A Control Method and System for Trackless Rubber-Tired Vehicles in Wells Based on Equipment Thermal State Sensing

Abstract

This application provides a control method and system for a trackless rubber-tired vehicle in underground mines based on equipment thermal state perception, belonging to the field of trackless rubber-tired vehicle control technology. The method includes: planning a travel path based on the target task type, a digital map of the roadway in the target travel area, and positioning information; controlling the trackless rubber-tired vehicle to travel along the travel path and collecting environmental image data, temperature data, and real-time operating condition data of the roadway ahead; obtaining the environmental thermal risk level based on the environmental image data and the expected travel trajectory corresponding to the travel path; inputting the temperature data, real-time operating condition data, and environmental thermal risk level into a thermal state prediction model to obtain a predicted temperature change curve; determining the thermal state assessment level based on the current temperature value of the target component and the predicted temperature change curve; and inputting the environmental thermal risk level, thermal state assessment level, target task type, target travel area, and real-time operating condition data into a collaborative decision engine to generate a collaborative control instruction set.

Description

Technical Field

[0001] This application relates to the field of trackless rubber-tired vehicle control technology, and in particular to a control method and system for a trackless rubber-tired vehicle in wells based on equipment thermal state sensing. Background Technology

[0002] Trackless rubber-tired vehicles are core equipment for material transportation, personnel transfer, and equipment inspection in underground mines. Their operational stability directly affects production efficiency and personnel safety. With the improvement of mine intelligence and the increase in underground mining depth, roadways are characterized by high temperature, high humidity, and narrow space, which places higher demands on the task adaptability, long-term operation capability, and thermal management reliability of rubber-tired vehicles.

[0003] In existing technologies, the control of rubber-tired vehicles revolves around path planning and basic operating condition adjustment. It employs UWB positioning combined with Dijkstra's algorithm to plan the shortest path, collects basic operating condition and environmental parameters, adjusts the vehicle speed according to a speed threshold, and uses a constant power mode for the cooling system. However, this approach has significant drawbacks in complex task switching and diverse tunnel environments: constant cooling can easily lead to component overheating and subsequent malfunctions, or excessive heat dissipation and energy waste. Furthermore, it lacks assessment of tunnel thermal risks, has no temperature prediction mechanism, and cannot adapt to new requirements.

[0004] Therefore, there is an urgent need for a control method and system for trackless rubber-tired vehicles in wells based on equipment thermal state sensing, in order to ensure the safe, efficient and energy-saving operation of trackless rubber-tired vehicles. Summary of the Invention

[0005] To address the aforementioned technical problems, this application provides a control method and system for a trackless rubber-tired vehicle in wells based on equipment thermal state sensing.

[0006] In a first aspect, this application provides a control method for a trackless rubber-tired vehicle in wells based on equipment thermal state sensing, including: In response to the received task instructions, determine the target task type and target driving area; Based on the target task type, the digital map of the alleyway in the target driving area, and the positioning information of the trackless rubber-wheeled vehicle, the driving route is planned; Control the trackless rubber-tired vehicle to travel along the driving path, and collect environmental image data of the road ahead, temperature data of the target components of the trackless rubber-tired vehicle, and real-time operating data of the trackless rubber-tired vehicle; Based on the environmental image data of the roadway ahead and the expected driving trajectory corresponding to the driving path, the thermal environment of the roadway ahead is assessed to obtain the environmental thermal risk level. Temperature data, real-time operating data, and environmental thermal risk level are input into a pre-set thermal state prediction model to obtain the predicted temperature change curve of the target component in the future. Based on the current temperature value of the target component and the predicted temperature change curve, the thermal state assessment level of the trackless rubber-wheeled vehicle is determined. The environmental thermal risk level, thermal status assessment level, target task type, target driving area and real-time operating condition data are input into the collaborative decision engine to generate a collaborative control instruction set, which includes motion control instructions and thermal management control instructions. Execute the coordinated control command set to adjust the driving status and thermal load status of the trackless rubber-wheeled vehicle.

[0007] Secondly, this application provides a control system for a trackless rubber-tired vehicle in wells based on equipment thermal state sensing, comprising: The task processing module is used to respond to received task instructions and determine the target task type and target driving area; The path planning module is used to plan the travel path based on the target task type, the digital map of the alleyway in the target travel area, and the positioning information of the trackless rubber-wheeled vehicle. The data acquisition module is used to control the trackless rubber-tired vehicle to travel along the driving path and to collect environmental image data of the road ahead, temperature data of the target components of the trackless rubber-tired vehicle, and real-time operating data of the trackless rubber-tired vehicle. The thermal environment assessment module is used to assess the thermal environment of the roadway ahead based on environmental image data of the roadway ahead and the expected driving trajectory corresponding to the driving path, and to obtain the environmental thermal risk level. The thermal condition assessment module is used to input temperature data, real-time operating condition data and environmental thermal risk level into the preset thermal condition prediction model to obtain the predicted temperature change curve of the target component in the future period. Based on the current temperature value of the target component and the predicted temperature change curve, the thermal condition assessment level of the trackless rubber-tired vehicle is determined. The collaborative decision engine is used to input environmental thermal risk level, thermal state assessment level, target task type, target driving area and real-time operating condition data into the collaborative decision engine to generate a collaborative control instruction set, which includes motion control instructions and thermal management control instructions. The control execution module is used to execute the collaborative control instruction set and adjust the driving status and thermal load status of the trackless rubber-wheeled vehicle.

[0008] Thirdly, this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of the above-described control method for a trackless rubber-tired vehicle in wells based on the thermal state perception of the device.

[0009] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the above-described control method for a trackless rubber-tired vehicle in wells based on equipment thermal state perception.

[0010] The beneficial effects of the control method and system for trackless rubber-tired vehicles based on equipment thermal state perception provided in this application are as follows: This application can effectively solve the pain points of trackless rubber-tired vehicles in high-temperature and complex underground environments, such as lagging thermal management, insufficient driving safety, and poor task adaptability, and has technical advantages and practical value; it combines task type and roadway digital map to plan the path, improving driving accuracy and path adaptability, and reducing ineffective driving energy consumption; by collecting environmental images, component temperature, and operating condition data in real time, it achieves dual accurate assessment of the roadway thermal environment and equipment thermal state, predicts component temperature trends in advance, avoids downtime due to overheating, and extends equipment service life; relying on a collaborative decision engine to generate composite control commands, it takes into account both driving status and thermal load adjustment, ensuring operational efficiency while reducing heat loss and safety risks, and improving the stability and reliability of underground operations. In addition, this application also realizes the integration of thermal state perception, path planning, and collaborative control, adapting to different underground working conditions, providing strong support for the intelligent and safe operation of trackless rubber-tired vehicles, and reducing the cost of manual intervention. Attached Figure Description

[0011] Figure 1 A flowchart illustrating a control method for a trackless rubber-tired vehicle in wellbore based on equipment thermal state sensing, provided in an embodiment of this application; Figure 2 A structural block diagram of a trackless rubber-tired vehicle control system based on equipment thermal state sensing provided in an embodiment of this application; Figure 3 This is a schematic block diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0012] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0013] To make the purpose, technical solution, and advantages of this application clearer, the following will be combined with Figures 1-3 The following is an explanation using specific examples.

[0014] Please refer to Figure 1 , Figure 1This is a flowchart illustrating a control method for a trackless rubber-tired vehicle in wellbore based on equipment thermal state sensing, according to an embodiment of this application. The method includes: S101: In response to the received task instruction, determine the target task type and target driving area; wherein, the target task type includes transportation task, transshipment task and inspection task.

[0015] In this embodiment, the task instruction refers to the command issued by the scheduling system or operator to instruct the trackless rubber-tired vehicle to perform operations. Its content includes information such as the task content, destination, and priority. The target task type refers to the specific type of operation the trackless rubber-tired vehicle needs to perform, such as a transportation task to move materials from point A to point B, a transfer task to move mined ore from the working face to a transfer point, and an inspection task to check underground equipment or the environment. Different task types have different requirements for the trackless rubber-tired vehicle's operating efficiency, heat load management, and safety.

[0016] Specifically, task instructions can be sent from the underground dispatch center to the onboard controller of the trackless rubber-tired vehicle via a wireless communication network, or manually entered by the operator through the onboard human-machine interface. The onboard controller parses the instruction content, extracts the operation type and its corresponding target travel area (the roadway section from the mining face to the transfer station).

[0017] S102: Based on the target task type, the digital map of the alleyway in the target driving area, and the positioning information of the trackless rubber-wheeled vehicle, plan the driving route.

[0018] In this embodiment, the tunnel digital map refers to a dataset that digitally models and represents the underground tunnel environment. This tunnel digital map includes information such as the tunnel's geometry, slope, width, obstacle distribution, ventilation conditions, and historical environmental thermal data.

[0019] Specifically, after obtaining the target task type and target travel area, starting from the current location of the trackless rubber-wheeled vehicle and aiming at the task endpoint, the A* algorithm is used to search for all reachable candidate paths on the topology map constructed from the digital map of the tunnel. The cost function for each candidate path is F=G+H, where F is the total cost of the candidate path from the current location to the task endpoint; G is the actual cost from the current location to the task endpoint, which is obtained by weighted summation of path length, estimated travel time, and cumulative heat load estimated based on historical environmental thermal data; the cumulative heat load is estimated by accumulating the product of the historical average temperature associated with each segment of the path and the segment length; H is the heuristic cost from the current location to the task endpoint, which is the Manhattan distance from the current location to the task endpoint (applicable to underground gridded tunnel layouts, the Manhattan distance reflects the actual cost of moving along the grid), or the estimated travel cost calculated in combination with the historical travel time of the corresponding segment.

[0020] When planning the route, candidate routes are first selected where the predicted temperature of the target component at the end of the route is below the first safety threshold throughout. Then, based on the pre-set weights for the target task type, the comprehensive evaluation value of each candidate route is calculated, and the route with the highest comprehensive evaluation value is selected as the driving route. The comprehensive evaluation value of each candidate route is calculated using the following formula: Comprehensive evaluation value of candidate route = Normalized driving time × Time weight + Normalized predicted temperature × Thermal safety weight + Normalized route energy consumption × Energy consumption weight. The normalization process uses the min-max standardization method to map each indicator value to the 0-1 range. The time weight, thermal safety weight, and energy consumption weight are pre-set according to the target task type, and the sum of the three weights is 1. For example, for transportation tasks: Time weight 0.6, Thermal safety weight 0.2, Energy consumption weight 0.2; for inspection tasks: Time weight 0.2, Thermal safety weight 0.6, Energy consumption weight 0.2; for reloading tasks: Time weight 0.4, Thermal safety weight 0.3, Energy consumption weight 0.3. In this embodiment, travel time, predicted temperature, and path energy consumption together constitute the comprehensive evaluation system for initial path planning. Travel time refers to the estimated time required to travel along the candidate path; predicted temperature refers to the temperature of the target component at the path's endpoint, estimated based on historical environmental thermal data and a thermal characteristic model; and path energy consumption refers to the total energy consumption required to travel along the candidate path, calculated based on path length, gradient, and estimated load. This comprehensive evaluation system is used in global path planning scenarios, focusing on comprehensively balancing travel efficiency, thermal safety, and energy economy during the task initiation phase, providing a unified quantitative basis for selecting the globally optimal path.

[0021] S103: Controls the trackless rubber-tired vehicle to travel along the driving path and collects environmental image data of the roadway ahead, temperature data of the target components of the trackless rubber-tired vehicle, and real-time operating data of the trackless rubber-tired vehicle.

[0022] In this embodiment, the trackless rubber-tired vehicle is controlled to travel along a planned path, and real-time environmental image data of the roadway ahead, temperature data of the target components of the trackless rubber-tired vehicle, and real-time operating condition data of the trackless rubber-tired vehicle are collected. The trackless rubber-tired vehicle travels automatically or semi-automatically according to the planned path.

