A multi-task-based thermal management method and device, electronic equipment and storage medium

Abstract

This application discloses a multi-task-based thermal management method, apparatus, electronic device, and storage medium. This application can perform real-time analysis of the pending task flow of a spacecraft to obtain task analysis results, wherein the pending task flow includes at least one pending task. Based on the task analysis results, the at least one pending task is divided into multiple task types, including heat-sensitive tasks, thermoelastic tasks, and heat-tolerant tasks. Sensor parameters of the spacecraft system are monitored; when sensor parameters are found to be inconsistent with sensor thresholds, a corresponding task scheduling strategy is determined for each pending task based on the sensor parameters and the task type of each pending task. The sensors include temperature parameters and / or power parameters. Based on the task scheduling strategy, each pending task is scheduled to achieve thermal shock suppression of the spacecraft.

Description

Technical Field

[0001] This application relates to the field of computer technology, and more specifically to a multi-task-based thermal management method, apparatus, electronic device, and storage medium. Background Technology

[0002] With the increasing integration of spacecraft payloads and the surge in computing power of edge computing devices, the internal thermal load of the system exhibits significant spatiotemporal nonuniformity and transient bursts. When the system faces instantaneous peak thermal shocks (such as high-power microwave transmission and high-intensity data processing), the hardware heat dissipation rate often lags behind the heat generation rate, leading to localized overheating of equipment, triggering protective throttling of the system, or even shutdown, severely affecting the continuity of high-priority tasks. Summary of the Invention

[0003] This application proposes a multi-task-based thermal management method, apparatus, electronic device, and storage medium, which aims to suppress thermal shock through software-defined task scheduling without increasing the hardware heat dissipation load.

[0004] This application provides a multi-task-based thermal management method, including:

[0005] The task flow to be processed of the spacecraft is analyzed in real time to obtain the task analysis result, wherein the task flow to be processed includes at least one task to be processed;

[0006] Based on the task parsing results, the at least one task to be processed is divided into multiple task types, including heat-sensitive tasks, thermally elastic tasks, and heat-tolerant tasks.

[0007] The system monitors the sensor parameters of the spacecraft system. When the sensor parameters are found to be inconsistent with the sensor threshold, the system determines the corresponding task scheduling strategy for each task based on the sensor parameters and the task type of each task to be processed. The sensors include temperature parameters and / or power parameters.

[0008] Based on the task scheduling strategy, each task to be processed is scheduled.

[0009] Accordingly, embodiments of this application also provide a multi-task-based thermal management device, including:

[0010] The parsing unit is used to perform real-time parsing of the spacecraft's pending task flow to obtain the task parsing result, wherein the pending task flow includes at least one pending task.

[0011] A partitioning unit is used to partition the at least one task to be processed based on the task parsing result to obtain multiple task types, wherein the multiple task types include heat-sensitive tasks, heat-elastic tasks, and heat-tolerant tasks.

[0012] A determination unit is used to monitor sensor parameters of the spacecraft system. When the sensor parameters are found to be inconsistent with the sensor threshold, the unit determines the corresponding task scheduling strategy for each task to be processed based on the sensor parameters and the task type of each task to be processed. The sensors include temperature parameters and / or power parameters.

[0013] The scheduling unit is used to schedule each task to be processed based on the task scheduling strategy.

[0014] In some embodiments, the partitioning unit includes:

[0015] The acquisition subunit is used to acquire, based on the task parsing result, the task time information, task hardware association information, task business logic information and task priority corresponding to each task to be processed.

[0016] The sub-unit is used to divide the at least one task to be processed according to at least one of the task time information, task hardware association information, task corresponding business logic information and task priority, so as to obtain multiple task types.

[0017] In some embodiments, the determining unit includes:

[0018] A comparison subunit is used to compare the temperature parameter with at least one preset threshold range;

[0019] A determination subunit is used to determine the corresponding task scheduling strategy for each task to be processed based on the comparison results of the temperature parameter and the at least one threshold interval and the task type of the task to be processed.

[0020] In some embodiments, the determining subunit includes:

[0021] The first execution module is used to perform time-power coupling control when the temperature parameter is within a first threshold range. The time-power coupling control includes performing dynamic voltage frequency adjustment on the thermal elastic task, reducing the operating frequency of the thermal elastic task, and pausing the background calculation of the thermal tolerance task.

