Movement control method and system for local heat accumulation of air cooling radiator

Abstract

The invention discloses a mobile control method and system for local heat accumulation of an air cooling radiator, and belongs to the technical field of automatic control, and the method comprises the steps: obtaining thermal imaging data of the surface of the air cooling radiator; dynamic modeling is carried out on the thermal imaging data, and a thermal aggregation state spatio-temporal evolution model including target area dynamic contour prediction, diffusion trend prediction and phase change type prediction is generated; generating a self-adaptive moving path matched with the prediction state of the target area and a preset operation mode corresponding to the phase change type according to the model, and integrating the self-adaptive moving path and the preset operation mode into a space-time cooperative control instruction; driving an anti-freezing execution unit to move along the path and preparing to execute an operation mode; collecting real-time temperature data to dynamically correct the model; and terminating the task when the temperature of the target area recovers to the safety threshold. Dynamic modeling and self-adaptive path planning are adopted, real-time feedback correction is combined, mobile operation cooperation of the anti-freezing execution unit can be achieved, local heat accumulation is restrained, and the operation safety of the air cooling system is improved.

Description

Technical Field

[0001] This invention relates to the field of automatic control technology, and in particular to a movement control method and system for localized heat accumulation in an air-cooled radiator. Background Technology

[0002] Air-cooled radiators, as key heat exchange devices in industries such as thermal power generation and chemical processing, primarily function to discharge waste heat generated during industrial processes into the atmosphere through air convection. In cold climates, especially in winter or high-altitude areas, the surface of air-cooled radiators is prone to uneven localized temperatures due to low ambient temperatures. This can lead to tube freezing, equipment damage, or a sharp decline in heat exchange efficiency. The formation and development of these low-temperature areas is known as localized heat accumulation. Therefore, effective monitoring and control of such phenomena are crucial for ensuring the safety and stability of industrial production.

[0003] In related technologies, Chinese invention patent CN120428835B discloses a server heat dissipation device and heat dissipation control method, including: synchronously collecting real-time temperature data of each partition through a temperature detection module to construct a three-dimensional temperature field distribution map containing temperature gradient and heat flow direction; then performing thermodynamic parameter calculation, extracting the temperature time-series change characteristics of each partition, and predicting the heat diffusion path within a future set time period by fusing the temperature time-series change characteristics and thermodynamic parameters; generating a control instruction set based on the predicted heat diffusion path, monitoring the actual heat dissipation response through an acoustic sensor during execution, dynamically comparing the monitoring data with the predicted heat diffusion path and updating the control parameters of the control instruction set until each partition reaches the target temperature threshold; predicting temperature abrupt changes and shortening the hotspot elimination response time.

[0004] However, the aforementioned existing technical solutions have the following technical shortcomings: Lack of proactive interception capability for dynamically moving targets: This solution primarily targets static or quasi-static hotspots such as server racks for heat dissipation. Its control logic adjusts the parameters of fixed heat dissipation actuators (such as fans) based on the prediction of heat diffusion paths. However, for "localized heat accumulation" on the surface of large equipment such as air-cooled radiators, where the location and shape may evolve rapidly, this solution lacks the ability to drive mobile actuators to perform proactive path planning and dynamic interception, making it difficult to achieve precise intervention "at the right location at the right time." Limited operating mode and lack of coordination with specific phase change types: The control command set of this solution mainly adjusts based on heat dissipation intensity, without considering the different handling strategies required for different icing types. Its relatively limited operating mode makes it unable to dynamically match and configure the collaborative operating parameters of the composite end effector based on the predicted phase change type, resulting in low energy utilization efficiency or poor de-icing effect. Sequential movement and operation preparation processes with large response delays: When executing control commands, this solution does not address the coordination between the moving platform and operation preparation. If applied to scenarios such as mobile robots, the traditional serial process of "moving to the position first and then starting work preparation" will introduce significant time delays, which may miss the golden window for dealing with local heat accumulation and fail to meet the requirements of air-cooled systems for rapid response. Summary of the Invention

[0005] To address the aforementioned issues, this invention provides a mobile control method and system for localized heat accumulation in air-cooled radiators. By employing dynamic modeling and adaptive path planning, combined with real-time feedback correction, it enables coordinated mobile operations of antifreeze execution units, suppresses localized heat accumulation, and enhances the operational safety of the air-cooling system.

[0006] The above objectives can be achieved through the following approach: A method and system for controlling the movement of localized heat accumulation in an air-cooled radiator includes: acquiring thermal imaging data of the radiator surface; dynamically modeling the thermal imaging data to generate a spatiotemporal evolution model of the heat accumulation state, including predictions of the dynamic contour of the target area, diffusion trends, and phase change types; generating an adaptive movement path matching the predicted state of the target area and a preset operating mode corresponding to the phase change type based on the model, and integrating them into a spatiotemporal coordinated control command; driving an antifreeze execution unit to move along the path and prepare to execute the operating mode; collecting real-time temperature data to dynamically correct the model; and terminating the task when the temperature of the target area returns to a safe threshold.

[0007] Optionally, the step of dynamically modeling the thermal imaging data to generate a spatiotemporal evolution model of thermal accumulation state that includes dynamic contour prediction, diffusion trend prediction, and phase transition type prediction of the target area includes: performing grid segmentation processing on the thermal imaging data to obtain temperature distribution information of multiple grid cells of interest; identifying abnormal grid cells with temperatures below a preset threshold based on the temperature distribution information; aggregating and tracking multiple abnormal grid cells captured in a continuous time series to generate an initial abnormal region contour and its historical change trajectory that records the state evolution of the region; calculating the future diffusion vector and rate of the initial abnormal region contour based on the historical change trajectory, and predicting the phase transition type of the initial abnormal region in combination with the acquired environmental state information; and integrating the initial abnormal region contour, the future diffusion vector and rate, and the phase transition type to construct and output the spatiotemporal evolution model of thermal accumulation state.

[0008] Optionally, the step of performing grid segmentation processing on the thermal imaging data to obtain temperature distribution information of multiple grid cells of interest includes: loading a virtual grid region that can digitize the surface of the air-cooled radiator; obtaining a region focus strategy that defines the scanning priority and scanning frequency of different grid regions; and dynamically adjusting the scanning parameters of different virtual grid regions according to the region focus strategy to obtain temperature distribution information of multiple grid cells of interest.

[0009] Optionally, the integrated spatiotemporal collaborative control command generation includes: acquiring the real-time position of the anti-freezing execution unit; loading and storing factory environment map information containing information on fixed obstacles and walkable paths; using the real-time position as the starting point and the future position corresponding to the dynamic contour prediction in the spatiotemporal evolution model of the thermal accumulation state as the dynamic endpoint, and combining the factory environment map information to perform spatiotemporal path search to generate an adaptive movement path; determining a preset operation mode based on the phase change type prediction; and binding the adaptive movement path with the preset operation mode to generate spatiotemporal collaborative control commands.

[0010] Optionally, the preset operation mode further includes: matching the phase change type prediction with the operation strategies in the preset multi-mode operation library to determine the dominant functional unit and the auxiliary functional unit; and generating a set of collaborative operation control parameters to define the start-up order, effect intensity and effect duration of the dominant functional unit and the auxiliary functional unit based on the matching result, wherein the collaborative operation control parameters are part of the preset operation mode.

[0011] Optionally, sending the spatiotemporal coordinated control command to the antifreeze execution unit to drive the antifreeze execution unit to move along the adaptive movement path and simultaneously prepare to execute the preset operation mode includes: parsing the spatiotemporal coordinated control command into motion control sub-commands to drive the antifreeze execution unit to move and operation preparation sub-commands for configuring its composite end effector; simultaneously executing the operation preparation sub-commands while the antifreeze execution unit moves along the adaptive movement path according to the motion control sub-commands; through the synchronous execution, the composite end effector completes the functional configuration corresponding to the preset operation mode before reaching the target area.