[0023] The environmental image data is collected using a bright, enhanced visible light camera and an industrial infrared thermal imager with a protective cover, capturing visible light and infrared images of the tunnel ahead. To cope with the low-light and high-dust environment underground, the Retinex algorithm is used to enhance the low-light performance of the visible light images, and an adaptive dust removal algorithm based on dark channel priors is applied. The infrared images undergo radiometric correction and temperature compensation based on real-time temperature, humidity, and dust concentration data. Temperature data is continuously monitored using temperature sensors placed on target components (motors, hydraulic pumps, battery packs) to track their surface or internal temperature. Real-time operating condition data includes parameters such as vehicle speed, acceleration, engine speed, load, and tunnel slope.

[0024] While controlling the trackless rubber-tired vehicle to travel along the driving path, based on the geometric characteristics of the driving path, the dynamic model of the trackless rubber-tired vehicle, and the task efficiency requirements, a spline curve generation method based on speed planning is adopted, combined with slope and curvature constraints, to generate reference speeds and expected travel times at each point along the path, thus forming the expected driving trajectory.

[0025] S104: Based on the environmental image data of the roadway ahead and the expected driving trajectory corresponding to the driving path, conduct a thermal environment assessment of the roadway ahead to obtain the environmental thermal risk level.

[0026] In this embodiment, the registered visible light and infrared images are fused at the pixel level (weighted average fusion). A temperature field distribution map is generated based on the infrared image, and areas with temperatures exceeding the dynamic ambient temperature threshold (set according to the real-time ambient temperature) are identified. Semantic segmentation is performed using the visible light image to identify and exclude non-heat source interference objects (rocks, support structures), confirming abnormal heat source areas. The highest temperature, area, shape factor, and shortest Euclidean distance to the expected driving trajectory are extracted for each abnormal heat source area. These features are input into a pre-trained environmental thermal risk assessment model (a classification model based on a multilayer perceptron). The model integrates factors such as heat source intensity, size, distance, and task type to output the environmental thermal risk level. The model training data comes from manually labeled thermal environment and corresponding component temperature rise data from historical tasks.

[0027] S105: Input temperature data, real-time operating condition data, and environmental thermal risk level into the preset thermal state prediction model to obtain the predicted temperature change curve of the target component in the future period. Based on the current temperature value of the target component and the predicted temperature change curve, determine the thermal state assessment level of the trackless rubber-wheeled vehicle.

[0028] In this embodiment, the thermal state prediction model adopts a hybrid architecture combining a long short-term memory network (LSM) and the device's thermal characteristic equation. The LSM learns the dynamic mapping relationship between operating conditions, environment, and temperature from historical time-series data; the thermal characteristic equation establishes a discretized thermal balance model of the target component based on the principle of energy conservation, providing physical constraints. The two are combined through feature concatenation and fusion via a fully connected layer, jointly outputting a predicted temperature change curve for the future time period (300 seconds ahead, with a step size of 10 seconds).

[0029] The training of the thermal state prediction model includes: Initial model training is performed under supervision using paired datasets (operating conditions, environment, and temperature) collected from historical tasks, with the loss function being mean squared error. During online runtime, an incremental learning update is triggered after each complete task cycle or after accumulating 8 hours of operation: based on the deviation between the latest collected real temperature data and the predicted value, a mini-batch gradient descent algorithm is used to adjust the LSTM network weights, with the learning rate set to 0.001 to prevent overfitting.

[0030] This embodiment pre-sets four temperature threshold levels (normal temperature threshold, attention temperature threshold, warning temperature threshold, and danger temperature threshold) corresponding to the target component. These four temperature threshold levels correspond one-to-one with the thermal status assessment level. The thermal status assessment level is uniformly divided into four levels: Normal, Attention, Warning, and Danger. The Normal level corresponds to a target component temperature below the Normal temperature threshold, indicating no thermal risk. The Attention level corresponds to a target component temperature between the Normal and Attention thresholds, requiring continuous monitoring of temperature changes. The Warning level corresponds to a target component temperature between the Attention and Warning thresholds, requiring the initiation of thermal management intervention measures. The Danger level corresponds to a target component temperature greater than the Warning temperature threshold but less than the Danger temperature threshold, requiring the implementation of mandatory protection measures.

[0031] The thermal condition assessment level is determined as follows: First, the current temperature is compared with a threshold to determine the baseline level. Then, the prediction curve is analyzed: if the prediction indicates that the temperature will exceed the next lower threshold corresponding to the current baseline level within the set warning time window Δt (e.g., 120 seconds), the thermal condition assessment level is increased by one level. The final output of the increased level is used as the real-time thermal condition assessment level of the trackless rubber-wheeled vehicle.

[0032] S106: Input the environmental thermal risk level, thermal status assessment level, target task type, target driving area and real-time operating condition data into the collaborative decision engine; the collaborative decision engine makes decisions based on the pre-set thermal management strategy library and decision mechanism, and generates a collaborative control instruction set, which includes motion control instructions for adjusting the driving parameters of the trackless rubber-tired vehicle and thermal management control instructions for adjusting the working status of the heat dissipation system.

[0033] In this embodiment, the collaborative decision engine adopts a hybrid decision architecture combining rule base and model predictive control (MPC). Based on the target task type and target driving area, a pre-set baseline rule base is invoked as the starting point for decision-making. With the core constraint that the predicted temperature will remain below a first safety threshold for the next few control cycles (the next 30 seconds), and with the optimization objectives of minimizing overall thermal load and maximizing task progress, the parameter sets of motion control commands (target vehicle speed) and thermal management control commands (target speed of the cooling fan) are solved online using an MPC rolling optimization algorithm to generate a collaborative control command set. The expected control objectives of the collaborative control command set generated by the collaborative decision engine include: controlling the temperature of the target component below the target temperature value within the future timeframe, while maintaining the driving speed near the target speed value.

[0034] The collaborative control instruction set includes motion control instructions for adjusting parameters such as the speed, acceleration, and braking intensity of the trackless rubber-wheeled vehicle, and thermal management control instructions for adjusting parameters such as the cooling fan speed and coolant flow rate. Thermal management control instructions include: engine power limiting instructions, used to dynamically limit engine output power based on the thermal state assessment level and the environmental thermal risk level; and cooling system regulation instructions, used to control the fan speed and / or water pump flow rate of the radiator and hydraulic oil radiator. The core hard constraints generated by the collaborative control instructions are: a predicted temperature consistently below the second warning threshold is a priority constraint, and a temperature consistently below the first safety threshold is an absolutely insurmountable hard constraint.

[0035] Specifically, the rule base (decision starting point and boundary constraints) calls a pre-set baseline strategy template based on the target task type and target driving area. This baseline strategy template defines the initial expected vehicle speed, the baseline speed of the cooling fan, and the maximum and minimum cooling power allowed under different thermal state assessment levels.

[0036] Model predictive control (MPC) includes: Prediction model: Integrating thermal state prediction model and longitudinal dynamics model of trackless rubber-tired vehicle.

[0037] Optimization variables: vehicle speed adjustment and cooling fan power adjustment within the next N control cycles (e.g., N=6, corresponding to the next 30 seconds).

[0038] Objective function:

[0039] Where J is the value of the comprehensive cost function, the collaborative decision engine seeks the optimal balance between thermal safety, driving efficiency and energy consumption by minimizing this value; The predicted temperature of the target component is the temperature value at a future moment or time period output by the thermal state prediction model. The first safety threshold (hazardous temperature threshold) is the target temperature reference value set to ensure the safe operation of the equipment. The thermal safety deviation term quantifies the degree to which the predicted temperature deviates from the safety target, penalizing any risk that could lead to overheating; v represents the speed of the trackless rubber-wheeled vehicle, the speed during the current or future control cycle. ,in For vehicle speed adjustment amount, The desired baseline speed; The expected driving speed is set as a reference speed based on the target task type, route planning, and task time limit requirements. This is the driving efficiency deviation term, which quantifies the degree to which the actual speed deviates from the expected speed, penalizing the decrease in driving efficiency caused by excessive thermal management; p is the power of the cooling fan. This is the power adjustment amount for the cooling fan, representing the magnitude of the power change of the cooling fan relative to the current operating point. , , These are weighting coefficients, representing the emphasis on thermal safety, driving efficiency, and energy consumption, respectively, with the sum of the three being 1.

[0040] Constraints include upper and lower speed limits, cooling system power limits, and absolute hard constraints ensuring the predicted temperature remains below the first safety threshold throughout the entire process. The optimization solution follows a priority rule: vehicle speed is adjusted first; only when speed adjustment fails to meet thermal safety constraints or task time limits is the cooling system power variable adjusted simultaneously.

[0041] The weight coefficients in the objective function of the collaborative decision engine MPC in this embodiment , , The system presets a base value based on the task type and dynamically adjusts it according to the real-time thermal state and environmental risk level.

[0042] In each control cycle, the MPC optimization problem is solved to obtain the initial control sequence. Then, compensation and adjustment are made based on real-time load and slope data (higher heat dissipation power is temporarily allowed when the vehicle is heavily loaded uphill). Finally, the motion control command (target vehicle speed) and thermal management control command (target speed / power of cooling fan) for the current cycle are generated.

[0043] S107: Execute the coordinated control instruction set to adjust the driving status and thermal load status of the trackless rubber-wheeled vehicle.

[0044] In this embodiment, after receiving the coordinated control command set, the on-board controller converts it into specific execution actions. Motion control commands are sent to the drive system of the trackless rubber-tired vehicle, such as adjusting the output of the motor controller to change the vehicle's speed or acceleration. Thermal management control commands are sent to the cooling system, such as controlling the start / stop or speed of the cooling fan, and adjusting the operating mode of the coolant pump. Through these adjustments, dynamic management of the trackless rubber-tired vehicle's operating status is achieved to cope with the challenges of the complex thermal environment underground.

[0045] The execution of the coordinated control instruction set to adjust the thermal load status of the trackless rubber-wheeled vehicle also includes: for the early warning level thermal status assessment, further setting three subdivided intervention levels according to the degree of temperature exceeding the threshold, namely Level 1 warning, Level 2 warning, and Level 3 warning, the corresponding thermal management control operations are as follows: When the thermal status assessment level is Level 1 warning, increase the base speed of the cooling fan in the cooling system; for example, increase the cooling fan speed to 120% of the base speed. When the thermal status assessment level is Level 2 warning, in addition to increasing the cooling fan speed, apply a first-level soft limit to the engine output power and start the air conditioning system to provide forced ventilation cooling for the critical electrical control cabinet; for example, increase the cooling fan speed to 150% of the base speed and limit the engine power to 85% of the rated power. When the thermal status assessment level is Level 3 warning, apply a second-level hard limit to the engine output power and adjust the cooling fan and water pump to the maximum operating mode; for example, limit the engine power to 70% of the rated power and switch the cooling system to the maximum power mode.

[0046] In this embodiment, the generation of thermal management control commands and motion control commands satisfies a cooperative constraint relationship. Specifically, when it is necessary to increase the intensity of thermal management to reduce the temperature of the target component, the cooperative decision engine prioritizes generating motion control commands that reduce the speed of the trackless rubber-wheeled vehicle or reduce the acceleration frequency to reduce the heat generated by the power system. If reducing the speed cannot meet the task time limit requirements, thermal management control commands that enhance heat dissipation intensity or temporarily limit the maximum power are generated simultaneously to allow the trackless rubber-wheeled vehicle to maintain a higher speed under acceptable thermal risks.

[0047] As can be seen from the above, this application dynamically adjusts the driving state and thermal load state of the trackless rubber-tired vehicle by responding to task commands, planning driving paths, collecting multi-source data in real time, assessing the thermal environment risks ahead, predicting component temperature trends, and making collaborative decisions based on multi-dimensional information. This effectively avoids component temperature exceeding limits caused by constant heat dissipation and lack of thermal risk assessment in high-temperature and high-humidity underground roadway environments, while reducing energy waste caused by excessive heat dissipation, thus ensuring the safe and efficient operation of the trackless rubber-tired vehicle in complex underground environments.