[0022] The second execution module is used to perform space power coupling control when the temperature parameter is in the second threshold range. The space power coupling control includes migrating the thermally tolerant task from the high-temperature core to the low-temperature core and down-clocking or delaying the thermal elastic task.

[0023] The third execution module is used to perform emergency protection control when the temperature parameter exceeds the third threshold. The emergency protection control includes suspending or terminating the heat-tolerant task, reducing the frequency of the thermal elastic task to the lowest level, keeping the heat-sensitive task running at full speed, or triggering hardware-level clock gating protection.

[0024] In some embodiments, the thermal management device proposed in this application further includes:

[0025] The acquisition unit is used to acquire task scheduling feedback information during the scheduling process of the task to be processed;

[0026] The adjustment unit is used to adjust the strategy determination model according to the task scheduling feedback information to obtain the adjusted strategy determination model.

[0027] An execution unit is used to update the task scheduling strategy corresponding to each pending task by using the adjusted strategy to determine the model.

[0028] In some embodiments, the adjustment unit includes:

[0029] The calculation subunit is used to calculate thermal risk information and performance loss information based on the task scheduling feedback information;

[0030] A sub-unit is constructed to construct target loss information based on the thermal risk information and the performance loss information;

[0031] The adjustment subunit is used to adjust the policy determination model based on the target loss information to obtain the adjusted policy determination model.

[0032] In some embodiments, the computing subunit includes:

[0033] The first calculation subunit is used to calculate the thermal risk information based on the temperature feedback information and the temperature gradient information;

[0034] The second calculation subunit is used to calculate the task execution quality information based on the task running frequency information and the task execution quality information.

[0035] This application also provides a computer program product or computer program that includes computer instructions stored in a computer-readable storage medium. A processor of an electronic device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the electronic device to perform the methods provided in the various alternative embodiments described above.

[0036] Accordingly, embodiments of this application also provide a storage medium storing instructions, which, when executed by a processor, implement any of the multi-task-based thermal management methods provided in embodiments of this application.

[0037] This application embodiment can perform real-time analysis of the spacecraft's pending task flow to obtain task analysis results. The pending task flow includes at least one pending task. Based on the task analysis results, the at least one pending task is divided into multiple task types, including heat-sensitive tasks, thermoelastic tasks, and heat-tolerant tasks. The sensor parameters of the spacecraft system are monitored. When the sensor parameters are found to be inconsistent with the sensor thresholds, a corresponding task scheduling strategy is determined for each pending task based on the sensor parameters and the task type of each pending task. The sensors include temperature parameters and / or power parameters. Based on the task scheduling strategy, each pending task is scheduled to achieve thermal shock suppression of the spacecraft. Attached Figure Description

[0038] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0039] Figure 1 This is a schematic diagram of a scenario for the multi-task-based thermal management method provided in an embodiment of this application;

[0040] Figure 2 This is a flowchart illustrating the multi-task-based thermal management method provided in an embodiment of this application;

[0041] Figure 3 This is another flowchart illustrating the multi-task-based thermal management method provided in this application embodiment;

[0042] Figure 4 This is another scenario illustration of the multi-task-based thermal management method provided in the embodiments of this application;

[0043] Figure 5 This is a schematic diagram of the structure of the multi-task-based thermal management device provided in the embodiments of this application;

[0044] Figure 6 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0045] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. However, the described embodiments are only some embodiments of this application, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0046] This application proposes a multi-task-based thermal management method, which can be executed by a multi-task-based thermal management device integrated into an electronic device. The electronic device may include at least one of a terminal and a server. That is, the multi-task-based thermal management method proposed in this application can be executed by a terminal, a server, or jointly by a terminal and a server capable of communicating with each other.

[0047] The terminals may include, but are not limited to, spacecraft, smartphones, tablets, laptops, personal computers (PCs), smart home appliances, wearable electronic devices, VR / AR devices, in-vehicle terminals, intelligent voice interaction devices, and so on.

[0048] A server can be an interconnecting server between multiple heterogeneous systems or a backend server. It can also be an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, and big data and artificial intelligence platforms, etc.

[0049] It should be noted that the embodiments of this application can be applied to various scenarios, including but not limited to commercial aerospace, cloud technology, artificial intelligence, smart transportation, and assisted driving.