[0012] Optionally, the step of dynamically correcting the spatiotemporal evolution model of the thermal accumulation state based on the real-time temperature data includes: comparing the collected real-time temperature data with the temperature change curve predicted by the spatiotemporal evolution model of the thermal accumulation state before operation to generate a deviation sequence of quantitative prediction and actual deviation; adjusting the internal model parameters in the spatiotemporal evolution model of the thermal accumulation state used to predict diffusion trends and phase transition types in reverse according to the deviation sequence; and updating the spatiotemporal evolution model of the thermal accumulation state using the adjusted internal model parameters to generate a corrected spatiotemporal evolution model of the thermal accumulation state.

[0013] Optionally, when it is determined that the temperature of the target area has recovered to a safe threshold defined by the system function based on the modified spatiotemporal evolution model of thermal accumulation state, generating and executing the task termination instruction includes: acquiring the predicted temperature value of the target area in real time based on the modified spatiotemporal evolution model of thermal accumulation state; comparing the predicted temperature value with the safe threshold; and generating and sending the task termination instruction to the antifreeze execution unit when the predicted temperature value continuously reaches or exceeds the safe threshold and remains at a preset duration.

[0014] Optionally, generating an adaptive movement path includes: calculating a series of dynamic candidate endpoints, ordered by time and representing the future evolution trajectory of the target area, based on the diffusion trend prediction; employing a search algorithm that incorporates a time dimension between the real-time location and the dynamic candidate endpoints; and using the search algorithm to find a path that comprehensively evaluates movement time, target coverage effectiveness, and energy consumption, while satisfying the physical constraints of the factory environment map information, as the adaptive movement path.

[0015] Based on the same inventive concept, this invention also provides a mobile control system for localized heat accumulation in an air-cooled radiator. The system includes: a data acquisition module for acquiring thermal imaging data of the air-cooled radiator surface; a state modeling and prediction module for dynamically modeling the thermal imaging data, generating a spatiotemporal evolution model of the heat accumulation state that includes dynamic contour prediction, diffusion trend prediction, and phase change type prediction of the target area; and a path planning and collaborative control module for generating an adaptive movement path that matches the arrival time of the antifreeze execution unit with the predicted state of the target area based on the spatiotemporal evolution model of the heat accumulation state, determining a preset operation mode corresponding to the phase change type prediction, and integrating and generating a spatiotemporal collaborative control system. The system includes: a control command module; an anti-freezing execution module, used to send the spatiotemporal coordinated control command to the anti-freezing execution unit to drive the anti-freezing execution unit to move along the adaptive movement path and simultaneously prepare to execute the preset operation mode; a feedback and adaptive correction module, used to collect real-time temperature data of the target area during and after the operation of the anti-freezing execution unit, and dynamically correct the spatiotemporal evolution model of the heat accumulation state based on the real-time temperature data to generate a corrected spatiotemporal evolution model of the heat accumulation state; and a task execution module, used to generate and execute a task termination command when it is determined, based on the corrected spatiotemporal evolution model of the heat accumulation state, that the temperature of the target area has recovered to the safety threshold defined by the system function.

[0016] Compared with the prior art, the present invention has the following advantages: This invention achieves proactive intervention and precise management of localized low-temperature regions in air-cooled radiators by constructing a predictive closed-loop control system. By establishing a spatiotemporal evolution model of heat accumulation, the system can predict the future location and diffusion trend of low-temperature regions and plan adaptive movement paths for early interception, transforming passive response into proactive prevention. This shortens the reaction time from hazard discovery to effective handling, preventing the occurrence or worsening of freezing problems.

[0017] This invention achieves a high degree of coordination and parallelism between the movement process and the work preparation process, thereby improving work efficiency and equipment utilization. While driving the anti-freezing execution unit to move along the path, the system simultaneously parses and executes the work preparation instructions, enabling it to enter the work state the instant it arrives at the target area. This eliminates the time delay caused by the traditional method of arriving first and then preparing, and increases the effective work time within the golden processing window.

[0018] This invention enhances the robustness and environmental adaptability of the control system by introducing dynamic correction and adaptive learning mechanisms. During and after operation, the system can reverse-correct and optimize its internal predictive model based on actual temperature change data, enabling the control strategy to continuously improve itself and more accurately adapt to current environmental conditions and equipment status, thus ensuring the continuous stability and high efficiency of the control effect.

[0019] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures pointed out in the description, claims and drawings. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a flowchart illustrating a method for controlling the movement of localized heat accumulation in an air-cooled radiator according to an embodiment of the present invention.

[0022] Figure 2 This is a temperature distribution diagram of an air-cooled radiator according to an embodiment of the present invention.

[0023] Figure 3 This is a diagram of collaborative operation control parameters according to an embodiment of the present invention.

[0024] Figure 4 This is a schematic diagram of the structure of a mobile control system for local heat accumulation in an air-cooled radiator according to an embodiment of the present invention. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0026] Reference Figure 1One embodiment of the present invention proposes a moving control method for local heat accumulation in air-cooled radiators. By employing dynamic modeling and adaptive path planning, combined with real-time feedback correction, it can achieve coordinated moving operations of antifreeze execution units, suppress local heat accumulation, and improve the safety of air-cooling system operation.

[0027] The method described in this embodiment specifically includes: S1. Acquire thermal imaging data of the surface of the air-cooled radiator; Specifically, acquiring thermal imaging data of the air-cooled radiator surface first requires establishing a digital workspace. This process is accomplished by loading a virtual grid region that perfectly corresponds to the physical air-cooled radiator in terms of three-dimensional geometry and topology. The virtual grid region is essentially a digital twin model constructed within the computer control system. It divides the physical surface into several logical grid units with unique spatial coordinates, thus providing a precise geospatial mapping reference for the sensor data. Next, a region-focused strategy is acquired and applied. This strategy is a set of preset logical rules used to define the scanning priority and frequency of different virtual grid regions. This strategy optimizes the allocation of sensor resources by identifying areas of the radiator prone to icing. For example, areas with large fluctuations in heat exchange efficiency, such as the edge tube bundles of the heat exchanger, are set as high priority. The scanning parameters of the thermal imaging sensor are dynamically adjusted based on the region-focused strategy. In practice, the final scanning command is determined by multiplying the preset base scanning frequency by the weighting coefficient corresponding to each region. The base scanning frequency is the minimum sampling rate required for the system to maintain routine monitoring; its value is usually preset based on the sensor's hardware characteristics and the minimum timeliness requirements of environmental monitoring. The weighting coefficient is a dimensionless value reflecting the importance of a specific region in the current environment, dynamically allocated by the region focus strategy. This weighting method allows high-priority regions to receive higher sampling densities. Furthermore, since the weighting coefficient is dimensionless, the calculation results remain in frequency units in physical logic, ensuring the calculation process conforms to objective laws and avoiding problems such as the inability to add or subtract different units or logical inconsistencies. Based on the calculated instructions, the thermal imaging sensor, driven by the actuator, captures infrared thermal images of the designated area. The acquired raw pixel-level temperature data is mapped in real-time to the corresponding virtual grid region, and after data aggregation processing, temperature distribution information for the focus grid cells is generated. This structured data output not only includes real-time temperature but also has a clear spatial index, providing a high-precision input source for subsequent thermal aggregation state prediction.