[0048] In one embodiment of this application, a travel path is planned based on a digital map of the alleyways and the target travel area, as well as the positioning information of the trackless rubber-wheeled vehicle, including: Several candidate routes that can be reached were identified based on the digital map of the alleyways; For each candidate path, based on the historical environmental thermal data associated with the digital map of the alleyway, the environmental thermal load that the trackless rubber-tired vehicle will experience when traveling along the candidate path is estimated. Based on the thermal characteristic model of the trackless rubber-tired vehicle, the environmental heat load, and the current temperature value of the target component, the temperature rise process of the target component is predicted when traveling along each candidate path, and the predicted temperature of the target component at the end of each candidate path is obtained. Based on the weight configuration of driving efficiency and thermal safety defined by the target task type, the candidate path with a predicted temperature below the first safety threshold throughout the journey and a travel time that meets the task requirements is selected as the driving path.

[0049] In this embodiment, when several accessible candidate paths are identified based on the digital map of the alleyway, the system can use breadth-first search in graph theory to search for all reachable destinations in the topology graph constructed by the digital map of the alleyway, starting from the current position of the trackless rubber-wheeled vehicle, and record all paths that meet the connectivity requirements as candidate paths.

[0050] For each candidate route, based on historical environmental thermal data associated with the digital map of the lanes, the environmental heat load that the trackless rubber-wheeled vehicle will experience when traveling along the candidate route is estimated. Each segment of the digital map of the lanes can be associated with data such as historical average environmental temperature, historical average environmental humidity, and historical average environmental wind speed. During the estimation, the historical data of all segments on the candidate route can be weighted and summed to obtain the overall estimated environmental heat load for the candidate route.

[0051] Specifically, the weighted summation calculation rule is as follows: using the road segment length as the basic weight, the historical average ambient temperature, historical average ambient humidity, and historical average ambient wind speed associated with each road segment are weighted and calculated separately. The weighted heat load value for a single road segment is... Calculate using the following formula: ; Among them, the weight of historical average ambient temperature is 80%, the weight of historical average ambient humidity is 15%, and the weight of historical average ambient wind speed is 5%. Let be the historical average ambient temperature of the i-th road segment. The historical average ambient humidity of the i-th road segment is... Let be the historical average ambient wind speed for the i-th road segment. The final estimated overall ambient heat load is... for: ;in, Let be the length of the i-th segment, and n be the total number of segments included in the candidate path.

[0052] Based on the thermal characteristic model of the trackless rubber-tired vehicle, the estimated environmental heat load, and the current temperature value of the target component, the model predicts the temperature rise process of the target component when traveling along each candidate path, and obtains the predicted temperature of the target component at the end of each candidate path. The thermal characteristic model of the trackless rubber-tired vehicle adopts a data-driven prediction model, such as a long short-term memory network. This model learns the relationship between environmental heat load, component heat generation, heat dissipation, and temperature change in historical operating data, and can predict the temperature change trend of the target component in the future period based on the current state and the estimated environmental heat load.

[0053] Based on the weighted configuration of driving efficiency and thermal safety defined by the target task type, when selecting candidate paths from the candidate paths where the predicted temperature is below the first safety threshold throughout and the travel time meets the task requirements, we can first filter out all candidate paths where the predicted temperature is below the second warning threshold throughout. If this cannot be met, we prioritize ensuring the absolute hard constraint that the predicted temperature is below the first safety threshold throughout. Then, we further filter out these candidate paths that meet the task requirements in terms of travel time. Finally, according to the preset weights of the target task type (e.g., transportation tasks focus more on efficiency, while inspection tasks focus more on safety), we perform a weighted score on the travel time and predicted temperature of the remaining paths and select the path with the best score as the driving path. The first safety threshold (corresponding to the upper limit of the hazard level) is set comprehensively based on the manufacturer's rated maximum operating temperature of the target component, historical fault data, and safety margin. For example, it can be set to 90% of the rated temperature. The second warning threshold (corresponding to the upper limit of the warning level) is set to a value lower than the first safety threshold (e.g., 15°C) to reserve a time window for thermal management response.

[0054] As can be seen from the above, through the aforementioned technical solution, this application's system can proactively consider the environmental heat load and the temperature prediction of target components when planning the travel path of the trackless rubber-tired vehicle. By identifying multiple candidate paths and performing heat load estimation and component temperature rise prediction for each path, the thermal risks of different paths can be comprehensively assessed. Based on this, combined with the different emphases on travel efficiency and thermal safety according to the task type, the optimal travel path is selected under the premise of meeting thermal safety thresholds and task time requirements; effectively avoiding the risk of overheating of target components due to improper path selection, ensuring the safe operation of the trackless rubber-tired vehicle in the complex thermal environment underground, while also taking into account the task execution efficiency, achieving deep integration of thermal management and path planning, thereby improving the reliability and economy of trackless rubber-tired vehicle operation.

[0055] In one embodiment of this application, before or simultaneously with identifying several accessible candidate paths based on a digital roadway map, the method further includes: Based on the historical task database, retrieve historical task records that are similar to the target task type, target driving area, and average ambient temperature within a predetermined time period before the retrieval time. Extract the historical collaborative control instruction set and its corresponding final thermal state assessment results from the historical task records; Based on the final thermal state evaluation results, the historical collaborative control instruction sets are evaluated for effectiveness. The policy parameters corresponding to the historical collaborative control instruction set with the highest score are used as the initial policy parameters of the collaborative decision engine in this task.

[0056] In this embodiment, a hierarchical similar task retrieval is performed based on the historical task database: First, historical task records with the same target task type and partially or completely overlapping target driving areas are retrieved; if no such records are found, the search is broadened to only historical records with the same target task type. From the retrieved records, their historical collaborative control instruction sets and corresponding final thermal state evaluation results are extracted. Based on the final thermal state evaluation results, the instruction sets are scored, and the strategy parameters corresponding to the historical collaborative control instruction set with the highest score are used as the initial strategy parameters of the collaborative decision engine in this task. If the retrieval process takes longer than a set threshold, the default strategy parameters are directly adopted to ensure the real-time start-up of the task.

[0057] In this embodiment, the historical task database is a database or data warehouse that stores past task execution data. This database records detailed information such as the type of each task, the driving area, ambient temperature, operating conditions of the trackless rubber-tired vehicle, the cooperative control instruction set used, and the thermal status assessment results of the target components at the end of the task.

[0058] In this embodiment, upon receiving a new task instruction, a hierarchical strategy is employed to search the historical task database based on the target task type and target driving area of ​​the current task. Priority is given to matching task type and region to improve search efficiency and hit rate. The search process aims to identify historical task records highly similar to the current task's context. For example, this can be achieved by defining a similarity function that integrates factors such as task type (e.g., through exact matching), driving area (e.g., through geographical overlap or lane ID matching), and average ambient temperature (e.g., through temperature difference within a preset threshold). The similarity function can employ weighted Euclidean distance to filter historical task records with similarity greater than a certain threshold.

[0059] Identifying similar historical task records allows for the extraction of the historical collaborative control instruction sets used during execution, as well as the resulting thermal status assessment results of the target components. For example, for each similar historical task record, its stored logs or data structures are parsed to obtain the sequence of all motion control and thermal management control instructions issued by the collaborative decision engine from the start to the end of the task. Simultaneously, the final thermal status assessment level of the target component upon task completion is obtained, such as normal, attention level, warning level, and danger level.

[0060] Subsequently, based on the extracted final thermal state assessment results, each historical collaborative control instruction set is scored for its effectiveness. This scoring mechanism aims to quantify the performance of different historical strategies in thermal management and task execution. For example, a scoring rule can be set: if the final thermal state assessment result is at the normal level, a first score is assigned; if it is at the concern level, a second score is assigned; if it is at the warning level, a third score is assigned; and if it is at the danger level, a fourth score is assigned; wherein, the first score is greater than the second score, the second score is greater than the third score, and the third score is greater than the fourth score.

[0061] Finally, the strategy parameters corresponding to the highest-scoring historical collaborative control instruction set were selected as the initial strategy parameters for the collaborative decision engine in this task. These strategy parameters include, but are not limited to, initial speed limits, cooling fan control curves, coolant pump flow settings, and response thresholds under different thermal risk levels.

[0062] As can be seen from the above, through the aforementioned technical solution, the collaborative decision-making engine of this application can fully utilize historical experience when handling new tasks, avoiding random or inefficient initial strategy settings due to a lack of reference. This enables the collaborative decision-making engine to start with a validated and optimized baseline strategy, thereby improving the efficiency and accuracy of decision-making.

[0063] In one embodiment of this application, the downhole trackless rubber-tired vehicle control method based on equipment thermal state sensing further includes: During operation, based on real-time collected data, the system continuously predicts the environmental thermal risk level sequence that the trackless rubber-tired vehicle will face during the future journey, as well as the temperature evolution curve of the target components. If the predicted temperature of the target component is determined to be greater than the second warning threshold before reaching the end of the mission based on the temperature evolution curve, then path replanning is triggered to obtain the updated driving path. The second warning threshold is lower than the first safety threshold.

[0064] Starting from the current position of the trackless rubber-wheeled vehicle, and using the constraint that the predicted temperature will remain below a first safety threshold throughout the journey, several alternative local paths are replanned. The comprehensive evaluation value of each alternative local path and the remaining portion of the original path is calculated. Only when the comprehensive evaluation value of the optimal alternative local path is significantly better than the comprehensive evaluation value of the remaining portion of the original path by a preset hysteresis threshold is the updated path established and the switching process executed. This prevents frequent path switching due to minor fluctuations in environmental parameters or prediction errors.

[0065] In this embodiment, the hysteresis threshold is not a fixed value, but a dynamic or semi-dynamic parameter based on a comprehensive evaluation of multiple factors. Its determination method includes: Baseline threshold setting: Based on historical operating data or simulation tests, statistically analyze the typical performance improvement (such as percentage reduction in thermal risk, energy savings, etc.) brought about by path replanning decisions under ideal, undisturbed conditions. Set a certain percentage of this improvement (e.g., 50%-70%) as the initial static hysteresis threshold, serving as a baseline value.

[0066] Dynamic adjustment factors include: Prediction model confidence: Real-time evaluation of the recent prediction accuracy of the thermal state prediction model and the environmental thermal risk assessment model. If the deviation between the model prediction results and the actual observations is consistently small (high confidence), the hysteresis threshold can be appropriately reduced to make the system more sensitive to path optimization; conversely, if the prediction fluctuates greatly and the confidence is low, the hysteresis threshold should be increased to enhance decision stability.

[0067] Environmental volatility: Monitor the real-time rate of change of parameters such as ambient temperature and abnormal heat source distribution in the roadway. If the environment is in a state of rapid or disordered change, the hysteresis threshold should be increased to prevent the system from overreacting to environmental noise.

[0068] Task execution phase: At the beginning of the task, the uncertainty of predicting the future path is relatively high, so a higher hysteresis threshold can be set; as the task progresses, the vehicle position and environmental information become more certain, so the threshold can be gradually reduced to allow for more refined path optimization.

[0069] Controlling switchover costs: Estimate the additional control overhead, time delay, and mechanical wear and tear incurred by the path switch itself. The hysteresis threshold should be set to ensure that the overall benefits of the new path outweigh the cost of the switch.

[0070] In actual operation, the baseline hysteresis threshold is first invoked, and then adjusted in real time according to the above dynamic factors (through predefined weights or fuzzy rules) to form the final effective hysteresis threshold applied to the current decision cycle.

[0071] In this embodiment, rolling predictions are continuously made based on real-time collected data during the driving process. The real-time collected data refers to various types of data continuously acquired by the onboard sensor system during the actual operation of the trackless rubber-tired vehicle, such as environmental image data of the road ahead, temperature data of the target components of the trackless rubber-tired vehicle, and real-time operating condition data of the trackless rubber-tired vehicle.

[0072] The predicted future timeframes are rolled forward as the trackless rubber-wheeled vehicle moves. For example, the prediction can be updated every 10 meters, forecasting the thermal risk and temperature trend for the next minute or the next 100 meters. Through this rolling prediction, the sequence of environmental thermal risk levels that the trackless rubber-wheeled vehicle will face in the future journey, as well as the temperature evolution curves of target components, can be obtained.