[0050] In one embodiment, such as Figure 1The multi-task-based thermal management device can be integrated into electronic devices such as terminals or servers to implement the multi-task-based thermal management method proposed in this application. Specifically, the electronic device can perform real-time analysis of the spacecraft's pending task flow to obtain task analysis results, wherein the pending task flow includes at least one pending task; based on the task analysis results, the at least one pending task is divided into multiple task types, wherein the multiple task types include heat-sensitive tasks, thermoelastic tasks, and heat-tolerant tasks; the sensor parameters of the spacecraft system are monitored, and when the sensor parameters are found to be inconsistent with the sensor thresholds, a corresponding task scheduling strategy is determined for each pending task based on the sensor parameters and the task type of each pending task, wherein the sensors include temperature parameters and / or power parameters; and each pending task is scheduled based on the task scheduling strategy.

[0051] The following will provide a detailed description of each example. It should be noted that the order of description of the following embodiments is not intended to limit the preferred order of the embodiments.

[0052] This application will describe the embodiments from the perspective of a multi-task-based thermal management device, which can be integrated into an electronic device, such as a server or a terminal.

[0053] like Figure 2 The present invention provides a multi-task-based thermal management method, the specific process of which includes:

[0054] 101. Perform real-time analysis on the unprocessed task flow of the spacecraft to obtain the task analysis results, wherein the unprocessed task flow includes at least one unprocessed task.

[0055] In some embodiments, a pending task flow refers to a set of tasks that the spacecraft needs to perform at the current moment or within a preset time window. It is typically organized in the form of a task queue, task list, or schedule, and contains at least one pending task. This task flow reflects the temporal distribution of the spacecraft's computational payload workload and serves as the input data source for subsequent task classification and thermal management decisions.

[0056] In some embodiments, a task to be processed refers to the smallest execution unit constituting the task flow, such as a computation process, a data processing job, a sensor readout command, or a communication transmission task. Each task to be processed has independent attribute parameters, including but not limited to deadline, period, resource consumption, computational intensity, priority label, etc.

[0057] In some embodiments, the task parsing result refers to the structured information output by the system after analyzing, extracting, and quantifying each task in the task flow. This result includes at least the real-time requirements of each task (e.g., hard real-time / soft real-time / best-effort delivery), thermal contribution (e.g., predicted power consumption increment or thermal shock intensity), resource dependencies (e.g., locked caches, dedicated hardware accelerators), and migration feasibility (e.g., whether context switching and checkpoint recovery are supported). The task parsing result serves as the basis for the execution of subsequent task classification and thermal shock suppression modules.

[0058] In some embodiments, when performing real-time parsing of the spacecraft's pending task flow, the system first accesses the task control block in the operating system kernel to directly extract the static attributes of each pending task that are determined at compile time or startup time, including deadline, cycle, priority, worst-case execution time, locked cache lines, dedicated hardware accelerator identifiers, and functional criticality level labels. Simultaneously, the system maintains historical execution data within a sliding window, statistically analyzing the average power consumption increment and temperature rise rate caused by the last execution of each task, and obtaining measured migration overhead through actual task migration tests. Dynamic relaxation time is calculated by combining the current time, remaining execution cycles, and worst-case execution time. Based on this, the system further establishes a directed dependency graph between tasks, identifies tasks on critical paths through reverse topology sorting, analyzes the holding relationships of semaphores or mutexes, and detects resource lock propagation to mark resource-critical tasks. Furthermore, the system combines a long short-term memory network load prediction model, inputting task scheduling logs and historical execution profiles to predict the types of tasks arriving at future times, estimate the load and thermal shock probability, and pre-inject the prediction results into the parsing result cache. Finally, the above parsing process outputs a structured parsing result vector for each task, including task classification, quantified thermal contribution value, migration overhead, and dependency flags, which serve as the input basis for subsequent task classification and thermal shock suppression modules.

[0059] 102. Based on the task parsing results, at least one task to be processed is divided into multiple task types, wherein the multiple task types include heat-sensitive tasks, heat-elastic tasks, and heat-tolerant tasks.

[0060] In some embodiments, based on the task's criticality to system survival, real-time requirements, and thermal management flexibility, the present invention classifies tasks to be processed into three levels:

[0061] The first level is for heat-sensitive tasks. These tasks involve the core functions of the system's survival and have zero tolerance for latency. They must not be migrated or suspended under any circumstances, and their computing resources and execution timing must be guaranteed first.