[0028] S2. Dynamically model the thermal imaging data to generate a spatiotemporal evolution model of thermal aggregation state that includes dynamic contour prediction of the target area, diffusion trend prediction, and phase change type prediction. Optionally, the step of dynamically modeling the thermal imaging data to generate a spatiotemporal evolution model of the thermal accumulation state, including dynamic contour prediction, diffusion trend prediction, and phase transition type prediction of the target region, includes: The thermal imaging data is divided into grids to obtain temperature distribution information for multiple grid cells of interest. Based on the temperature distribution information, abnormal grid cells with temperatures below a preset threshold are identified; Multiple anomalous grid cells captured within a continuous time series are aggregated and tracked to generate an initial anomalous region profile and its historical change trajectory that record the state evolution of the region. Based on the historical change trajectory, the future diffusion vector and rate of the initial anomaly region contour are calculated, and the phase transition type of the initial anomaly region is predicted by combining the acquired environmental state information. By integrating the initial anomaly region contour, the future diffusion vector and rate, and the phase transition type, a spatiotemporal evolution model of the thermal accumulation state is constructed and output.

[0029] Specifically, the system achieves efficient processing and analysis of thermal imaging data by discretizing the continuous temperature field into computable units through gridding. The system loads a virtual grid that precisely corresponds to the physical structure of the air-cooled radiator surface, with the size of each grid cell set according to the required resolution. By registering and segmenting the thermal imaging data stream covering the entire radiator surface (e.g., acquiring 5 frames of infrared thermal images per second), the temperature value of each pixel is assigned to its corresponding virtual grid cell, typically using regional temperature averaging or weighted averaging algorithms. This ultimately generates a matrix of temperature distribution information for the grid cells of interest, indexed by the grid cells and containing the temperature values.

[0030] The system quickly identifies localized areas at risk of icing from global temperature distribution information. It compares the real-time temperature value of each grid cell of interest with a preset anti-freezing safety threshold. This threshold is not fixed but dynamically adjusted based on factors such as ambient humidity and air pressure, typically set between 2 and 5 degrees Celsius. When the real-time temperature of a grid cell falls below the anti-freezing safety threshold, that cell is marked as an abnormal grid cell. The input to this step is a temperature distribution information matrix, and the output is a list recording the coordinates of all abnormal grid cells.

[0031] Discrete anomalous mesh cells are aggregated into physically meaningful low-temperature regions, and their changes over time are tracked to form a historical understanding of the development of anomalous states. The system processes a list of anomalous mesh cells captured within a continuous time series, such as three consecutive acquisition cycles with 10-second intervals. A density-based spatial clustering algorithm, such as DBSCAN, is used to aggregate spatially adjacent anomalous mesh cells to form one or more initial anomalous region contours. This contour can be represented by a closed polygon. Simultaneously, by matching the centroid position, area, and shape features of these contours over time, the system constructs a historical trajectory recording the evolution of the region's state from its generation to the current moment.

[0032] Based on historical data, the system predicts the future spatiotemporal evolution of anomaly regions, including their expansion trends and potential physical phase transitions. The system analyzes the displacement changes of contour boundary points recorded in historical change trajectories. By establishing kinematic models, such as using Kalman filters or optical flow methods, it calculates the future diffusion vector and rate at key points on the contour. The future diffusion vector is a two-dimensional vector indicating the main expansion direction of the contour on the radiator surface, while the rate quantifies the speed of expansion. Simultaneously, the system combines environmental state information obtained from external sensors, such as ambient humidity and wind speed, with the temperature drop rate of the anomaly region itself. Through a pre-defined phase transition type decision model, it predicts the phase transition type. For example, a rapid temperature drop below freezing in a low-humidity environment might predict frost, while a slow temperature decrease in a high-humidity environment might predict clear ice. This phase transition type prediction provides a basis for subsequent selection of operating modes.

[0033] The above analysis and prediction results are structured and encapsulated to form a unified data model that can be directly called by the control system. The system integrates the initial anomaly region outline at the current moment, the predicted future diffusion vector and rate, and the predicted phase transition type to construct and output a spatiotemporal evolution model of the thermal accumulation state. This model is dynamically updated, and its output directly serves as the core input for generating adaptive movement paths and determining preset operation modes, ensuring that the actions of the anti-freezing execution unit are predictable and targeted.

[0034] For example, in the process of moving and controlling local heat accumulation in an air-cooled radiator, the system first loads a virtual grid region that precisely corresponds to the physical structure of the radiator and performs grid segmentation to obtain temperature distribution information. Based on the region focus strategy, the system sets the base scanning frequency to 0.5 Hz and assigns a weight coefficient of 4 to the high-priority region of the windward edge. By multiplying the base scanning frequency by the region weight, the final scanning frequency for this region is calculated to be 2.0 Hz, thus achieving high-frequency thermal imaging data acquisition twice per second in this risk area. When the temperature distribution matrix of the acquired focus grid unit shows a real-time temperature of 2.5 degrees Celsius, and this value is lower than the dynamically adjusted antifreeze safety threshold of 3.0 degrees Celsius, the system identifies it as an abnormal grid unit. Subsequently, the system uses a density-based spatial clustering algorithm to aggregate and track multiple abnormal grid units captured in three consecutive acquisition cycles, generating an initial abnormal region contour and its historical change trajectory that records the evolution of the region's state. The system analyzes the contour displacement recorded in the historical change trajectory and calculates the future diffusion vector and rate of the key points of the contour by establishing a kinematic model. For example, if the displacement vectors in both the horizontal and vertical directions of the region show an increasing trend, the system will calculate the specific rate value representing the expansion speed. Simultaneously, combining information from external sensors indicating an ambient humidity of 85% and a slowly decreasing regional temperature, the system predicts the phase transition type of the region to be open ice using a pre-set phase transition type decision model. Finally, the system integrates the initial anomaly region outline, future diffusion vectors and rates, and phase transition type prediction results to construct and output a spatiotemporal evolution model of the thermal accumulation state. This model will serve as the core basis for subsequently generating adaptive movement paths and determining operational modes.

[0035] Optionally, the step of performing grid-based segmentation on the thermal imaging data to obtain temperature distribution information for multiple grid cells of interest includes: The loading process enables the digitalization of the virtual mesh region corresponding to the surface of the air-cooled radiator. Obtain the region focus strategy that defines the scanning priority and scanning frequency for different grid regions; Based on the region focus strategy, the scanning parameters of different virtual grid regions are dynamically adjusted to obtain temperature distribution information of multiple focus grid cells.

[0036] Specifically, the first step is to establish a digital workspace. This is accomplished by loading a virtual mesh region that perfectly corresponds to the actual air-cooled radiator's geometric dimensions and topology in three-dimensional space. This virtual mesh is created based on the device's computer-aided design model or through reverse engineering using 3D laser scanning, ensuring a high-precision mapping between the physical and digital worlds. The size of the mesh cells, for example, set to 0.25 × 0.25 meters, is determined by a trade-off between the required monitoring accuracy and the spatial resolution of the thermal imaging sensor. After loading, each individual mesh cell is assigned a unique spatial coordinate identifier, providing a foundation for subsequent temperature data loading and regional management. The temperature distribution of the air-cooled radiator is shown as follows: Figure 2 As shown.

[0037] A core scheduling logic, the region focus strategy, is adopted. This strategy is not a simple configuration file, but a dynamic rule base that guides the perception system on how to treat different parts of the radiator differently. Based on historical icing data, fluid dynamics simulation results, and real-time environmental parameters, the strategy divides virtual mesh regions into different priorities. For example, edge pipes on the windward side and structural corners prone to vortices are defined as high-priority regions, while the central stable flow field region may be defined as a low-priority region. The strategy explicitly defines the scanning frequency and scanning priority corresponding to each priority level.