[0073] The environmental thermal risk level sequence refers to the predicted environmental thermal risk level of each road segment that a trackless rubber-wheeled vehicle will traverse sequentially over a future journey, representing the dynamic distribution of the thermal environment along the path ahead. It is implemented by combining digital roadway maps with historical environmental thermal data and real-time environmental image data, using machine learning models to assess the thermal environment along the future path, and generating a risk level list sorted by time or distance. The functions of the environmental thermal risk level sequence include: providing environmental thermal input parameters for the rolling prediction of the temperature evolution curve of target components; providing a basis for triggering path replanning and assessing the thermal risk of alternative paths; and providing advance prediction support for the real-time control strategy optimization of the collaborative decision engine to predict the environmental thermal risk ahead.

[0074] The temperature evolution curve of the target component refers to the predicted trajectory of the temperature change of the target component of the trackless rubber-tired vehicle over time or distance during a future journey.

[0075] If the predicted temperature of the target component, based on the temperature evolution curve, is expected to exceed the second warning threshold before reaching the mission endpoint, a path replanning evaluation process is triggered. Using the current position of the trackless rubber-wheeled vehicle as the new starting point, and with the predicted temperature consistently below the first safety threshold as a hard constraint, several alternative local paths are replanned. The replanned alternative local paths are compared with the remaining portion of the original path to determine whether a path switch should be executed. To ensure consistency in evaluation standards, the comparison process uniformly uses the evaluation value system from the replanning phase (thermal risk index, estimated travel time, and path length) to calculate the comprehensive evaluation value for both. The thermal risk index, estimated travel time, and path length of the remaining portion of the original path are obtained through rolling prediction based on the current vehicle position, remaining path information, and real-time collected environmental and operating condition data. Only when the comprehensive evaluation value of the optimal alternative local path is significantly better than the comprehensive evaluation value of the remaining portion of the original path by a pre-set hysteresis threshold is it identified as the updated travel path and the switch executed, preventing frequent path switching due to minor fluctuations in environmental parameters or prediction errors. The second warning threshold is a pre-set temperature value, lower than the first safety threshold but higher than the normal operating temperature. When the predicted temperature of the target component is greater than or equal to the second warning threshold, it indicates a potential risk of thermal overload, requiring early intervention. The second warning threshold can be set based on factors such as the component's heat resistance limit, safety margin, and allowable response time.

[0076] Triggered path replanning refers to the process of starting a new path planning process instead of continuing along the original route when certain conditions are met (i.e., the predicted temperature exceeds the second warning threshold).

[0077] As can be seen from the above, through the aforementioned technical solution, this application, based on rolling prediction and conditional triggering mechanisms, achieves dynamic monitoring and path adjustment of the thermal state of the trackless rubber-tired vehicle during operation, thereby effectively ensuring thermal safety. During operation, the rolling prediction of the environmental thermal risk level sequence and the temperature evolution curve of the target component based on real-time collected data can promptly capture environmental changes and temperature trends, providing a basis for dynamic decision-making and solving the problem that initial predictions may fail due to sudden environmental changes. If, based on the temperature evolution curve, it is determined that the predicted temperature of the target component will exceed the second warning threshold before reaching the mission endpoint, the path replanning evaluation process is triggered. Through comprehensive evaluation value comparison and hysteresis thresholds, the necessity and stability of the replanning decision are ensured.

[0078] In one embodiment of this application, if the predicted temperature of the target component is determined to be greater than a second warning threshold before reaching the mission endpoint based on the temperature evolution curve, path replanning is triggered to obtain an updated driving path, including: A path optimization model is established with the weighted sum of task completion time and total energy consumption as the optimization objective and the predicted temperature being less than the first safety threshold throughout the process as a hard constraint. When the path replanning condition is triggered, at least one alternative local path is generated in the digital map of the alleyway based on the current position of the trackless rubber-tired vehicle and the destination of the task. Based on the updated environmental thermal risk level, thermal state assessment level, real-time operating data, and path optimization model, the comprehensive evaluation value of each alternative local path is calculated. The optimal candidate local path is selected as the updated driving path based on the comprehensive evaluation value. The optimal candidate path is determined as the updated driving path only if the comprehensive evaluation value of the optimal candidate path is better than the remaining part of the original path by a factor greater than the preset hysteresis threshold.

[0079] In this embodiment, the path optimization model aims to provide a quantitative decision-making framework for path selection. This model weights and sums task completion time and total energy consumption to form a comprehensive optimization objective function, balancing efficiency and economy. For example, different weighting coefficients can be set to adjust the relative importance of task completion time and total energy consumption in the optimization objective; for instance, in emergency tasks, the weight of task completion time can be higher. Simultaneously, the model is based on a hard constraint that the predicted temperature must remain below a first safety threshold throughout the entire journey. Any selected path must ensure that the predicted temperature of the target component remains below the first safety threshold throughout the entire journey, thus forcibly guaranteeing the thermal safety of the trackless rubber-wheeled vehicle. The path length is a simplified substitute for the total energy consumption in local short-distance replanning scenarios, while the total energy consumption index is still used in long-distance replanning scenarios.

[0080] When the path replanning condition is triggered, based on the trackless rubber-tired vehicle's current position and the task endpoint, at least one alternative local path is generated using a path search algorithm within a pre-constructed digital map of the lanes. The digital map of the lanes includes the lane topology, geometric information, and information on existing obstacles. The path search algorithm, such as the A* algorithm, can find feasible paths from the current position to the task endpoint in complex lane networks. Generating alternative local paths aims to provide multiple possible travel options.

[0081] Subsequently, based on the updated environmental thermal risk level, thermal state assessment level, real-time operating data, and the aforementioned path optimization model, a comprehensive evaluation value was calculated for each candidate local path. Based on the updated environmental thermal risk level, thermal state assessment level, real-time operating data, and in conjunction with the optimization objectives and hard constraints defined in the path optimization model, a simulation evaluation was performed on each candidate local path, and the comprehensive evaluation value was calculated.

[0082] Finally, based on the calculated comprehensive evaluation values ​​of each candidate local path, the optimal candidate local path is selected as the updated driving path. The selection criterion for the optimal path is the minimum (or maximum, depending on the definition of the optimization objective function) comprehensive evaluation value, that is, the path that minimizes the weighted sum of task completion time and total energy consumption while satisfying the thermal safety hard constraints.

[0083] As can be concluded from the above, this application provides a path replanning mechanism through the aforementioned technical solution. This path replanning mechanism, when the trackless rubber-tired vehicle faces potential overheating risks, comprehensively considers task efficiency, energy consumption, and thermal safety. It establishes a multi-objective optimization model and combines it with real-time data for path evaluation and selection. This not only avoids the inefficiency or increased energy consumption that may result from simple replanning, but more importantly, by using a hard constraint that the predicted temperature remains below a first safety threshold throughout the entire process, it forcibly ensures the operational safety of the trackless rubber-tired vehicle in the complex thermal environment downhole. Simultaneously, it dynamically generates alternative local paths and calculates evaluation values ​​based on updated environmental thermal risk levels, thermal state assessment levels, and real-time operating data, ensuring the real-time nature, accuracy, and comprehensiveness of path selection, thereby improving the overall operational efficiency and thermal management level of the trackless rubber-tired vehicle in downhole operations.

[0084] In one embodiment of this application, the downhole trackless rubber-tired vehicle control method based on equipment thermal state sensing further includes: Several virtual evaluation points are set at equal intervals on the alternative local paths or according to the characteristics of the roadway environment; For each alternative local path, the thermal state prediction model, the estimated environmental heat load corresponding to each virtual evaluation point and the thermal characteristic model of the trackless rubber-tired vehicle are used to simulate and deduce the temperature of the target component when the trackless rubber-tired vehicle travels to each virtual evaluation point, and obtain the component temperature sequence. Based on the component temperature sequence, a thermal risk index is calculated to characterize the overall thermal safety level of the alternative local path. In the route optimization model, the thermal risk index, route length, and estimated travel time are combined to form a multi-objective optimization function for solving, thereby obtaining a comprehensive evaluation value of the candidate local routes.

[0085] In this embodiment, several virtual assessment points are set up along the candidate local path to extend the thermal risk assessment of the path from a single endpoint to the entire path range. These virtual assessment points can be set up at equal intervals, for example, one assessment point is set up at every fixed distance (such as 10 meters or 20 meters) to ensure uniform coverage of the path. Alternatively, they can be set up according to the characteristics of the roadway environment. For example, in areas with known or predicted high temperatures, poor ventilation, long uphill sections, or other areas with high heat loads, the density of virtual assessment points can be increased, while in areas with relatively stable thermal environments, they can be appropriately decreased.

[0086] For each alternative local path, the thermal state prediction model, the estimated environmental heat load corresponding to each virtual evaluation point, and the thermal characteristic model of the trackless rubber-tired vehicle are used to conduct simulation to obtain the temperature of the target component when the trackless rubber-tired vehicle travels to each virtual evaluation point, and thus obtain the component temperature sequence.

[0087] The thermal characteristic model of the trackless rubber-tired vehicle includes information such as the heat generation characteristics, heat dissipation characteristics, and heat transfer paths of each component. By inputting this information into the simulation model, the temperature change process of the target component as it travels along the path can be dynamically simulated, thus obtaining a sequence characterizing the temperature change of the component with position.

[0088] Based on the component temperature sequence, a thermal risk index is calculated to characterize the overall thermal safety level of the candidate local path. This thermal risk index is a comprehensive quantitative value used to assess the thermal safety level of the entire path. For example, the thermal risk index can comprehensively consider factors such as the highest temperature value in the component temperature sequence, the duration of temperature exceeding the warning threshold, and the cumulative deviation between the temperature and the safety threshold. Specifically, a weighted function can be designed to sum the above factors to obtain the thermal risk index.

[0089] In scenarios triggering route replanning, since the vehicle is already in operation and the replanning focuses on local route adjustments, the decision-making objective emphasizes rapid response to sudden changes in the thermal environment ahead while meeting thermal safety constraints. Therefore, the comprehensive evaluation system for the route replanning stage is adjusted to include three parameters: thermal risk index, estimated travel time, and route length. The thermal risk index is a quantitative value of the overall thermal safety level calculated based on the predicted temperature sequence of target components at each virtual evaluation point along the route. Its function corresponds to the "predicted temperature" in the initial planning, but it more comprehensively reflects the heat load distribution of the entire route. The estimated travel time uses the same calculation method as the travel time in the initial planning, both based on the path geometry and vehicle dynamics model. Route length serves as a simplified substitute for route energy consumption in local route replanning because, within the local adjustment range, route length and energy consumption are strongly positively correlated, and the calculation is more suitable for real-time dynamic scenarios. This parameter adjustment is to adapt to the specific needs of replanning scenarios for real-time performance and refined assessment of local thermal risks, rather than replacing the initial planning system. In the route optimization model, thermal risk indicators, route length, and estimated travel time are combined to form a multi-objective optimization function for solution. Route selection no longer merely pursues the shortest path or the shortest time, but instead considers thermal safety as an optimization objective of equal importance to efficiency. The multi-objective optimization function can employ a weighted sum method, that is, by assigning different weights to thermal risk indicators, route length, and estimated travel time, multiple objectives are transformed into a single optimization objective for solution. Specifically, the multi-objective optimization function is transformed into a single-objective optimization problem using a linear weighted method. Solving this problem yields the comprehensive evaluation value of each candidate local path, which is used for path selection. The expression of the multi-objective optimization function is: Comprehensive evaluation value of candidate local path = Normalized value of thermal risk index × Thermal risk weight + Normalized value of estimated travel time × Time weight + Normalized value of path length × Path length weight; where the sum of the three weights is 1. Normalization is performed using the min-max standardization method, mapping each index value to the 0-1 interval. The solution process uses the A* algorithm, with the goal of minimizing the objective value, to obtain the optimal solution and corresponding comprehensive evaluation value for each candidate local path. For the three parameters in the comprehensive evaluation value of the replanning stage, the weights are preset according to the target task type, and the specific configurations are as follows: Transportation task: thermal risk weight 0.4, time weight 0.5, path length weight 0.1; Inspection task: thermal risk weight 0.7, time weight 0.2, path length weight 0.1; Re-transfer task: thermal risk weight 0.5, time weight 0.3, path length weight 0.2.During operation, the thermal risk weight is dynamically adjusted based on the real-time thermal status assessment level (for example, when the thermal status assessment level is upgraded from normal to warning level, the thermal risk weight can be temporarily increased by 10%-20%). While the thermal risk weight is increased, the time weight and path length weight are simultaneously reduced according to the initial weight ratio to ensure that the sum of the three weights is always 1, thereby enhancing the system's response sensitivity to thermal risks.