[0062] The second level is thermally elastic tasks. These tasks have a certain degree of time redundancy, allowing for short-term delays during thermal shocks, and support reducing power consumption through dynamic voltage and frequency adjustments, thereby providing buffering capabilities for the system.

[0063] The third level is thermally tolerant tasks. These tasks are non-critical background tasks with the lowest real-time requirements. They support migration across cores to low-temperature regions or complete suspension when necessary, and can serve as a flexible resource for system thermal capacity adjustment.

[0064] In some embodiments, the step "based on the task parsing results, dividing at least one task to be processed to obtain multiple task types" includes:

[0065] Based on the task parsing results, obtain the task time information, task hardware association information, task business logic information, and task priority for each task to be processed;

[0066] Based on at least one of the following: task time information, task hardware association information, task corresponding business logic information, and task priority, at least one task to be processed is divided into multiple task types.

[0067] The task time information can include the task's slack time, execution time, etc. Task hardware association information can include whether the task is directly bound to physical sensors, has a dedicated driver interface for the actuator, has locked a specific core's cache, uses a non-virtualizable dedicated hardware accelerator, or is positively correlated with processor frequency, etc. The corresponding business logic information can include whether the task belongs to any critical business logic, etc.

[0068] In some embodiments, when a task's deadline meets hard real-time constraints and missing the deadline would cause system malfunction, or the task is directly bound to a dedicated driver interface of a physical sensor or actuator, or the task locks a cache of a specific core or uses a dedicated hardware accelerator that cannot be virtualized, the task is marked as a heat-sensitive task and given the highest execution priority and exclusive computing resources.

[0069] When the relaxation time of a task is greater than the preset hot adjustment time window, or the execution time of a task is positively correlated with the processor frequency and the throughput after frequency reduction can still meet the minimum threshold of the business function, or the task belongs to non-critical business logic, the task will be marked as a hot elastic task, supporting time-triggered reordering of execution.

[0070] When a task has the lowest priority in the system and no user interface, or when the task's data is stored in shared memory or remote storage and the migration overhead is less than a preset percentage of the recalculation overhead, or when the task supports checkpointing and recovery mechanisms, the task is marked as a hot-tolerant task, supporting cross-core migration or suspension.

[0071] In a specific implementation, this invention provides a multi-criteria task classification method based on Task Control Blocks (TCBs). The system first parses the Task Control Block of each task to be executed to obtain its deadline parameter. Periodic parameters The remaining number of execution cycles and the current time. For those satisfying hard real-time constraints ( Tasks that will cause system malfunction if they miss their deadline, or tasks that are directly bound to dedicated driver interfaces of physical sensors or actuators, are part of the system watchdog monitoring chain (an interruption of such a task will cause the system to enter safe mode), or tasks that lock the L1 / L2 cache of a specific core, or tasks that use non-virtualized dedicated hardware accelerators (such as encryption chips) and cannot undergo context switching migration, are marked as L1 hot-sensitive tasks by the system. L1 tasks are given the highest execution priority and exclusive access to computing resources, and their operation is guaranteed in any hot management actions.

[0072] For tasks that do not meet the L1 condition, the system further determines the L2 thermoelastic task based on relaxation time and DVFS adaptability. The relaxation time of the task is calculated as follows:

[0073] ,

[0074] like Greater than the system's preset thermal conditioning time window (For example, 500ms) indicates that the task allows for a delay; or the task's execution time is positively correlated with the processor frequency (i.e., computational load). If a task is classified as an L2 thermal elastic task, and its throughput still meets the minimum threshold Xn, (n=1,2,…,n) of the business function after the frequency is reduced (where n identifies different business functions); or if the task belongs to data buffering, prefetching, or non-core business logic processing, and its priority is lower than the system's critical task threshold, then the system marks it as an L2 thermal elastic task. L2 tasks support time-triggered execution order reordering and can be down-frequencyed or briefly delayed during thermal shocks.