[0038] Based on the acquired area focus strategy, the thermal imaging sensing system is dynamically scheduled to perform scanning tasks and generate temperature distribution information for the focus grid cells. If the system is equipped with a programmable gimbal, the controller adjusts the gimbal's rotation sequence and dwell time according to the scanning priority instructions, so that high-priority areas are placed more frequently at the center of the field of view. Simultaneously, the scanning frequency parameter directly determines the acquisition and processing rate of image data for a specific area. For example, the data refresh rate for high-priority areas can reach 2 Hz, while it drops to 0.5 Hz for low-priority areas. The scanning frequency can be determined by a function: , in, Representing the The final scan frequency of each virtual grid region It is the system's base scanning frequency. For example, setting it to 0.5 Hz represents the minimum guaranteed scanning rate for all areas. It is determined based on the regional focus strategy and allocated to the first The dimensionless weight coefficients for each region are determined by the scanning priority of that region; for example, high, medium, and low priorities correspond to weights of 4, 2, and 1, respectively. This method outputs a non-uniformly updated temperature distribution information matrix for the grid cells of interest, where the temperature data in the risk region has higher temporal resolution, thus providing high-quality data input for the rapid and accurate construction of a spatiotemporal evolution model of thermal aggregation states.

[0039] For example, the system replicates the digital mapping of the air-cooled radiator surface by establishing a digital workspace. First, a virtual mesh region that perfectly corresponds to the actual radiator's geometric dimensions and topology in three-dimensional space is loaded, and the mesh cell size is set to... Each grid cell is assigned a unique spatial coordinate identifier. The system then acquires a region focus strategy defining different grid area scanning priorities and frequencies. This strategy, based on historical icing data and real-time environmental parameters, classifies areas prone to vortex generation, such as the windward edge of pipes, as high-priority areas. At this point, the system's base scanning frequency is set. equal Hertz represents the minimum guaranteed scan rate for all regions. For a virtual grid region marked as high priority... In other words, its corresponding dimensionless weighting coefficient Assigned to The system dynamically adjusts the scanning parameters of the region based on the acquired region focus strategy, and performs a calculation process according to the formula to obtain the final scanning frequency of the region. Hertz. Using this scanning parameter, the system drives the programmable gimbal to adjust the rotation sequence, increasing the data acquisition rate of the risk area to twice per second. The final output is a non-uniformly updated temperature distribution information matrix for the grid cells of interest, providing a high-temporal-resolution data foundation for subsequently constructing a spatiotemporal evolution model of thermal aggregation states.

[0040] S3. Based on the spatiotemporal evolution model of the heat accumulation state, generate an adaptive movement path that can match the arrival time of the antifreeze execution unit with the predicted state of the target area, determine the preset operation mode corresponding to the phase change type prediction, and integrate and generate spatiotemporal collaborative control instructions. Optionally, the integrated spatiotemporal collaborative control commands include: Obtain the real-time position of the antifreeze execution unit; Load the factory environment map information that stores information on fixed obstacles and walkable paths; Starting from the real-time location and taking the future location corresponding to the dynamic contour prediction in the spatiotemporal evolution model of the thermal aggregation state as the dynamic endpoint, a spatiotemporal path search is performed in conjunction with the factory environment map information to generate an adaptive movement path. Based on the phase transition type prediction, a preset operation mode is determined; The adaptive movement path is data-bound with the preset operation mode to generate spatiotemporal collaborative control commands.

[0041] Specifically, the starting point for the planning is established. The system uses an onboard real-time positioning system, such as a LiDAR-based instant localization and mapping (SLAM) algorithm or ultra-wideband (UWB) positioning technology, to obtain the real-time location of the anti-freezing execution unit. This location information includes its two-dimensional coordinates and heading angle in the global coordinate system, and the positioning accuracy is typically required to be within 5 centimeters.

[0042] Construct the constraint environment for motion planning. Load pre-built factory environment map information from memory or a database. This map is typically a high-precision 2D raster map with a grid resolution of around 0.1 meters, clearly marking static environmental elements such as fixed obstacles, equipment bases, pipe supports, and walkable paths. This map provides the physical boundaries and constraints of the world for subsequent path search.

[0043] An adaptive movement path is generated to accurately intercept dynamic targets. This step addresses the problem of "chasing" moving targets, rather than simply moving to a static point. The path planner starts from the real-time position of the antifreeze execution unit. Its endpoint is not fixed but dynamically determined based on the future position predicted by the dynamic contour in the spatiotemporal evolution model of the thermal aggregation state. Specifically, the planner employs a spatiotemporal path search algorithm, such as a time-enhanced A* algorithm. This algorithm operates in two-dimensional space... The time dimension was introduced on the basis of this. The search space becomes three-dimensional. The algorithm aims to find a path such that when the antifreeze execution unit reaches the predicted position of the target region's contour boundary, its movement time is exactly equal to the time required for the model to predict the contour's evolution to that position. Its path cost function... Defined as: , in, From the starting point to the current node The actual travel time cost is calculated by dividing the path length by the average speed of the unit. It is a heuristic function used to predict the behavior of nodes. The minimum time to reach the point where the future trajectory intersects with the target area is estimated by combining node-specific parameters. The Euclidean distance to the current centroid of the target area, the diffusion vector of the target area, and its velocity are considered. Under the physical constraints of the factory environment map information, the planner iteratively searches and ultimately outputs an adaptive movement path that combines optimal time and the highest interception success rate.

[0044] Based on the phase transition type prediction provided by the spatiotemporal evolution model of thermal accumulation state, specific response strategies are determined to ensure a precise match between operational methods and the nature of the problem. The system maintains a pre-set operational mode library, which stores operational strategies for different ice types, such as frost, clear ice, and mixed ice. For example, if the phase transition type is predicted to be "frost," the system will query the library and determine the pre-set operational mode as "Mode." "High-flow-rate, gentle hot air sweeping"; if the prediction is "Mingbing", then it is determined to be "Model". "High-pressure de-icing fluid injection and local infrared heating coordinated operation", the control parameters for the coordinated operation are as follows: Figure 3 As shown.

[0045] The independent path planning results and operation mode selection results are encapsulated into a unified command. The system binds the generated adaptive movement path, i.e., a series of spatiotemporal waypoint sequences, with a predetermined operation mode identifier. This binding is represented in the data structure as a composite object containing path data structure and operation mode parameter set, ultimately generating a complete spatiotemporal cooperative control command, ready to be issued to the anti-freeze execution unit.

[0046] For example, the system reproduces the generation process of spatiotemporal coordinated control commands by simulating the motion trajectory of the antifreeze actuator in a digital simulation environment. First, the system uses LiDAR real-time positioning and mapping technology to obtain the real-time position of the antifreeze actuator in the global coordinate system, setting its coordinates to 0.0 on the horizontal axis and 0.0 on the vertical axis. Then, the system loads high-precision grid-based factory environment map information storing information on fixed obstacles and walkable paths, providing physical boundary constraints for path planning. Starting from the real-time position and using the future position corresponding to the dynamic contour prediction in the spatiotemporal evolution model of the thermal aggregation state as the dynamic endpoint, the system employs a search algorithm that incorporates a time dimension to perform spatiotemporal path search. When the algorithm evaluates a path node with a horizontal axis coordinate of 2.0 and a vertical axis coordinate of 1.5... At that time, the system calculates the node's position based on the 2.5-meter path length between the node and the starting point and the average speed of the anti-freeze execution unit at 0.5 meters per second. Simultaneously, a heuristic function is set based on the predicted intersection point between this node and the evolution trajectory of the target region. The estimated time is 5.0. Calculate the total cost of this node according to the path cost function formula defined in the instruction manual. Through this search process, the system iteratively finds an adaptive movement path that comprehensively evaluates both movement time and coverage effectiveness. Simultaneously, based on the phase transition type prediction results provided by the thermal aggregation state spatiotemporal evolution model, the system matches and determines the corresponding preset operation mode from the pre-defined multi-mode operation library. Finally, the system will adapt the waypoint sequence and pattern in the mobile path. Data binding is performed to integrate and generate spatiotemporal collaborative control commands containing motion and operation coordination information, which are then sent to the airborne main controller of the execution unit.