[0090] As can be seen from the above, this application, through the aforementioned technical solution, overcomes the limitation of incomplete thermal risk assessment in traditional path optimization. By setting virtual assessment points on candidate local paths and simulating the temperature sequence of target components using thermal state prediction models and thermal characteristic models, a refined perception of the dynamic changes in thermal load along the path is achieved. The thermal risk index calculated based on this temperature sequence can comprehensively and quantitatively characterize the overall thermal safety level of candidate local paths, avoiding the one-sidedness of relying solely on a single temperature point. Finally, this thermal risk index, along with path length and estimated travel time, is incorporated into a multi-objective optimization function for solution. This ensures that path optimization decisions not only consider travel efficiency but also take thermal safety as the core constraint and optimization objective. Consequently, when path replanning is triggered, an optimal path can be selected that meets mission requirements while effectively avoiding thermal risks, ensuring that the temperature of the trackless rubber-tired vehicle's target components remains within a safe range throughout the entire journey. This improves the operational reliability, safety, and mission adaptability of the underground trackless rubber-tired vehicle in complex thermal environments.

[0091] In one embodiment of this application, the environmental image data includes visible light images and infrared images; Based on the environmental image data of the roadway ahead and the expected driving trajectory corresponding to the driving path, a thermal environment assessment of the roadway ahead is performed to obtain the environmental thermal risk level, including: The acquired visible light and infrared images are fused together to identify abnormal heat source areas in the roadway ahead. Based on the temperature, geometric dimensions of the abnormal heat source area and its relative position to the expected driving trajectory, the environmental thermal risk level is calculated using an environmental thermal risk assessment model.

[0092] In this embodiment, abnormal heat source regions refer to areas with temperatures higher than the surrounding environment or expected normal temperatures; these are potential thermal risk points. The fusion processing can employ a feature-level fusion method. Visible light and infrared images are aligned using image registration techniques, and then algorithms such as weighted averaging, wavelet transform, and gradient domain fusion are used to generate a fused image that includes visible light details and infrared temperature information. On this fused image, image processing techniques such as threshold segmentation and connected component analysis are used to identify regions with abnormal temperatures and specific geometric features as abnormal heat source regions.

[0093] Furthermore, based on the temperature, geometric dimensions, and relative positional relationship between the abnormal heat source region and the expected driving trajectory, the environmental thermal risk level is calculated using an environmental thermal risk assessment model. This step aims to transform the objective physical attributes of the identified abnormal heat source region into a quantified environmental thermal risk level, providing a basis for decision-making regarding trackless rubber-wheeled vehicles. By comprehensively considering temperature, size, and positional relationships, the impact of the heat source on the thermal safety of the trackless rubber-wheeled vehicle can be fully assessed.

[0094] The environmental thermal risk assessment model is a classification model based on a multilayer perceptron, used to quantify the thermal environment characteristics of the roadway ahead into discrete environmental thermal risk levels.

[0095] Feature extraction and input include: heat source features, basic environmental features, and task context features. Specifically, for heat source features, an improved YOLOv5 detection framework is used to identify abnormal heat source regions in the registered and fused visible-infrared images. The basic YOLOv5 detection framework is a well-known target detection technology. The improved YOLOv5 framework is primarily designed for scenarios with low illumination, high dust levels, and small-scale heat source targets in underground mines. Specific improvements include: adding an image preprocessing module at the input end, integrating Retinex low-light enhancement and dust removal algorithms to improve input image quality; re-clustering and generating suitable anchor box parameters based on the size characteristics of underground heat source targets to improve the detection accuracy of small target heat sources; and optimizing the loss function by introducing Focal Loss to address the imbalance between positive and negative samples, thereby improving the recall rate for identifying abnormal heat sources. For each detection region, the following features are extracted: the region's highest temperature, average temperature, pixel area (converted to actual area), shape factor (perimeter squared / area), and the shortest Euclidean distance from its centroid to the expected trajectory. Basic environmental characteristics: Real-time collected data on tunnel ambient temperature, humidity, and wind speed. Task context characteristics: Target task type (e.g., transportation, inspection) encoded as a one-hot vector.

[0096] The features of all identified heat source regions are aggregated (e.g., the maximum value or average value of each feature is taken) and concatenated with the basic environmental features and task context features to form a fixed-dimensional input feature vector.

[0097] The feature vector is input into the trained classification model, and the output layer of the model uses the Softmax function to obtain the confidence level of each risk level. The level with the highest confidence level is taken as the environmental thermal risk level of the roadway ahead.

[0098] The environmental thermal risk assessment model is trained using manually labeled thermal environment-risk level data from historical tasks. The labeling is based on the actual temperature rise rate of components and whether overheating alarms are triggered after the trackless rubber-wheeled vehicle operates in that environment. Based on preset risk tolerance coefficients for different task types, the weights of temperature-related features used in model classification or the confidence thresholds in post-processing are dynamically adjusted to ensure that the risk assessment results match the task safety strategy.

[0099] As can be seen from the above, the present application can improve the accuracy and reliability of thermal environment assessment in underground roadways through the above technical solution.

[0100] In one embodiment of this application, the acquired visible light image and infrared image are fused to identify abnormal heat source areas in the roadway ahead, including: A temperature field distribution map is generated based on infrared images, and areas with temperatures higher than the ambient temperature threshold are marked. Spatial registration and semantic analysis were performed based on visible light images to eliminate interference and confirm the final abnormal heat source area. Extract the extreme temperature values, area, shape factor, and shortest distance from the expected driving trajectory of each abnormal heat source region.

[0101] In this embodiment, when generating a temperature field distribution map based on infrared images and identifying areas with temperatures exceeding the ambient temperature threshold, an infrared thermal imager can be used to acquire infrared images. Algorithms such as radiometric correction and temperature conversion are then used to convert the pixel values ​​of the images into actual temperature values, thereby constructing a temperature field distribution map of the roadway ahead. When identifying areas with temperatures exceeding the ambient temperature threshold, a fixed ambient temperature threshold based on the average underground environmental temperature or a safety standard can be preset. All areas in the temperature field distribution map exceeding this threshold are marked as potential heat sources. Alternatively, the ambient temperature threshold can be dynamically adjusted based on real-time collected data from underground environmental temperature sensors to adapt to environmental changes and improve the accuracy of the calibration.

[0102] In spatial registration and semantic analysis based on visible light images to eliminate interfering objects and confirm the final anomalous heat source region, spatial registration can employ a feature point matching method to precisely align the visible light image and infrared image geometrically, ensuring a pixel-level correspondence between them in space. Semantic analysis can utilize semantic segmentation models from deep learning to analyze the registered visible light image and identify potential non-heat source objects, such as rocks, support structures, water stains, other trackless rubber-tired vehicle components, and personnel. When eliminating interfering objects and confirming the final anomalous heat source region, a logical comparison is performed between the potential heat source region initially identified in the infrared image and the non-heat source object region identified by the semantic analysis of the visible light image. If the potential heat source region overlaps with the non-heat source object region, the overlapping portion or the entire potential heat source region is excluded from the anomalous heat source list, thus confirming and retaining those anomalous heat regions truly caused by heat sources.

[0103] When extracting the extreme temperatures, area, shape factor, and shortest distance from the expected travel trajectory for each abnormal heat source region, the highest temperature value within each confirmed abnormal heat source region can be extracted from the temperature field distribution map as its thermal intensity characteristic. For area and shape factor, the actual physical area can be converted by calculating the number of pixels in the abnormal heat source region and combining it with the pixel size. The shape factor can be described by calculating geometric features such as the ratio of the region's perimeter to its area or the ratio of its principal axis length, to characterize the heat source's morphological features. For the shortest distance from the expected travel trajectory, based on the real-time positioning information of the trackless rubber-wheeled vehicle and the pre-planned expected travel trajectory, the Euclidean distance from the geometric center or boundary point of each abnormal heat source region to the expected travel trajectory is calculated, and the minimum value is taken as the shortest distance to assess the potential impact of the heat source on the trackless rubber-wheeled vehicle's movement.

[0104] As can be seen from the above, this application effectively solves the problem of misjudgment that occurs when relying solely on infrared images to identify abnormal heat sources. First, a temperature field distribution map is generated based on the infrared image, and potential heat source areas are initially identified, providing a foundation for heat source identification. Subsequently, spatial registration and semantic analysis using visible light images can accurately identify and eliminate non-heat source interference objects that may exist in the infrared image, such as rocks, support structures, or water stains, thereby ensuring that the identified abnormal heat source areas are genuine heat sources. Based on this, key features such as the temperature extremes, area, shape factor, and shortest distance from the expected travel trajectory of each abnormal heat source area are further extracted, providing more accurate and reliable input data for subsequent environmental thermal risk assessment. This further improves the accuracy of the thermal environment assessment of the roadway ahead, avoids excessive or insufficient thermal management due to misjudgment, and thus improves the reliability of thermal risk control for trackless rubber-wheeled vehicles.

[0105] In one embodiment of this application, based on the temperature, geometric dimensions of the abnormal heat source area, and the relative positional relationship between the abnormal heat source area and the expected driving trajectory, an environmental thermal risk level is calculated using an environmental thermal risk assessment model, including: For each abnormal heat source region, a regional risk feature vector is constructed based on its extracted features; The regional risk feature vector is input into the environmental thermal risk assessment model to obtain the local risk level of each region. By integrating the local risk levels and spatial distribution of all abnormal heat source areas, the global environmental thermal risk value is calculated. Based on the risk tolerance coefficient corresponding to the target task type, the global environmental thermal risk value is corrected and mapped to the environmental thermal risk level.

[0106] In this embodiment, parameters such as the extreme temperature value (e.g., the highest temperature), geometric dimensions (e.g., area, aspect ratio) of the abnormal heat source region, and its shortest distance from the expected driving trajectory are used as features and combined into a multi-dimensional vector. For example, this vector can be represented as [highest temperature, area, shape factor, shortest distance].

[0107] Subsequently, the regional risk feature vector is input into the environmental thermal risk assessment model to obtain the local risk level of each region. The environmental thermal risk assessment model is a mathematical or algorithmic model used to transform the quantified regional risk feature vector into an understandable local risk level, which characterizes the impact of a single anomalous heat source region on the thermal safety of trackless rubber-tired vehicles. This model can be a machine learning model, such as a decision tree, trained on historical data to learn the mapping relationship between anomalous heat source characteristics and local risk levels. After inputting the regional risk feature vector, the model outputs the corresponding local risk level, such as a continuous value from 0 to 100 or discrete low, medium, and high levels.

[0108] Based on this, the local risk levels and spatial distribution of all abnormal heat source areas are integrated to calculate the global environmental thermal risk value. The global environmental thermal risk value aims to comprehensively consider the overall risk level of all abnormal heat source areas in the roadway ahead, rather than just the risk of a single heat source. Fusion needs to be based on the spatial distribution of heat sources, as multiple low-risk heat sources may constitute a high-risk area in a specific spatial arrangement. One implementation method is to perform a weighted summation of all local risk levels, with the weights determined based on factors such as the distance and size of the heat sources. Simultaneously, a spatial density factor can be used; for example, if multiple locally high-risk areas are densely distributed within a short distance of the expected travel trajectory, the risk value of that area is additionally weighted. Another implementation method is to use all local risk levels and their coordinates in the roadway digital map as input, and fuse them through a grid-based analysis algorithm. For example, the roadway is divided into several grids, the sum or maximum value of the local risk levels within each grid is calculated, and then the grid risks over the entire expected travel trajectory are integrated or averaged to obtain the global environmental thermal risk value.