[0075] For tasks that do not belong to L1 or L2, the system uses background attribute and location independence rules to determine whether they are L3 heat-tolerant tasks. If the task has the lowest priority level in the system (such as IDLE or BEST_EFFORT in Linux), has no user interface, and is a background maintenance, log archiving, or idle resource filling task; or if the task's data is stored in shared memory or remote storage, and its migration overhead is less than 20% of the recalculation overhead, or it does not depend on a specific core's local cache; or if the task supports checkpointing / recovery mechanisms, its execution state can be quickly saved to non-volatile storage, and it can be suspended at any time, then the system marks it as an L3 heat-tolerant task. During thermal shocks, L3 tasks can be migrated to a low-temperature core, downclocked, or suspended to prioritize the operation of L1 and L2 tasks. Through this multi-criteria step-by-step determination, the system achieves a fine-grained classification of task thermal sensitivity, providing accurate decision-making basis for subsequent collaborative thermal shock suppression.

[0076] 103. Monitor the sensor parameters of the spacecraft system. When the sensor parameters are found to be inconsistent with the sensor threshold, determine the corresponding task scheduling strategy for each task to be processed based on the sensor parameters and the task type of each task to be processed. The sensors include temperature parameters and / or power parameters.

[0077] In some embodiments, temperature parameters refer to thermal state data collected by temperature sensors located on the various computing cores, key electronic components, or heat dissipation channels of the spacecraft system, such as core junction temperature, shell temperature, and radiator inlet / outlet coolant temperature. This parameter characterizes the current system's thermal load level and heat distribution uniformity, and is one of the core bases for determining whether thermal shock suppression is triggered and selecting scheduling strategies.

[0078] In some embodiments, the power parameter refers to the instantaneous total power consumption of the spacecraft system or the power consumption value of a critical mission subsystem, such as the real-time power calculated by current and voltage sensors. This parameter reflects the current electrothermal conversion intensity of the system and is used to predict thermal shock trends (such as instantaneous power exceeding a preset threshold), and together with the temperature parameter, determines the triggering timing and intensity of the mission scheduling strategy.

[0079] In some embodiments, the task scheduling strategy refers to a set of control actions dynamically determined based on the current sensor parameters (temperature, power) and the task type (L1 thermally sensitive, L2 thermoelastic, L3 thermally tolerant) of each task to be processed. This strategy includes at least a combination of the following three types of actions: time-dimensional scheduling (e.g., postponing L2 tasks to low-power time slots), spatial-dimensional migration (e.g., migrating L3 tasks from high-temperature cores to low-temperature backup units), and fine-tuning of power consumption (e.g., dynamic voltage and frequency adjustment to reduce the system's instantaneous total power consumption while ensuring L1 task latency). The specific strategy combination used is determined in real-time by a preset rule engine or online learning model based on whether the current sensor parameters exceed thresholds and the degree of deviation.

[0080] In some embodiments, the system uses temperature and power sensors deployed on the spacecraft's computing cores, power distribution units, and critical heat dissipation paths to collect temperature parameters (e.g., core junction temperature) and power parameters (instantaneous total system power consumption) in real time. The sensor sampling frequency is matched to the task scheduling cycle (e.g., once every 100 milliseconds), and the collected parameters are sent to the thermal management controller after analog-to-digital conversion. The controller has preset sensor thresholds matching the task level, including temperature and power thresholds. When any sensor parameter is detected to exceed the corresponding threshold, the system immediately triggers the scheduling strategy decision process.

[0081] In some embodiments, such as Figure 3 As shown in the embodiments of this application, the method proposed may include a multi-dimensional task feature classification module, a collaborative thermal shock suppression module, and an incremental online learning strategy library. The task feature classification module is used to analyze the task flow of tasks to be executed by the system in real time, and based on preset real-time indicators and thermal sensitivity indicators, statically or dynamically classifies tasks into thermally sensitive tasks, thermally elastic tasks, and thermally tolerant tasks. The thermal shock suppression module is connected to the classification module and is used to execute collaborative control actions, including task time-domain peak shifting, spatial computing power migration, and dynamic voltage and frequency adjustment, when the instantaneous thermal load of the system is detected to exceed a preset threshold. The online learning strategy module is used to store multi-condition thermal management templates and iteratively update the parameters of the collaborative control actions based on the historical thermal response data of the system operation using a reinforcement learning algorithm.