[0047] Optionally, the preset operation mode further includes: The phase transition type prediction is matched with the operation strategies in the preset multi-mode operation library to determine the dominant functional unit and the auxiliary functional unit. Based on the matching results, a set of collaborative operation control parameters is generated to define the start-up order, intensity, and duration of the dominant functional unit and the auxiliary functional unit. These collaborative operation control parameters are part of the preset operation mode.

[0048] Specifically, abstract icing types are bound to specific execution units. The phase transition type prediction output by the spatiotemporal evolution model of thermal accumulation states is used as a query index for matching within a pre-defined multi-mode operation library. This library stores various operation strategies, each corresponding to one or a class of phase transition types. The matching process is essentially a table lookup or rule-based reasoning process. Once a match is successful, the system extracts the dominant and auxiliary functional units for this operation from the strategy. The dominant functional unit is the equipment responsible for the main de-icing task, such as a high-pressure de-icing fluid injector; the auxiliary functional units play a supporting, pre-processing, or post-processing role, such as an infrared heating module or a hot air purging device. For example, when the phase transition type is predicted as "open ice," the system matches the strategy... This strategy defines the high-pressure de-icing fluid injector as the primary functional unit and the infrared heating module as the auxiliary functional unit.

[0049] The identified functional unit combinations are transformed into a precise set of timing and intensity control parameters. Based on the matching results from the previous step, the system further generates a set of collaborative operation control parameters. These parameters are not isolated numerical values, but a structured set used to precisely define how two or more functional units work collaboratively. This set of collaborative operation control parameters can be represented as a vector. As a core component of the pre-set operation mode: , in, This is the job strategy matched in the previous step. This function... This represents the process of extracting and organizing parameters from the strategy library. Specifically, this parameter set includes: the startup delay times of each unit representing the startup order, such as the startup time of the dominant functional unit. Startup time of auxiliary function units Their time difference determines the order and interval of activation. For example, the strategy for "Mingbing". Possible settings =0, A value of 3 means the auxiliary unit will start for 3 seconds to warm up. The power or flow rate settings for each unit represent the intensity of its function, such as the intensity of the dominant functional unit. Intensity of interaction with auxiliary functional units This intensity value is typically a normalized value between 0 and 1, and is converted into a specific physical quantity, such as injection pressure, heating power, or fan speed, when sent to the actuator. The duration of operation represents the continuous working time of each unit, such as the duration of operation of the dominant functional unit. Duration of operation with auxiliary functional units These durations can be unequal to allow for flexible task combinations. Ultimately, this set of generated collaborative task control parameters is integrated into the data structure of the preset task mode, providing precise and quantifiable instructions for subsequent anti-freezing execution units to prepare for tasks before arriving at the target area and to immediately execute collaborative tasks upon arrival.

[0050] For example, the system reproduces the precise operation configuration process for specific ice conditions by simulating the retrieval and parameter generation logic of a multi-mode operation library in a digital management platform. First, the system retrieves the prediction results for phase transition types from the spatiotemporal evolution model of thermal accumulation. When the prediction result is open ice, the system performs a table lookup and matching in the preset multi-mode operation library and extracts the strategy. Based on this strategy, the system identifies the high-pressure de-icing fluid injector as the primary functional unit for this operation and the infrared heating module as the secondary functional unit. The system then executes a parameter integration function to calculate the specific execution quantities. In this example, the collaborative operation control parameters generated by the system include the start-up delay time of the primary functional unit. Seconds and auxiliary function unit startup delay time =0 seconds, used to achieve timing control for preheating followed by de-icing. The action intensity parameter is set as the action intensity of the dominant functional unit. Intensity of interaction with auxiliary functional units Meanwhile, the system specifies the duration parameter as the duration of the dominant functional unit's operation. Seconds and duration of auxiliary function unit operation Seconds. The system imports all the generated collaborative operation control parameters into the data structure of the preset operation mode to ensure that the anti-freezing execution unit can strictly complete the automated de-icing task according to the predetermined start-up sequence, action intensity, and action duration after arriving at the target area.

[0051] Optionally, generating the adaptive movement path includes: Based on the aforementioned diffusion trend prediction, a series of dynamic candidate endpoints representing the future evolution trajectory of the target area are calculated in chronological order. A search algorithm incorporating a time dimension is employed between the real-time location and the dynamic candidate endpoint. Using the search algorithm, under the physical constraints of the factory environment map information, a path that comprehensively evaluates movement time, target coverage effectiveness, and energy consumption is found as an adaptive movement path.

[0052] Specifically, the continuous evolution prediction is discretized into a set of specific target points that can be used by the path planner. Based on the diffusion trend prediction in the spatiotemporal evolution model of thermal agglomeration, i.e., the diffusion vector and rate, the system calculates the future evolution trajectory of the target region. This calculation starts from the current moment and iterates backward at a fixed time step, such as 1.5 seconds, calculating the position of the centroid or contour of the target region at a series of future time points. The output of this process is a series of dynamic candidate endpoints representing the future evolution trajectory of the target region, ordered by time. The data structure of each endpoint is as follows: That is, in At that time, the target area is expected to move to Location.

[0053] The system searches for feasible paths in an extended space that incorporates the time dimension. It employs a time-enhanced search algorithm, such as a time-based A* algorithm, to perform path searching between the real-time location of the anti-freeze execution unit and a series of dynamic candidate endpoints. This differs from traditional two-dimensional search methods. Unlike traditional search algorithms, this algorithm has a three-dimensional search space. Its state nodes represent not only spatial location but also the time of arrival at that location. At each step of the search, the algorithm rigorously checks whether the path collides spatially with fixed obstacles defined in the factory environment map information.

[0054] The algorithm finds the optimal path from among numerous feasible paths. This search is achieved through a multi-objective path cost function. To evaluate the merits of each potential path, this function comprehensively assesses travel time, target coverage effectiveness, and energy consumption. This cost function... The expression can be: , in, This represents the total travel time from the starting point to the current evaluation point, calculated by dividing the path length by the average cruising speed of the antifreeze execution unit. This represents the estimated energy consumption, which is usually simplified in engineering terms as a function of travel time. A function that is directly proportional to the input. It is a cost item for measuring the effectiveness of target coverage, and its core is the accuracy of time alignment. For example, it can be defined as the timestamp of the time when the anti-freeze execution unit arrives at the candidate endpoint and the timestamp corresponding to that candidate endpoint. The absolute value of the difference between them indicates that the interception is more accurate and the cost item is lower. These are three preset dimensionless weighting coefficients, with values ​​between 0 and 1, configured by the system engineer according to task priority. For example, they can be increased in urgent tasks. The proportion should be increased to achieve the fastest arrival, while in regular tasks, the proportion can be increased. The proportion of energy consumption is adjusted to save energy. The search algorithm will eventually find a path that makes the overall cost function... Find the shortest path and output that path as the final adaptive movement path.

[0055] For example, the system reproduces the adaptive movement path generation process by performing multi-dimensional spatiotemporal path search in the path planning and collaborative control module. First, based on the diffusion trend prediction in the spatiotemporal evolution model of thermal agglomeration, the system calculates a series of dynamic candidate endpoints, ordered by time, representing the future evolution trajectory of the target area, and sets one of the time steps to a specific value. The dynamic candidate endpoint coordinates after seconds are Subsequently, the system acquires the real-time position of the anti-freezing execution unit in the global coordinate system and employs a search algorithm incorporating a time dimension to optimize the path while satisfying the physical constraints of the factory environment map information. When the algorithm evaluates a potential path, the system calls a multi-objective path cost function to calculate the total cost of that path. Set the total travel time from the starting point to the evaluation point. for Seconds, estimated energy consumption The value after function simplification is The cost item for measuring the target coverage effectiveness of time alignment accuracy. for The system engineer configures the dimensionless weighting coefficients according to task priority as follows: , , The system performs a comprehensive evaluation calculation process based on the formulas defined in the instruction manual. By iteratively searching different path nodes, the system eventually finds a path that satisfies the overall cost function. The minimum trajectory serves as the adaptive movement path, achieving an optimal balance between movement time, target coverage effectiveness, and energy consumption.