[0109] Finally, the global environmental thermal risk value is corrected based on the risk tolerance coefficient corresponding to the target task type and mapped to an environmental thermal risk level. The risk tolerance coefficient is a parameter that adjusts the level of acceptance of environmental thermal risk based on the current task type performed by the trackless rubber-wheeled vehicle. Different task types have different emphases on thermal safety and driving efficiency, and therefore different tolerances for risk. The corrected global environmental thermal risk value will ultimately be mapped to discrete environmental thermal risk levels.

[0110] In this embodiment, a risk tolerance coefficient for different task types is preset as a parameter to quantify the acceptable degree of thermal risk (or safety redundancy) for different work tasks. This parameter directly affects the output correction of the environmental thermal risk assessment model and the tendency of the collaborative decision engine to balance thermal safety and driving efficiency.

[0111] The method for setting and determining the risk tolerance coefficient is as follows: Preset coefficients based on task attributes: For transportation tasks, efficiency and timeliness are prioritized. A higher risk tolerance coefficient (e.g., 1.1-1.3) is preset, meaning that in environmental thermal risk assessment and collaborative control decisions, a relatively high thermal load can be tolerated to maintain a higher travel speed, or a relatively low risk level can be assigned to the same thermal environment during assessment. For transshipment tasks, efficient turnover between fixed points is emphasized, requiring a balance between efficiency and continuous equipment operational stability. A medium risk tolerance coefficient (e.g., 1.0) is preset as a baseline balance point. For inspection tasks, safety and equipment integrity are the highest priority, with relatively relaxed travel speed requirements. A lower risk tolerance coefficient (e.g., 0.7-0.9) is preset, making the system more sensitive to thermal anomalies and tending to adopt more conservative thermal management strategies and safer travel modes, even at the cost of some efficiency. For emergency or special tasks: depending on the specific urgency and safety requirements, extremely high or extremely low coefficients may be preset.

[0112] Under the same task type, the risk tolerance coefficient is adjusted according to real-time conditions. If the vehicle's thermal condition assessment level is already high, the risk tolerance coefficient can be temporarily lowered, prioritizing safety regardless of the task type. When an extremely high thermal risk level is detected in the roadway environment, the coefficient benchmark for all task types can be automatically lowered, entering a global conservative mode.

[0113] The formula for correcting the global environmental thermal risk value is: Corrected global environmental thermal risk value = Global environmental thermal risk value × Risk tolerance coefficient. Then, based on the corrected risk value and a preset threshold range, it is mapped to low, medium, or high environmental thermal risk levels. When the thermal status assessment level reaches the warning level or above, the risk tolerance coefficient is forcibly corrected to ≤1.0, prioritizing thermal safety.

[0114] As can be seen from the above, through the aforementioned technical solution, this application, by constructing regional risk feature vectors, can quantitatively analyze the specific attributes of each anomalous heat source region, ensuring the targeted nature of risk assessment and avoiding errors caused by general assessments. Inputting these feature vectors into the environmental thermal risk assessment model yields the local risk level of each region, achieving scientific calculation and standardized processing of risk, facilitating the identification of local high-risk points. Furthermore, by integrating the local risk levels and spatial distribution of all anomalous heat source regions, the global environmental thermal risk value is calculated, avoiding the neglect of the complexity of the global thermal environment. Finally, based on the risk tolerance coefficient corresponding to the target task type, the global environmental thermal risk value is corrected and mapped to a level, enabling the risk level to dynamically adapt to the safety requirements of different tasks.

[0115] In one embodiment of this application, environmental thermal risk level, thermal state assessment level, target task type, target driving area, and real-time operating condition data are input into the collaborative decision engine to generate a collaborative control instruction set, including: Based on the target task type and target driving area, the preset baseline thermal management strategy is invoked as the starting point for decision-making; Based on the environmental thermal risk level and thermal state assessment level, the control parameters in the baseline thermal management strategy are optimized at the first level through a collaborative decision engine to obtain the first-level optimized parameter set. Based on the load and roadway slope data in the real-time operating data, the first-level optimization parameter set is adjusted in the second level to generate the final parameters for motion control commands and thermal management control commands.

[0116] In this embodiment, when a preset baseline thermal management strategy is invoked as the starting point for decision-making based on the target task type and target driving area, the baseline thermal management strategy aims to provide an initial, general control framework for subsequent decisions, ensuring the efficiency of the decision-making process and its compatibility with task characteristics. For example, a strategy library can be pre-established, including thermal management strategy templates for different target task types and different target driving areas. When a specific task instruction is received, the most suitable strategy is selected from this strategy library as the initial decision starting point based on the task type and driving area.

[0117] Based on this, and considering the current environment and the thermal state of the trackless rubber-tired vehicle itself, the baseline thermal management strategy is initially adjusted to ensure the stability of the equipment when facing thermal risks. For example, a multi-objective optimization algorithm can be used, taking the environmental thermal risk level and thermal state assessment level as input variables, to dynamically adjust various control parameters in the baseline strategy (such as heat dissipation power, fan speed, coolant flow rate, engine load limit, etc.) to generate a preliminary optimized parameter set.

[0118] Subsequently, based on the load and roadway gradient data from real-time operating conditions, a second-level adjustment is performed on the primary optimization parameter set to generate the final parameters for motion control and thermal management control commands. This second-level adjustment aims to incorporate real-time operating conditions and refine the parameters after primary optimization to maximize driving efficiency while ensuring thermal safety. The adjustment process follows a priority rule of first reducing speed and heat generation, and then enhancing heat dissipation only when time constraints cannot be met. For example, when real-time operating conditions show the trackless rubber-wheeled vehicle is heavily loaded uphill, even if primary optimization has considered thermal risks, the motion control commands still need to be adjusted to allow for a short-term increase in engine torque output to overcome the gradient, while simultaneously enhancing thermal management control commands, such as instantaneously increasing the cooling fan speed or coolant pump power, to cope with the additional heat generated. Conversely, when descending a slope unloaded, the workload of the cooling system is reduced to save energy. This adjustment can be achieved through a preset operating condition correction factor table, which adjusts parameters such as speed, torque, and cooling power in the primary optimization parameter set by percentage or absolute value according to different combinations of load and gradient. Another approach is to use predictive or adaptive control algorithms, combined with the dynamic and thermal characteristic models of the trackless rubber-tired vehicle, to iteratively adjust the primary optimization parameter set in real time based on load and gradient data, thereby generating motion control commands and thermal management control commands that are best suited to the current working conditions.

[0119] As can be seen from the above, through the aforementioned technical solution, this application uses a pre-set baseline thermal management strategy as the starting point for decision-making based on the target task type and target driving area, providing an efficient initial framework for decision-making and ensuring the initial matching of the strategy with task characteristics and regional environment. Secondly, a first-level optimization is performed based on the environmental thermal risk level and thermal state assessment level, enabling the control strategy to prioritize thermal safety risks and ensure the operational stability of the trackless rubber-tired vehicle in complex thermal environments. Finally, a second-level adjustment is performed based on the load and roadway slope data in real-time operating data, further refining the control parameters. This allows the generated motion control commands and thermal management control commands to more accurately adapt to actual driving conditions, thereby effectively balancing driving efficiency and energy consumption while ensuring equipment thermal safety, and improving the adaptive control capability and overall operating performance of the trackless rubber-tired vehicle under complex and variable underground conditions.

[0120] In one embodiment of this application, the downhole trackless rubber-tired vehicle control method based on equipment thermal state sensing further includes: Collect actual operating data, actual temperature data of target components, and environmental data after executing the collaborative control instruction set, as strategy execution feedback data; The strategy execution feedback data is compared with the expected control objectives to generate evaluation indicators that include prediction deviation and control deviation. Based on evaluation indicators, online optimization of parameters or logic is performed on the decision rule base in the thermal state prediction model, environmental thermal risk assessment model, and / or collaborative decision engine.

[0121] In this embodiment, actual operating data refers to the actual motion state and working condition information of the trackless rubber-tired vehicle after executing the coordinated control instruction set. The actual temperature data of the target component refers to the actual temperature value of the target component of the trackless rubber-tired vehicle (e.g., motor, battery pack, hydraulic system, transmission system, etc.) during operation, which is monitored and collected in real time by temperature sensors (e.g., thermocouples, thermistors, infrared temperature sensors) installed on the surface of the target component.

[0122] Environmental data refers to the real-time physical parameters of the operating environment of the trackless rubber-tired vehicle, including but not limited to ambient temperature, humidity, wind speed, and air pressure within the tunnel. This data can be collected through onboard environmental sensors (e.g., temperature and humidity sensors, wind speed sensors) or underground environmental monitoring systems.

[0123] The strategy execution feedback data is a collection of the above-mentioned types of actual collected data, used to comprehensively characterize the trackless rubber-tired vehicle's operating status, component thermal state, and environmental conditions after the execution of the collaborative control instruction set.

[0124] The expected control objective refers to the ideal operating state and thermal management objective set when the coordinated control instruction set is generated.

[0125] Prediction bias refers to the difference between the prediction results of a thermal state prediction model or an environmental thermal risk assessment model and the actual observed values. For example, the difference between the predicted temperature and the actual temperature of a target component, or the difference between the predicted environmental thermal risk level and the actual assessed environmental thermal risk; this can be obtained by calculating the absolute error between the predicted and actual values. Control bias refers to the difference between the control quantity set in the cooperative control instruction set and the control quantity actually executed by the trackless rubber-wheeled vehicle.

[0126] Evaluation metrics are measures used to quantify and comprehensively assess prediction and control deviations. These metrics can be single numerical values ​​(e.g., the root mean square value of temperature prediction error) or multi-dimensional vectors used to comprehensively measure the performance of a control system.

[0127] Online optimization refers to the process of dynamically adjusting model parameters or decision logic based on real-time collected feedback data and generated evaluation indicators during the operation of a trackless rubber-tired vehicle.

[0128] Thermal state prediction models are used to predict the temperature trend of target components over future periods. Online optimization of their parameters or logic based on evaluation metrics can improve prediction accuracy.

[0129] An environmental thermal risk assessment model is used to evaluate the environmental thermal risk level of the roadway ahead. Online optimization of its parameters or logic based on evaluation indicators can make the assessment results more consistent with reality. For example, the temperature thresholds, feature weights, or risk factor calculation formulas used to classify risk levels in the model can be adjusted according to the actual temperature rise rate and risk tolerance.

[0130] The collaborative decision engine's decision rule base is responsible for generating collaborative control instruction sets based on various inputs. Online optimization of the decision rule base's parameters or logic can make the generated instruction sets more effective and robust. For example, the decision weight allocation between thermal management control instructions and motion control instructions can be adjusted based on prediction deviation, control deviation, and task type, or the decision logic such as decision trees and fuzzy logic rules can be modified.

[0131] As can be seen from the above, this application solves the adaptability problem of the model and strategy in dynamic environments through feedback acquisition, deviation comparison, and online optimization mechanisms, realizing the self-learning and real-time adjustment of the control system. This improves prediction accuracy and control stability, ensuring the safe and efficient operation of the trackless rubber-tired vehicle in complex downhole environments.

[0132] In one embodiment of this application, online parameter optimization of the thermal state prediction model is performed based on evaluation metrics, including: Compare the actual temperature change curve of the target component with the previously predicted temperature evolution curve to calculate the temperature prediction error; If the temperature prediction error exceeds the allowable range for multiple consecutive control cycles, the parameters of the thermal state prediction model will be updated online.

[0133] In this embodiment, the temperature prediction error can be obtained by calculating the absolute difference or relative difference between the actual temperature value and the predicted temperature value. For example, statistical indicators such as mean square error, mean absolute error or root mean square error can be used to quantify the prediction error over a period of time, or the instantaneous error of each sampling point can be simply calculated.

[0134] The allowable range is a preset threshold representing the upper limit of acceptable prediction error, while multiple consecutive control cycles represent the persistence of the error. For example, an error threshold (such as ±5℃ or ±10%) can be set, and a counter can be maintained. When the error exceeds the threshold, the counter increments by one. When the counter reaches a preset number of consecutive cycles (3 or 5 control cycles), the condition is met. When the above condition is met, the parameters of the thermal state prediction model will be updated online.