[0082] Among them, the collaborative thermal shock suppression module monitors the total power consumption and critical node temperature of the system in real time. When a thermal shock with instantaneous power consumption exceeding a preset threshold (e.g., 2.1kW) is predicted, three-dimensional collaborative control is executed: in the time dimension, the L2 thermal elasticity task is postponed to the valley range of the power consumption curve; in the spatial dimension, the L3 thermal tolerance task is migrated from the high-temperature core to a standby unit with thermal redundancy; at the same time, the collaborative dynamic voltage and frequency adjustment algorithm dynamically compresses the instantaneous total entropy increase of the system while ensuring the delay of the L1 thermally sensitive task. In addition, the system is equipped with an incremental online learning strategy library, in which the static library pre-stores 137 preset thermal feature templates for typical operating conditions (such as deep space maneuvers, high-throughput data transmission, etc.) for quickly matching the current operating state and calling the corresponding strategy.

[0083] In some embodiments, when the sensor parameters include temperature parameters, the step "determine the corresponding task scheduling strategy for each task to be processed based on the sensor parameters and the task type of each task to be processed" may include:

[0084] The temperature parameter is compared with at least one preset threshold range;

[0085] Based on the comparison results of the temperature parameter and the at least one threshold range, as well as the task type of the task to be processed, a corresponding task scheduling strategy is determined for each task to be processed.

[0086] In some embodiments, the step "determining the corresponding task scheduling strategy for each task to be processed based on the comparison result of the temperature parameter and the at least one threshold range and the task type of the task to be processed" may include:

[0087] When the temperature parameter is within the first threshold range, time-power coupling control is executed, wherein the time-power coupling control includes performing dynamic voltage frequency adjustment on the thermal elastic task, reducing the operating frequency of the thermal elastic task, and pausing the background calculation of the thermal tolerance task.

[0088] When the temperature parameter is in the second threshold range, space power coupling control is executed, wherein the space power coupling control includes migrating the thermally tolerant task from the high-temperature core to the low-temperature core, and down-clocking or delaying the scheduling of the thermally elastic task.

[0089] When the temperature parameter exceeds the third threshold, emergency protection control is executed. The emergency protection control includes suspending or terminating the heat-tolerant task, reducing the frequency of the thermal elastic task to the lowest level, keeping the heat-sensitive task running at full speed, or triggering hardware-level clock gating protection.

[0090] In specific implementation methods, such as Figure 4As shown, this invention proposes a three-dimensional coordinated control mechanism of time, space, and power consumption to overcome the limitations of traditional thermal management that relies on only a single dimension (such as simple frequency reduction or simple task migration). The core of this mechanism lies in coupling task scheduling and power consumption adjustment methods based on the dynamic relationship between temperature gradient and multi-level thresholds to achieve Pareto optimality of system thermal safety and task performance.

[0091] The system first performs real-time thermal sensing and gradient calculation. It collects the real-time temperature of each computational core. Calculate the average temperature of the system and highest temperature And obtain the temperature gradient .like Greater than the preset non-uniformity threshold If a local hotspot is identified, spatial dimension task migration will be triggered first; if and Approaching the middle threshold If the temperature rises, it is determined to be an overall temperature rise, and frequency modulation operation in the time dimension is triggered first.

[0092] After obtaining the temperature gradient and the maximum temperature, the system... The interval in which it is located executes hierarchical coupling control. When Within the first threshold range ( When the temperature begins to rise but has not yet affected the L1 thermally sensitive task, the system enters the pre-conditioning phase. At this time, the temperature begins to rise but has not yet affected the L1 thermally sensitive task. The system adopts a time-power coupling strategy: for the L2 thermally elastic task, a mild dynamic voltage-frequency adjustment (DVFS) is initiated, reducing the operating frequency to half of the nominal frequency. This can reduce dynamic power consumption by a factor of 100 (e.g., 80%). Simultaneously, background calculations for L3 heat tolerance tasks are suspended. This phase aims to suppress temperature rise through time-based "peak shaving" without interfering with core business operations.

[0093] when Within the second threshold range ( When the system enters the active migration phase, local hotspots have formed, potentially impacting hardware lifespan. The system employs a space-power coupling strategy: forcibly migrating L3 tasks from high-temperature cores to low-temperature cores (requiring the target core temperature to be within a certain range). Simultaneously, L2 tasks are combined with migration and frequency adjustment—if the target core load is relatively light, some L2 tasks are migrated there; if the target core is equally busy, L2 tasks are deeply down-clocked or delayed. This stage utilizes the "valley filling" effect in the spatial dimension to balance the chip's heat distribution and avoid localized overheating.