[0056] S4. Send the spatiotemporal collaborative control command to the antifreeze execution unit to drive the antifreeze execution unit to move along the adaptive movement path and simultaneously prepare to execute the preset operation mode; Optionally, sending the spatiotemporal coordinated control command to the antifreeze execution unit to drive the antifreeze execution unit to move along the adaptive movement path and simultaneously prepare to execute the preset operation mode includes: The spatiotemporal coordinated control command is parsed into motion control sub-commands that drive the antifreeze execution unit to move and operation preparation sub-commands for configuring its composite end effector. During the process of the antifreeze execution unit moving along the adaptive movement path according to the motion control sub-instruction, the operation preparation sub-instruction is executed synchronously; Through the synchronous execution, the composite end effector completes the functional configuration corresponding to the preset operation mode before arriving at the target area.

[0057] Specifically, the system parses and distributes composite instructions issued by the upper-level system. Upon receiving the complete spatiotemporal coordinated control instructions, the airborne main controller of the anti-freeze execution unit immediately initiates a parsing program. This program decomposes the composite instructions into two independent but logically related instruction streams. The first is a motion control sub-instruction, which contains all spatiotemporal waypoint information defined by the adaptive movement path. The second is a job preparation sub-instruction, which contains the coordinated operation control parameters defined in the preset job mode, such as the activation sequence, intensity, and duration of each functional unit. After parsing, the main controller sends the motion control sub-instruction to the underlying mobile platform controller and the job preparation sub-instruction to the controller of the composite end effector.

[0058] The anti-freezing actuator moves precisely along the planned path, while simultaneously pre-configuring the operational functions. This step is the core of the parallel operation. Upon receiving the motion control sub-command, the mobile platform controller immediately drives the motors, controlling the anti-freezing actuator to move along the adaptive movement path according to the waypoint sequence. At the same time, upon receiving the operation preparation sub-command, the controller of the composite end effector also begins executing the corresponding pre-configuration actions. This synchronous execution process can be coordinated through a time synchronization mechanism to ensure that the progress of operation preparation matches the movement progress. For example, the system can adjust the remaining movement time based on the current position and the target area. and the time required for homework preparation To dynamically trigger preparation actions: , in, It is based on collaborative operation control parameters The calculated total preparation time includes the pump's pre-pressurization time, the heating element's preheating time, etc. The trigger condition can be set as follows: , When this condition is met, the job preparation sub-instruction will be executed. It is a safety margin time, such as 1-2 seconds.

[0059] This ensures that the antifreeze actuator is fully operational the instant it arrives at the target area. Through the aforementioned synchronous execution mechanism, while the antifreeze actuator is moving, its composite end effector has already completed all necessary configurations according to the work preparation sub-instructions. For example, the de-icing fluid pipeline has established the preset pressure, the surface temperature of the infrared heating module has reached the set operating temperature, and the fan speed has stabilized at the target value. Therefore, when the real-time position sensor of the antifreeze actuator confirms that it has reached the end of the adaptive movement path, i.e., the target area, the composite end effector can immediately begin efficient antifreeze operations without any waiting, based on the start-up sequence and intensity defined in the collaborative operation control parameters. This process achieves seamless integration of movement and preparation.

[0060] For example, the system reproduces the synchronous preparation process of the anti-freezing execution unit during movement through the instruction parsing and parallel execution logic of the main controller. First, the airborne main controller receives the spatiotemporal cooperative control instruction containing the path waypoints and operational parameters, and parses it into motion control sub-instructions composed of coordinate sequences and cooperative operation control parameters. The mobile platform controller, based on the motion control sub-instruction, drives the unit to move towards the target area at a speed of 0.5 meters per second. During this process, the controller calculates the total preparation time required to complete the task configuration using a function. The total preparation time for this task includes pump pre-pressurization and infrared preheating. Set to 5.0 seconds, with a safety margin time. The remaining travel time is calculated as 2.0 seconds. When the positioning sensor reports that the current location is 4.0 meters away from the destination path, the remaining travel time is calculated. Seconds. The trigger condition is not met at this time. That is, numerical value Greater than and The system remains in standby mode. When the anti-freeze execution unit continues to move to the remaining distance... Measured in meters Seconds. The trigger condition at this time. Upon activation, the composite end effector immediately initiates the work preparation sub-instruction. Through this time synchronization mechanism, the antifreeze actuator completes the establishment of pipeline pressure and preheating of heating elements during its movement before reaching the target area, ensuring that the composite end effector is ready to perform the work immediately upon arrival at the destination.

[0061] S5. During and after the operation of the antifreeze execution unit, real-time temperature data of the target area is collected, and the spatiotemporal evolution model of the heat accumulation state is dynamically corrected based on the real-time temperature data to generate a corrected spatiotemporal evolution model of the heat accumulation state. Optionally, the step of dynamically correcting the spatiotemporal evolution model of the thermal accumulation state based on the real-time temperature data includes: The collected real-time temperature data is compared with the temperature change curve predicted by the spatiotemporal evolution model of the thermal accumulation state before the operation to generate a deviation sequence of quantitative prediction and actual deviation. Based on the deviation sequence, the internal model parameters used to predict diffusion trends and phase transition types in the spatiotemporal evolution model of the thermal accumulation state are adjusted in reverse. The thermal aggregation state spatiotemporal evolution model is updated using the adjusted internal model parameters to generate a corrected thermal aggregation state spatiotemporal evolution model.

[0062] Specifically, this involves quantifying the gap between predictions and reality. During the operation of the anti-freezing unit, the system continuously collects real-time temperature data of the target area at a high frequency, such as twice per second, using thermal imaging sensors. The system then processes this series of collected real-time temperature data points. Compared with the predicted temperature change curve generated by the spatiotemporal evolution model of thermal accumulation state before the operation. Perform time alignment and comparison. This is done by calculating the time points at each time point. Difference on : , The system generates a time series of quantitative deviations between the prediction and the actual results, known as the deviation series. This series visually reflects the dynamic changes in the accuracy of the model's predictions as the operation progresses.

[0063] Based on the quantified deviations, the system traces back and adjusts the internal root causes of inaccurate predictions within the model. The system takes the deviation sequence as input and initiates a parameter optimization algorithm to adjust the internal model parameters used to predict diffusion trends and phase transition types in the spatiotemporal evolution model of thermal accumulation. These internal model parameters are key coefficients used to describe physical processes during model construction; for example, coefficients representing heat diffusion rates, or environmental humidity weights affecting icing type in phase transition type prediction models. The adjustment process can employ a gradient descent-based optimization method, with the following adjustment logic: , in, It is the adjusted internal model parameter vector. It is the parameter vector before adjustment. It is the learning rate, a dimensionless hyperparameter between 0.01 and 0.1, which controls the step size of each adjustment. It is a cost function calculated based on the bias sequence, such as the root mean square error of the bias, which quantifies the overall prediction error under the current model parameters. It is a cost function For parameters The gradient of the formula indicates the direction of parameter adjustment that will most quickly reduce prediction error. Using this formula, the system can automatically fine-tune its internal model parameters in the direction that minimizes prediction error.