[0135] As can be seen from the above, by setting trigger conditions where the error exceeds the allowable range within multiple consecutive control cycles, this application effectively filters out instantaneous fluctuations or measurement noise, ensuring that parameter updates are only initiated when the prediction model exhibits persistent and systematic deviations. This improves the stability and efficiency of system updates and avoids unnecessary frequent adjustments. Finally, the mechanism for online updating of thermal state prediction model parameters enables the model to adaptively adjust in real time based on actual operating data, ensuring that the model can always accurately represent the thermal characteristics under the current downhole environment and equipment operating conditions, continuously improving prediction accuracy.

[0136] In one embodiment of this application, if the temperature prediction error exceeds the allowable range within multiple consecutive control cycles, the parameters of the thermal state prediction model are updated online, including: Construct a loss function based on temperature prediction error; The impact of environmental thermal risk level and operating condition data on the prediction confidence of thermal state prediction model was analyzed, and the parameter update intensity was determined. An optimization algorithm is adopted, guided by the parameter update intensity, to iteratively adjust the parameters of the thermal state prediction model to minimize the loss function and complete the thermal state prediction model update.

[0137] In this embodiment, the loss function is constructed to quantify the difference between the predicted values ​​of the thermal state prediction model and the actual temperature values ​​of the target component. This loss function provides a clear optimization objective for the optimization algorithm: minimizing the function value to make the model's predictions closer to reality. For example, mean squared error can be used as the loss function, calculated as the average of the sum of squares of the prediction errors.

[0138] In determining the parameter update intensity, this application analyzes the impact of environmental thermal risk level and operating condition data on the prediction confidence of the thermal state prediction model. Prediction confidence characterizes the model's degree of confidence in the reliability of its predictions. By identifying under what external conditions the model's predictions are less accurate, the parameter update intensity can be adjusted accordingly. For example, historical data analysis can establish a statistical relationship between environmental thermal risk level, specific operating conditions, and model prediction errors. When the environmental thermal risk level is high or the operating condition data indicates that the trackless rubber-tired vehicle is under high load, the model's prediction confidence is low. In this case, a higher parameter update intensity can be set to correct model bias more quickly. Conversely, when environmental conditions are stable and the operating load is low, a lower update intensity can be used to maintain model stability.

[0139] Subsequently, an optimization algorithm is employed, guided by the parameter update intensity, to iteratively adjust the parameters of the thermal state prediction model to minimize the loss function. The optimization algorithm's role here is to adjust the internal parameters of the thermal state prediction model based on the error direction indicated by the loss function and the adjustment magnitude determined by the parameter update intensity. The iterative adjustment algorithm repeatedly executes the parameter update steps, gradually approaching the optimal solution. For example, gradient descent can be used to calculate the gradient of the loss function with respect to the model parameters, and then adjust the parameters in the opposite direction of the gradient with the parameter update intensity (i.e., the learning rate) as the step size to complete the thermal state prediction model update. This means applying the new parameters adjusted by the optimization algorithm to the thermal state prediction model, giving it higher prediction accuracy. This can be achieved by directly replacing the old parameter set in the model, or by using a smooth update strategy, such as weighted averaging of the old and new parameters, to avoid drastic fluctuations in the model during the update process.

[0140] As can be seen from the above, through the aforementioned technical solution, this application constructs a loss function based on temperature prediction error, providing a clear quantitative target for parameter optimization, avoiding blind updates, and making model correction directional. Simultaneously, by analyzing the impact of environmental thermal risk level and operating condition data on prediction confidence, the parameter update intensity is dynamically determined, effectively preventing model instability caused by excessive or insufficient updates, and ensuring the model's adaptability under different operating conditions and environments. The use of optimization algorithms and iterative adjustments guided by parameter update intensity ensures efficient convergence of the parameter optimization process, thereby improving the model's prediction accuracy. Finally, the thermal state prediction model is updated, achieving real-time adaptation of model parameters and greatly enhancing the reliability of predictions.

[0141] In one embodiment of this application, online parameter optimization of the environmental thermal risk assessment model is performed based on evaluation indicators, including: Statistics were compiled on the actual temperature rise rate of the target component after executing the coordinated control instruction set under different environmental thermal risk levels. Based on the risk tolerance of the target task type, the temperature threshold or feature weights used to classify risk levels in the environmental thermal risk assessment model are dynamically adjusted.

[0142] In this embodiment, the actual temperature rise rate of the target component after executing the coordinated control command set is first statistically analyzed under different environmental thermal risk levels. Statistical analysis can be understood as the process of collecting, organizing, and analyzing historical operational data. For example, the system can continuously record the temperature change of the target component over time after executing the coordinated control command set under different environmental thermal risk levels for the trackless rubber-tired vehicle, thereby calculating its actual temperature rise rate within a specific time period. This data can be stored in a historical database and indexed and queried based on dimensions such as time, environmental conditions, and task type.

[0143] Building upon this, this application further dynamically adjusts the temperature thresholds or feature weights used to classify risk levels in the environmental thermal risk assessment model based on the risk tolerance of the target task type. This step aims to enable the environmental thermal risk assessment model to adaptively adjust according to the characteristics of the current task, thereby improving the accuracy and practicality of the assessment. The risk tolerance of the target task type refers to the different levels of tolerance and priority that different task types have for thermal risks.

[0144] Dynamic adjustment can be achieved by using machine learning algorithms to train an adaptive model based on historical task data and corresponding risk tolerance. This model can automatically output adjusted temperature thresholds or feature weights according to the type of input target task. For example, adaptive control methods can be used to continuously optimize these parameters based on task objectives and actual operational feedback.

[0145] As can be seen from the above, the technical solution described herein effectively solves the problem of inaccurate assessment results caused by fixed parameters in existing environmental thermal risk assessment models. This dynamic adjustment mechanism improves the accuracy and task adaptability of environmental thermal risk assessment, avoids the risk of excessive heat dissipation or thermal runaway due to inaccurate assessment, and optimizes the generation of collaborative control instruction sets, ensuring that the trackless rubber-wheeled vehicle can achieve safe, efficient, and energy-saving operation under different task and environmental conditions.

[0146] In one embodiment of this application, the decision weights or rules in the collaborative decision engine are optimized based on evaluation metrics, including: Based on the thermal state prediction bias, environmental risk assessment bias, and target task type, adjust the decision weight of thermal management control commands relative to motion control commands in the collaborative decision engine.

[0147] In this embodiment, the thermal state prediction deviation refers to the difference between the actual temperature change curve of the target component and the previously predicted temperature change curve. This deviation quantifies the accuracy of the thermal state prediction model in predicting future temperature trends. This deviation can be obtained in various ways, such as by calculating the mean square error, mean absolute error, or maximum absolute error between the actual temperature and the predicted temperature at a specific time point or time period.

[0148] Environmental risk assessment bias refers to the difference between the environmental thermal risk level output by the environmental thermal risk assessment model and the actual environmental thermal load experienced by the trackless rubber-tired vehicle during actual operation. This bias characterizes the accuracy of the model's identification and assessment of the thermal environment of the roadway ahead. This bias can be indirectly assessed by comparing the environmental thermal risk level assessed by the model with indicators such as the temperature rise rate of the trackless rubber-tired vehicle components and the load on the heat dissipation system after actual operation.

[0149] The target task type refers to the category of the task that the trackless rubber-tired vehicle is currently performing, such as a transportation task, a transshipment task, or an inspection task. Different task types have different priorities regarding driving efficiency and thermal safety. The target task type can be explicitly included in the task instruction as a task type identifier, which can be directly read and identified.

[0150] Adjusting the decision weight of thermal management control commands relative to motion control commands in the collaborative decision engine refers to dynamically changing the relative importance of thermal management control commands and motion control commands when generating collaborative control command sets, based on the aforementioned deviations and task types. A rule-based expert system can be used to dynamically switch preset weight configuration schemes according to the severity of the deviation and the task type. For example, increasing the weight of thermal management commands when thermal risk is high, and increasing the weight of motion control commands when task efficiency requirements are high and thermal risk is controllable.

[0151] Through the above technical solution, this application can effectively solve the problem of unreasonable allocation of decision weights, ensuring that the priority of thermal management control commands and motion control commands can be optimized according to real-time deviations and task types. By using thermal state prediction deviations and environmental risk assessment deviations, the difference between model predictions and actual conditions can be perceived in real time, thus providing an objective basis for adjusting decision weights. This weight adjustment method enables the trackless rubber-tired vehicle to achieve more intelligent and adaptive control in complex and ever-changing downhole environments, thereby effectively improving the equipment's operational reliability, safety, and energy efficiency.

[0152] In one embodiment of this application, the decision weights of thermal management control commands relative to motion control commands in the collaborative decision engine are adjusted based on thermal state prediction deviations, environmental risk assessment deviations, and target task types, including: Based on the preset basic weights of the target task type, the severity and trend of thermal state prediction deviation and environmental risk assessment deviation, and real-time operating data, the relative importance of thermal safety management and driving efficiency management in the current decision-making cycle is comprehensively evaluated to obtain a comprehensive evaluation result. Based on the comprehensive evaluation results, the decision weight allocation between thermal management control commands and motion control commands in the collaborative decision engine was adjusted.

[0153] In this embodiment, the preset base weight based on the target task type refers to a set of weight parameters pre-set according to the specific task type performed by the trackless rubber-tired vehicle before the task begins. These weight parameters characterize the inherent priority or emphasis of different task types on thermal safety management and driving efficiency management. For example, this can be achieved by consulting a preset weight configuration table or database; a higher driving efficiency weight may be preset for transportation tasks, while a higher thermal safety weight may be preset for inspection tasks.

[0154] The severity and trend of thermal state prediction deviation and environmental risk assessment deviation are considered. Thermal state prediction deviation refers to the difference between the actual and predicted temperatures of the target component, while environmental risk assessment deviation refers to the difference between the actual and assessed environmental thermal risks. Severity refers to the absolute magnitude of these deviations or the degree to which they exceed a preset threshold. Trend refers to the direction and rate of evolution of these deviations over a continuous control period, such as whether the deviations continuously increase, decrease, or fluctuate. Severity can be determined by calculating the absolute or relative value of the deviation and comparing it with a preset severity level threshold. Trend can be obtained through time series analysis of deviation values ​​over multiple consecutive control periods.

[0155] A comprehensive assessment of the relative importance of thermal safety management and driving efficiency management within the current decision-making cycle yields a comprehensive evaluation result. This involves integrating and analyzing the aforementioned inputs (including basic weights, the severity and trend of deviations, and real-time operating data) to quantify which of the two—thermal safety management and driving efficiency management—should be prioritized at the current moment, or how the two should be balanced. The comprehensive evaluation result is the output of this quantitative analysis. This can be achieved by establishing a multi-objective decision-making model, such as the Analytic Hierarchy Process (AHP), by weighted summing of the various input factors and their weights to obtain a comprehensive priority score or relative importance index.

[0156] Adjusting the decision weight allocation of thermal management control commands and motion control commands in the collaborative decision engine based on the comprehensive evaluation results means dynamically adjusting the relative priority of thermal safety and driving efficiency objectives when generating these commands. This can be achieved by adjusting the weight coefficients in the optimization objective function within the collaborative decision engine. For example, if the comprehensive evaluation results indicate that thermal safety is more important, the weight coefficients of thermal management-related objectives are increased, while the weight coefficients of driving efficiency-related objectives are decreased.

[0157] As can be seen from the above, through the above technical solution, this application improves the system's adaptability and overall performance in dynamic environments, enabling the collaborative decision engine to more accurately balance the thermal load and driving performance of the trackless rubber-tired vehicle, avoiding overly conservative or overly aggressive control strategies, thereby maximizing operating efficiency and energy utilization while ensuring the safe operation of the equipment.

[0158] Corresponding to the control method for trackless rubber-tired vehicles in wells based on equipment thermal state sensing in the above embodiment, Figure 2 This is a structural block diagram of a trackless rubber-tired vehicle control system for underground wells based on equipment thermal state sensing, provided as an embodiment of this application. For ease of explanation, only the parts relevant to the embodiment of this application are shown. References Figure 2The underground trackless rubber-tired vehicle control system 20 based on equipment thermal state perception includes: a task processing module 21, a path planning module 22, a data acquisition module 23, a thermal environment assessment module 24, a thermal state assessment module 25, a collaborative decision engine 26, and a control execution module 27.

