear industry robot embodied intelligent model based on a large model for implementing the method as claimed in claim 7, characterized in that: include: Nuclear power plant equipment maintenance system, used to realize equipment inspection, fault diagnosis and maintenance tasks of nuclear industrial robots in nuclear power plants; Nuclear waste treatment system, used to achieve precise operation of nuclear industrial robots in nuclear waste handling and treatment tasks; Nuclear reactor maintenance system, used to realize the replacement and maintenance tasks of reactor components by nuclear industrial robots in high radiation environments; Nuclear accident emergency response system, used to enable nuclear industrial robots to enter dangerous areas and perform emergency response tasks after a nuclear accident occurs; The nuclear power plant equipment maintenance system comprises: Equipment inspection unit, used to regularly inspect key equipment in nuclear power plants; Fault diagnosis unit, used to identify equipment anomalies and generate maintenance plans; Maintenance execution unit, used to control nuclear industry robots to complete equipment maintenance tasks; The nuclear waste treatment system comprises: Nuclear waste identification unit, used to accurately identify the location and status of nuclear waste; The handling and packaging unit is used to control the nuclear industry robot to complete the nuclear waste handling and packaging tasks; Safety inspection unit, used to perform radiation detection on the packaging area to ensure there is no leakage.

Images