[0064] The adjusted parameters are applied to generate a revised prediction model that better reflects the current physical reality. The system uses the internal model parameter vector obtained after the above adjustments. A new spatiotemporal evolution model of the thermal accumulation state is constructed. This process does not start from scratch, but rather updates relevant parts of the existing model with new parameters. For example, the updated diffusion rate is calculated using new coefficients. Ultimately, the system outputs a revised spatiotemporal evolution model of the thermal accumulation state. This revised model provides a more accurate temperature evolution prediction benchmark for subsequent operational effectiveness assessments, ensuring the system can accurately determine whether the task has been completed.

[0065] For example, the system reproduces the dynamic correction process of the spatiotemporal evolution model of the thermal accumulation state by running a parameter optimization algorithm in the feedback and adaptive correction module. First, during the operation of the antifreeze execution unit, the system continuously collects real-time temperature data of the target area using a thermal imaging sensor. A sampling point is set... Real-time temperature data collected at any time for The temperature change curve predicted by the model before the operation was measured in degrees Celsius. The value at that moment Celsius. The system generates a sequence of deviation values ​​between the quantified prediction and the actual deviation by comparing and calculating. The system then uses this deviation sequence as input to a parameter optimization algorithm to reverse-adjust the internal model parameters describing the heat diffusion rate. The internal model parameter vector before adjustment is set. for Learning rate for And the gradient of the current cost function with respect to the parameters Calculated as The system parameter optimization logic executes the calculation process. The system finally applies the adjusted internal model parameters. The heat diffusion coefficient in the model is updated to generate and output a corrected spatiotemporal evolution model of the heat accumulation state. Through this correction process, the system can use real-time feedback to narrow the gap between predictions and physical reality, thereby providing a more accurate temperature evolution benchmark for judging the effectiveness of subsequent operations.

[0066] S6. When it is determined, based on the modified spatiotemporal evolution model of thermal accumulation state, that the temperature of the target area has recovered to the safe threshold defined by the system function, a task termination instruction is generated and executed.

[0067] Optionally, when it is determined, based on the modified spatiotemporal evolution model of thermal accumulation state, that the temperature of the target region has recovered to a safe threshold defined by the system function, generating and executing a task termination instruction includes: Based on the modified spatiotemporal evolution model of thermal accumulation state, the predicted temperature value of the target area is obtained in real time; The predicted temperature value is compared with the safety threshold. When the predicted temperature value continuously reaches or exceeds the safety threshold and remains at the preset duration, a task termination command is generated and sent to the antifreeze execution unit.

[0068] Specifically, a smooth and forward-looking temperature assessment value is obtained, rather than directly using raw sensor data that may contain glitches. Based on the spatiotemporal evolution model of the heat accumulation state, which is continuously dynamically corrected from real-time temperature data during anti-freezing operations, the predicted temperature value of the target area at the next moment is calculated and output in real time. This predicted value is the model's inference of the current temperature state evolution trend based on its internally updated physical parameters. It naturally filters out high-frequency noise and better reflects the true temperature recovery trend.

[0069] The system confirms the temperature recovery status and filters out unstable transient changes. The real-time predicted temperature value is continuously compared to a defined safety threshold. This safety threshold is set based on the radiator's material properties and operating conditions; for example, a typical safety threshold is 5 degrees Celsius. To prevent false completion judgments due to brief temperature rises, a time-accumulation judgment logic is integrated. Specifically, it monitors the duration for which the predicted temperature value remains above the safety threshold. This judgment logic can be defined as: if and only if... Consecutive validity, and duration When this condition is met, the termination condition is triggered. At a certain point in time The predicted temperature value, It is a safety threshold. It is a preset duration to ensure the stability of the state. Its engineering value range is usually between 5 and 15 seconds. It is used to confirm that the temperature rise is not a temporary phenomenon, but that the operation has achieved fundamental results.

[0070] The antifreeze operation is officially concluded. Once the aforementioned time accumulation judgment condition is met, the system's main control logic determines that the antifreeze task has been successfully completed and immediately generates a standardized task termination command. This command, as a command data packet, is sent to the onboard main controller of the antifreeze execution unit via the wireless communication network. After receiving and verifying the command, the execution unit immediately stops all operational functions of its composite end effector, such as shutting down the heating module and stopping the spraying of de-icing fluid, and enters standby or returns to the charging station state according to the system's higher-level scheduling instructions.

[0071] For example, the system reproduces the automated closed-loop management of antifreeze operations by simulating task termination logic based on predicted temperature trends in the task execution module. First, the system calculates and obtains the predicted temperature value of the target area in real time based on a modified spatiotemporal evolution model of heat accumulation state. This value can naturally filter out high-frequency noise in the original sensor data and reflect the true temperature recovery trend. Set the system-defined safe threshold for anti-freezing operations. for The temperature is measured in degrees Celsius, and the preset time required to determine the stability of the state is determined. for Seconds. During continuous monitoring, when the system obtains the predicted temperature value at the current sampling time... At 100 degrees Celsius, the system performs a temperature state comparison process. That is, numerical calculation is The basic condition is met. The system then uses time-accumulation logic to check the duration for which this condition is continuously satisfied. The current status is fed back by the timer. The cumulative total has reached Seconds. The system then executes the duration determination logic. That is, numerical calculation is Established. Since the predicted temperature value consistently reaches the safety threshold and remains there for the preset duration, the system's main control logic determines that the anti-freeze task has been successfully completed. It then generates a standardized task termination command and sends it to the anti-freeze execution unit. Upon receiving and verifying the command, the anti-freeze execution unit immediately stops the heating and spraying functions of its composite end effector, successfully concluding the anti-freeze operation.

[0072] Based on the same inventive concept, such as Figure 4 As shown, the present invention also provides a mobile control system for localized heat accumulation in an air-cooled radiator, the system comprising: The data acquisition module is used to acquire thermal imaging data of the surface of the air-cooled radiator; The state modeling and prediction module is used to dynamically model the thermal imaging data and generate a spatiotemporal evolution model of thermal aggregation state that includes dynamic contour prediction of the target area, diffusion trend prediction, and phase change type prediction. The path planning and collaborative control module is used to generate an adaptive movement path that matches the arrival time of the antifreeze execution unit with the predicted state of the target area based on the spatiotemporal evolution model of the heat accumulation state, and to determine the preset operation mode corresponding to the phase change type prediction, and to integrate and generate spatiotemporal collaborative control instructions. The anti-freeze execution module is used to send the spatiotemporal coordinated control command to the anti-freeze execution unit to drive the anti-freeze execution unit to move along the adaptive movement path and simultaneously prepare to execute the preset operation mode. The feedback and adaptive correction module is used to collect real-time temperature data of the target area during and after the operation of the antifreeze execution unit, and dynamically correct the spatiotemporal evolution model of the heat accumulation state based on the real-time temperature data to generate the corrected spatiotemporal evolution model of the heat accumulation state. The task execution module is used to generate and execute a task termination command when it is determined, based on the modified spatiotemporal evolution model of thermal accumulation state, that the temperature of the target area has recovered to a safe threshold defined by the system function.

[0073] It should be noted that the electrical connections between the various units described above do not necessarily represent direct or indirect connections. Any indirect connection method can be applied to the embodiments of the present invention as long as it achieves the purpose of the present invention. The above descriptions are merely exemplary embodiments of the present invention and should not be construed as limiting the scope of the present invention.

[0074] All equivalent changes and modifications made in accordance with the teachings of this invention are still within the scope of this invention. Those skilled in the art will readily conceive of other embodiments of this invention upon considering the specification and the disclosure of practical truth. This application is intended to cover any variations, uses, or adaptations of this invention that follow the general principles of this invention and include common knowledge or conventional techniques in the art not described herein.