[0159] Among them, the task processing module 21 is used to determine the target task type and target driving area in response to the received task instruction; The path planning module 22 is used to plan the driving path based on the target task type, the digital map of the alleyway in the target driving area, and the positioning information of the trackless rubber-wheeled vehicle. The data acquisition module 23 is used to control the trackless rubber-tired vehicle to travel along the driving path and to collect environmental image data of the road ahead, temperature data of the target components of the trackless rubber-tired vehicle, and real-time operating data of the trackless rubber-tired vehicle. The thermal environment assessment module 24 is used to assess the thermal environment of the roadway ahead based on environmental image data of the roadway ahead and the expected driving trajectory corresponding to the driving path, and obtain the environmental thermal risk level. The thermal condition assessment module 25 is used to input temperature data, real-time operating condition data and environmental thermal risk level into the preset thermal condition prediction model to obtain the predicted temperature change curve of the target component in the future period. Based on the current temperature value of the target component and the predicted temperature change curve, the thermal condition assessment level of the trackless rubber-tired vehicle is determined. The collaborative decision engine 26 is used to input environmental thermal risk level, thermal state assessment level, target task type, target driving area and real-time operating condition data into the collaborative decision engine to generate a collaborative control instruction set, which includes motion control instructions and thermal management control instructions. The control execution module 27 is used to execute the collaborative control instruction set to adjust the driving status and thermal load status of the trackless rubber-wheeled vehicle.

[0160] See Figure 3 , Figure 3 This is a schematic block diagram of an electronic device provided according to an embodiment of this application. Figure 3 The electronic device 300 in this embodiment may include one or more processors 301, one or more input devices 302, one or more output devices 303, and one or more memories 304. The processors 301, input devices 302, output devices 303, and memories 304 communicate with each other via a communication bus 305. The memories 304 store computer programs, including program instructions. The processors 301 execute the program instructions stored in the memories 304. Specifically, the processors 301 are configured to invoke the program instructions to perform the functions of the modules in the aforementioned device embodiments, for example... Figure 2The functions of the task processing module 21, path planning module 22, data acquisition module 23, thermal environment assessment module 24, thermal state assessment module 25, collaborative decision engine 26, and control execution module 27 are shown.

[0161] It should be understood that, in the embodiments of this application, the processor 301 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0162] Input device 302 may include a touchpad, a fingerprint sensor (for collecting the user's fingerprint information and fingerprint orientation information), a microphone, etc., and output device 303 may include a display (LCD, etc.), a speaker, etc.

[0163] Memory 304 may include read-only memory and random access memory, and provides instructions and data to processor 301. A portion of memory 304 may also include non-volatile random access memory. For example, memory 304 may also store device type information.

[0164] In specific implementations, the processor 301, input device 302, and output device 303 described in the embodiments of this application can execute the implementation methods described in any embodiment of the control method for trackless rubber-tired vehicles in wells based on equipment thermal state perception provided in the embodiments of this application, or they can execute the implementation methods of the electronic devices described in the embodiments of this application, which will not be repeated here.

[0165] This application also provides a computer-readable storage medium storing a computer program. The computer program includes program instructions, which, when executed by a processor, implement all or part of the processes in the methods described above. The computer program can also instruct related hardware to implement these processes. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer-readable storage medium is used to store the computer program and other programs and data required by the electronic device.

[0166] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A control method for a trackless rubber-tired vehicle in wells based on equipment thermal state sensing, characterized in that, include: In response to the received task instructions, determine the target task type and target driving area; Based on the target task type, the digital map of the alleyways in the target driving area, and the positioning information of the trackless rubber-wheeled vehicle, a driving route is planned; Control the trackless rubber-tired vehicle to travel along the driving path, and collect environmental image data of the road ahead, temperature data of the target components of the trackless rubber-tired vehicle, and real-time operating data of the trackless rubber-tired vehicle; Based on the environmental image data of the roadway ahead and the expected driving trajectory corresponding to the driving path, a thermal environment assessment of the roadway ahead is performed to obtain the environmental thermal risk level. The temperature data, the real-time operating condition data, and the environmental thermal risk level are input into a preset thermal state prediction model to obtain the predicted temperature change curve of the target component in the future period. Based on the current temperature value of the target component and the predicted temperature change curve, the thermal state assessment level of the trackless rubber-wheeled vehicle is determined. The environmental thermal risk level, the thermal state assessment level, the target task type, the target driving area, and the real-time operating condition data are input into the collaborative decision engine to generate a collaborative control instruction set, which includes motion control instructions and thermal management control instructions. The coordinated control instruction set is executed to adjust the driving status and thermal load status of the trackless rubber-wheeled vehicle.

2. The method for controlling a trackless rubber-tired vehicle in wells based on equipment thermal state sensing according to claim 1, characterized in that, The step of planning the travel route based on the target task type, the digital map of the alleyway in the target travel area, and the positioning information of the trackless rubber-wheeled vehicle includes: Several candidate routes that can be reached were identified based on the digital map of the alleyways; For each candidate path, based on the historical environmental thermal data associated with the roadway digital map, the environmental thermal load that the trackless rubber-tired vehicle will experience when traveling along the candidate path is estimated. Based on the thermal characteristic model of the trackless rubber-tired vehicle, the environmental heat load, and the current temperature value of the target component, the temperature rise process of the target component when traveling along each candidate path is predicted, and the predicted temperature of the target component at the end of each candidate path is obtained. Based on the weighted configuration of driving efficiency and thermal safety defined by the target task type, the candidate path with a predicted temperature below the first safety threshold throughout the journey and a travel time that meets the task requirements is selected from the candidate paths as the driving path.

3. The method for controlling a trackless rubber-tired vehicle in wells based on equipment thermal state sensing according to claim 2, characterized in that, Before or simultaneously with identifying several accessible candidate paths based on the digital map of the alleyways, the method further includes: Based on the historical task database, retrieve historical task records that are similar to the target task type, the target driving area, and the average ambient temperature within a predetermined time period before the retrieval time. Extract the historical collaborative control instruction set and its corresponding final thermal state assessment results from the historical task records; Based on the final thermal state assessment results, the historical collaborative control instruction set is evaluated for effectiveness. The strategy parameters corresponding to the historical collaborative control instruction set with the highest score are used as the initial strategy parameters of the collaborative decision engine in this task.

4. The method for controlling a trackless rubber-tired vehicle in wells based on equipment thermal state sensing according to claim 2, characterized in that, Also includes: During operation, based on real-time collected data, the system continuously predicts the environmental thermal risk level sequence that the trackless rubber-tired vehicle will face during the future journey, as well as the temperature evolution curve of the target components. If the predicted temperature of the target component is determined to be greater than the second warning threshold before reaching the end of the mission based on the temperature evolution curve, then path replanning is triggered to obtain the updated driving path. Wherein, the second warning threshold is less than the first safety threshold.

5. The method for controlling a trackless rubber-tired vehicle in wells based on equipment thermal state sensing according to claim 4, characterized in that, If, based on the temperature evolution curve, it is determined that the predicted temperature of the target component will exceed the second warning threshold before reaching the mission endpoint, path replanning is triggered to obtain an updated driving path, including: A path optimization model is established with the weighted sum of task completion time and total energy consumption as the optimization objective and the predicted temperature being less than the first safety threshold throughout the process as a hard constraint. When the path replanning condition is triggered, at least one alternative local path is generated in the digital map of the alleyway based on the current position of the trackless rubber-tired vehicle and the destination of the task. Based on the updated environmental thermal risk level, thermal state assessment level, real-time operating data, and the path optimization model, the evaluation value of each alternative local path is calculated. The optimal alternative local path is selected as the updated driving path based on the evaluation value.

6. The control method for a trackless rubber-tired vehicle in wells based on equipment thermal state sensing according to claim 5, characterized in that, Also includes: Several virtual evaluation points are set at equal intervals or according to the characteristics of the roadway environment on the alternative local path; For each alternative local path, the thermal state prediction model, the estimated environmental heat load corresponding to each virtual evaluation point and the thermal characteristic model of the trackless rubber-tired vehicle are used to simulate and deduce the temperature of the target component when the trackless rubber-tired vehicle travels to each virtual evaluation point, and the component temperature sequence is obtained. Based on the component temperature sequence, a thermal risk index is calculated to characterize the overall thermal safety level of the alternative local path. In the path optimization model, the thermal risk index, path length, and estimated travel time are combined to form a multi-objective optimization function for solving, thereby obtaining a comprehensive evaluation value of the candidate local path.

7. The method for controlling a trackless rubber-tired vehicle in wells based on equipment thermal state sensing according to claim 1, characterized in that, The environmental image data includes visible light images and infrared images; The thermal environment assessment of the ahead alleyway is performed based on the environmental image data of the ahead alleyway and the expected driving trajectory corresponding to the driving path, to obtain the environmental thermal risk level, including: The acquired visible light and infrared images are fused together to identify abnormal heat source areas in the roadway ahead. Based on the temperature, geometric dimensions, and relative positional relationship between the abnormal heat source region and the expected driving trajectory, the environmental thermal risk level is calculated using an environmental thermal risk assessment model.

8. The method for controlling a trackless rubber-tired vehicle in wells based on equipment thermal state sensing according to claim 1, characterized in that, The step of inputting the environmental thermal risk level, the thermal state assessment level, the target task type, the target driving area, and the real-time operating condition data into the collaborative decision engine to generate a collaborative control instruction set includes: Based on the target task type and target driving area, a preset baseline thermal management strategy is invoked as the starting point for decision-making; Based on the environmental thermal risk level and the thermal state assessment level, the control parameters in the baseline thermal management strategy are optimized in the first level through the collaborative decision engine to obtain the first-level optimization parameter set. Based on the load and roadway slope data in the real-time operating data, the first-level optimization parameter set is adjusted in the second level to generate the final parameters of the motion control command and the thermal management control command.

9. The control method for a trackless rubber-tired vehicle in wells based on equipment thermal state sensing according to claim 1, characterized in that, Also includes: The actual operating data of the trackless rubber-wheeled vehicle, the actual temperature data of the target components, and the environmental data after the execution of the collaborative control instruction set are collected and used as strategy execution feedback data. The strategy execution feedback data is compared with the expected control target to generate evaluation indicators including prediction deviation and control deviation. Based on the evaluation indicators, the parameters or logic of the thermal state prediction model, the environmental thermal risk assessment model, and / or the decision rule base in the collaborative decision engine are optimized online.

10. A control system for a trackless rubber-tired vehicle in wells based on equipment thermal state sensing, characterized in that, include: The task processing module is used to respond to received task instructions and determine the target task type and target driving area; The path planning module is used to plan the travel path based on the target task type, the digital map of the alleyways in the target travel area, and the positioning information of the trackless rubber-wheeled vehicle. The data acquisition module is used to control the trackless rubber-tired vehicle to travel along the driving path and to collect environmental image data of the road ahead, temperature data of the target components of the trackless rubber-tired vehicle, and real-time operating data of the trackless rubber-tired vehicle. The thermal environment assessment module is used to assess the thermal environment of the roadway ahead based on the environmental image data of the roadway ahead and the expected driving trajectory corresponding to the driving path, and to obtain the environmental thermal risk level. The thermal state assessment module is used to input the temperature data, the real-time operating condition data and the environmental thermal risk level into a preset thermal state prediction model to obtain the predicted temperature change curve of the target component in the future period. Based on the current temperature value of the target component and the predicted temperature change curve, the thermal state assessment level of the trackless rubber-tired vehicle is determined. The collaborative decision engine is used to input the environmental thermal risk level, the thermal state assessment level, the target task type, the target driving area and the real-time operating condition data into the collaborative decision engine to generate a collaborative control instruction set, which includes motion control instructions and thermal management control instructions. The control execution module is used to execute the collaborative control instruction set and adjust the driving state and thermal load state of the trackless rubber-wheeled vehicle.

Images