[0094] when Exceeding the third threshold At this point, the system enters the emergency protection phase. The temperature is approaching the edge of thermal runaway, and the system performs a hard shutdown: immediately suspending or terminating the L3 task, forcing the L2 task to reduce its frequency to the lowest level or even suspending non-critical calculations, while maintaining the L1 task running at full speed (if hardware allows) or triggering hardware-level clock gating protection. This phase prioritizes ensuring the survival of the L1 task, ensuring the system does not crash due to overheating.

[0095] After the above-mentioned hierarchical control is implemented, the system further introduces a feedback compensation mechanism: monitoring the rate of temperature change. If the temperature drop slope is insufficient and L3 resources are exhausted, the priority of L2 tasks will be increased in reverse, allowing them to occupy more cryogenic resources, thus forming a closed-loop regulation.

[0096] Based on the above three-dimensional collaborative control, this invention achieves the following technical effects: First, by utilizing the portability of L3 tasks as thermal capacity, the elasticity of L2 tasks as a buffer, and protecting the rigidity of L1 tasks as a bottom line, differentiated thermal management based on task hierarchy is realized, avoiding the problem of critical tasks being stuck due to traditional "one-size-fits-all" frequency reduction; Second, through conditional judgment ( and The solution prioritizes either spatial migration or time-based frequency modulation, decoupling and recoupling control actions across different dimensions. This avoids bus congestion caused by simple migration and performance loss caused by simple frequency modulation. Third, the hierarchical triggering mechanism (pre-adjustment, active migration, and emergency protection) effectively prevents frequent system jumps near the threshold, ensuring the stability of thermal control. Fourth, experimental data shows that this solution can guarantee 100% L1 task latency compliance in high-temperature scenarios, while traditional solutions typically drop to 70%-80%. At the same time, by balancing hotspots through spatial migration, it significantly extends hardware lifespan.

[0097] In a specific implementation, the online learning strategy library of this invention pre-sets feature templates for various typical operating conditions. Each template includes parameters in four dimensions: heat source distribution characteristics, load fluctuation frequency, thermal time constant, and recommended scheduling strategy. By collecting system operation data in real time and matching it with the templates, the system can quickly identify the current operating condition and invoke the corresponding optimal thermal management strategy, thereby avoiding the problem of untimely response caused by the lag in traditional PID control.

[0098] Specifically, when the system is in a steady-state, high-load cruise mode with continuous high computing power output (such as continuous satellite Earth observation or high-concurrency server processing), the load rate remains above 85%, the temperature shows a monotonically increasing trend and eventually tends to thermal equilibrium, and the thermal time constant is relatively large. For this mode, the system prioritizes a spatial dimension strategy, focusing on utilizing the task migration of L3 heat-tolerant tasks to evenly distribute heat from the hot core to the entire chip surface, thereby avoiding localized overheating.

[0099] When the system encounters a transient peak impact condition (such as radar pulse emission or instantaneous startup of the encryption module), the load rate jumps from 10% to 100% in a very short time (usually less than 5 seconds), with a huge temperature gradient and an extremely fast local temperature rise rate, and a very small thermal time constant. At this time, the system prioritizes a time-dimensional strategy, reserving thermal capacity in advance, forcibly pausing L2 and L3 tasks before the thermal impact arrives, and using dynamic voltage and frequency adjustment to quickly reduce background power consumption, leaving sufficient thermal margin for the peak impact.

[0100] When the system is in a periodic pulsating condition (such as the periodic movement of the robotic arm or the timed data transmission), the load rate oscillates in a square wave pattern between 20% and 90%, with a fixed period (e.g., 60 seconds / cycle). The temperature fluctuates in a sawtooth pattern and cannot reach thermal equilibrium. The system adopts a phase staggered peak strategy to adjust the execution phase of the L2 thermoelastic task, so that it is executed during the trough of the system load, avoiding the generation of higher temperature peaks due to load superposition.

[0101] During cold start conditions, the system wakes up from hibernation or shutdown. The ambient temperature is low (e.g., -20°C), but the system's own heat generation increases sharply. At startup, all cores are fully loaded, resulting in a large temperature difference between the inside and outside of the chip and a high risk of thermal stress. The system implements a slope limiting strategy to restrict the startup speed of tasks other than L1 thermally sensitive tasks, and uses a stepped wake-up of L2 and L3 tasks to prevent thermal shock from causing material fatigue.