Claims

1. A method for controlling the movement of air-cooled radiators to prevent localized heat accumulation, characterized in that, The method includes: Acquire thermal imaging data of the surface of an air-cooled radiator; Dynamic modeling is performed on the thermal imaging data to generate a spatiotemporal evolution model of thermal aggregation state that includes dynamic contour prediction, diffusion trend prediction, and phase transition type prediction of the target area. Based on the spatiotemporal evolution model of the heat accumulation state, an adaptive movement path is generated that can match the arrival time of the antifreeze execution unit with the predicted state of the target area, and a preset operation mode corresponding to the phase change type prediction is determined, and spatiotemporal collaborative control instructions are generated. Send the spatiotemporal coordinated control command to the antifreeze execution unit to drive the antifreeze execution unit to move along the adaptive movement path and simultaneously prepare to execute the preset operation mode; During and after the operation of the antifreeze execution unit, real-time temperature data of the target area is collected, and the spatiotemporal evolution model of the heat accumulation state is dynamically corrected based on the real-time temperature data to generate a corrected spatiotemporal evolution model of the heat accumulation state. When it is determined, based on the modified spatiotemporal evolution model of thermal accumulation state, that the temperature of the target area has recovered to the safe threshold defined by the system function, a task termination instruction is generated and executed.

2. The method for controlling the movement of localized heat accumulation in an air-cooled radiator according to claim 1, characterized in that, The step of dynamically modeling the thermal imaging data to generate a spatiotemporal evolution model of the thermal aggregation state, which includes dynamic contour prediction, diffusion trend prediction, and phase transition type prediction of the target area, includes: The thermal imaging data is divided into grids to obtain temperature distribution information for multiple grid cells of interest. Based on the temperature distribution information, abnormal grid cells with temperatures below a preset threshold are identified; Multiple anomalous grid cells captured within a continuous time series are aggregated and tracked to generate an initial anomalous region profile and its historical change trajectory that record the state evolution of the region. Based on the historical change trajectory, the future diffusion vector and rate of the initial anomaly region contour are calculated, and the phase transition type of the initial anomaly region is predicted by combining the acquired environmental state information. By integrating the initial anomaly region contour, the future diffusion vector and rate, and the phase transition type, a spatiotemporal evolution model of the thermal accumulation state is constructed and output.

3. The method for controlling the movement of localized heat accumulation in an air-cooled radiator according to claim 2, characterized in that, The step of performing grid segmentation on the thermal imaging data to obtain temperature distribution information for multiple grid cells of interest includes: The loading process enables the digitalization of the virtual mesh region corresponding to the surface of the air-cooled radiator. Obtain the region focus strategy that defines the scanning priority and scanning frequency for different grid regions; Based on the region focus strategy, the scanning parameters of different virtual grid regions are dynamically adjusted to obtain temperature distribution information of multiple focus grid cells.

4. The method for controlling the movement of localized heat accumulation in an air-cooled radiator according to claim 1, characterized in that, The integrated spatiotemporal collaborative control commands include: Obtain the real-time position of the antifreeze execution unit; Load the factory environment map information that stores information on fixed obstacles and walkable paths; Starting from the real-time location and taking the future location corresponding to the dynamic contour prediction in the spatiotemporal evolution model of the thermal aggregation state as the dynamic endpoint, a spatiotemporal path search is performed in conjunction with the factory environment map information to generate an adaptive movement path. Based on the phase transition type prediction, a preset operation mode is determined; The adaptive movement path is data-bound with the preset operation mode to generate spatiotemporal collaborative control commands.

5. The method for controlling the movement of localized heat accumulation in an air-cooled radiator according to claim 4, characterized in that, The preset operation mode also includes: The phase transition type prediction is matched with the operation strategies in the preset multi-mode operation library to determine the dominant functional unit and the auxiliary functional unit. Based on the matching results, a set of collaborative operation control parameters is generated to define the start-up order, intensity, and duration of the dominant functional unit and the auxiliary functional unit. These collaborative operation control parameters are part of the preset operation mode.

6. The method for controlling the movement of localized heat accumulation in an air-cooled radiator according to claim 1, characterized in that, Sending the spatiotemporal coordinated control command to the antifreeze execution unit to drive the antifreeze execution unit to move along the adaptive movement path and simultaneously prepare to execute the preset operation mode includes: The spatiotemporal coordinated control command is parsed into motion control sub-commands that drive the antifreeze execution unit to move and operation preparation sub-commands for configuring its composite end effector. During the process of the antifreeze execution unit moving along the adaptive movement path according to the motion control sub-instruction, the operation preparation sub-instruction is executed synchronously; Through the synchronous execution, the composite end effector completes the functional configuration corresponding to the preset operation mode before arriving at the target area.

7. The method for controlling the movement of localized heat accumulation in an air-cooled radiator according to claim 1, characterized in that, The dynamic correction of the spatiotemporal evolution model of the thermal accumulation state based on the real-time temperature data includes: The collected real-time temperature data is compared with the temperature change curve predicted by the spatiotemporal evolution model of the thermal accumulation state before the operation to generate a deviation sequence of quantitative prediction and actual deviation. Based on the deviation sequence, the internal model parameters used to predict diffusion trends and phase transition types in the spatiotemporal evolution model of the thermal accumulation state are adjusted in reverse. The thermal aggregation state spatiotemporal evolution model is updated using the adjusted internal model parameters to generate a corrected thermal aggregation state spatiotemporal evolution model.

8. The method for controlling the movement of localized heat accumulation in an air-cooled radiator according to claim 1, characterized in that, When the temperature of the target region has recovered to the safe threshold defined by the system function based on the modified spatiotemporal evolution model of thermal accumulation state, the generation and execution of the task termination instruction includes: Based on the modified spatiotemporal evolution model of thermal accumulation state, the predicted temperature value of the target area is obtained in real time; The predicted temperature value is compared with the safety threshold. When the predicted temperature value continuously reaches or exceeds the safety threshold and remains at the preset duration, a task termination command is generated and sent to the antifreeze execution unit.

9. The method for controlling the movement of localized heat accumulation in an air-cooled radiator according to claim 4, characterized in that, The generation of the adaptive movement path includes: Based on the aforementioned diffusion trend prediction, a series of dynamic candidate endpoints representing the future evolution trajectory of the target area are calculated in chronological order. A search algorithm incorporating a time dimension is employed between the real-time location and the dynamic candidate endpoint. Using the search algorithm, under the physical constraints of the factory environment map information, a path that comprehensively evaluates movement time, target coverage effectiveness, and energy consumption is found as an adaptive movement path.

10. A movement control system for localized heat accumulation in an air-cooled radiator, applied to the movement control method for localized heat accumulation in an air-cooled radiator as described in any one of claims 1-9, characterized in that, The system includes: The data acquisition module is used to acquire thermal imaging data of the surface of the air-cooled radiator; The state modeling and prediction module is used to dynamically model the thermal imaging data and generate a spatiotemporal evolution model of thermal aggregation state that includes dynamic contour prediction of the target area, diffusion trend prediction, and phase change type prediction. The path planning and collaborative control module is used to generate an adaptive movement path that matches the arrival time of the antifreeze execution unit with the predicted state of the target area based on the spatiotemporal evolution model of the heat accumulation state, and to determine the preset operation mode corresponding to the phase change type prediction, and to integrate and generate spatiotemporal collaborative control instructions. The anti-freeze execution module is used to send the spatiotemporal coordinated control command to the anti-freeze execution unit to drive the anti-freeze execution unit to move along the adaptive movement path and simultaneously prepare to execute the preset operation mode. The feedback and adaptive correction module is used to collect real-time temperature data of the target area during and after the operation of the antifreeze execution unit, and dynamically correct the spatiotemporal evolution model of the heat accumulation state based on the real-time temperature data to generate the corrected spatiotemporal evolution model of the heat accumulation state. The task execution module is used to generate and execute a task termination command when it is determined, based on the modified spatiotemporal evolution model of thermal accumulation state, that the temperature of the target area has recovered to a safe threshold defined by the system function.

Images