[0102] When the system encounters limited heat dissipation or fan failure conditions (such as fan speed lock, dust accumulation on the heatsink, or aging of thermal grease), the total thermal resistance of the system... The temperature rises significantly, with the steady-state temperature under the same load being 10-15°C higher than normal. The system adopts a conservative threshold reduction strategy, dynamically lowering the intermediate threshold. and high temperature threshold The trigger value is used to intervene in the frequency adjustment of L2 tasks in advance, sacrificing some performance for system security.

[0103] For non-periodic random fluctuations (such as sudden network traffic spikes or random user operations), the load rate is randomly distributed without a fixed period, and the temperature curve is chaotic and difficult to predict. In this case, the system no longer relies on fixed rules, but instead uses the online learning module to make dynamic decisions based on real-time feedback—choosing whether to migrate L3 tasks or reduce the frequency of L2 tasks, thereby adapting to unpredictable load changes.

[0104] When the system experiences concentrated multi-core hotspots (e.g., a specific algorithm causes a few cores to be under high load for an extended period while other cores remain idle), the spatial temperature distribution becomes extremely uneven, resulting in a significant temperature gradient. When the temperature exceeds 20°C, persistent hot spots form in certain areas. The system executes a forced load balancing strategy, breaking task affinity constraints and forcibly filling L3 tasks, and even some L2 tasks, into idle low-temperature cores, using these idle cores as auxiliary heat sinks to reduce the maximum temperature gradient.

[0105] In extreme high-temperature environments (such as summer surface operations or poorly ventilated equipment compartments), when the ambient temperature approaches or exceeds the chip junction temperature warning line (e.g., 85°C), the heat dissipation capacity approaches zero, relying mainly on thermal capacity buffering. The system initiates survival mode, maintaining only the L1 heat-sensitive tasks, suspending all L3 tasks and deeply downclocking L2 tasks, and suspending all unnecessary background activities.

[0106] When the system is under mixed critical task concurrency conditions, L1 tasks and a large number of L3 tasks arrive simultaneously, causing the task waiting queue to overflow. This resource contention leads to frequent context switching and generates additional switching power consumption and heat. The system adopts an L1 isolation strategy, using core isolation technology on the physical cores to run only L1 tasks, relegating all L2 and L3 tasks to the remaining cores, and implementing strict power capping.

[0107] For aging or performance degradation conditions (increased leakage rate of components and higher power consumption at the same frequency after long-term operation), the system's energy efficiency ratio decreases, and the temperature remains high even at low frequencies. The system implements a frequency upper limit locking strategy, no longer pursuing high-frequency performance, and limits the system's highest frequency to the optimal energy efficiency range to avoid unnecessary heat generation caused by "high frequency and low efficiency".

[0108] Through the feature-template-strategy mapping mechanism described above, the system can quickly identify complex thermal environments and invoke the optimal solution. For example, when the system detects a temperature gradient... When the load exceeds a preset value and is mainly concentrated on a specific core, the system matches the multi-core hotspot scenario, and the policy library outputs a forced load balancing instruction to forcibly migrate L3 tasks from the low-temperature zone to the high-temperature zone to assist in heat dissipation. When the system detects that the ambient temperature is close to the junction temperature warning line, the system matches the extreme high-temperature environment scenario, and the policy library outputs a survival mode instruction to directly suspend all L3 tasks and lock the L2 task frequency at the lowest level. This mechanism significantly improves the system's adaptability to complex thermal environments.

[0109] In some embodiments, the step "determining the task scheduling strategy corresponding to each pending task" includes:

[0110] The strategy determination model is used to determine the corresponding task scheduling strategy for each task to be processed.

[0111] The method proposed in this application also includes:

[0112] During the scheduling of the tasks to be processed, task scheduling feedback information is collected;

[0113] Based on the task scheduling feedback information, the strategy determination model is adjusted to obtain the adjusted strategy determination model;

[0114] The adjusted strategy is used to determine the updated task scheduling strategy for each pending task.

[0115] The strategy determination model can be an artificial intelligence model.

[0116] Artificial intelligence (AI) refers to the theories, methods, technologies, and application systems that utilize digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to achieve optimal results. In other words, AI is a comprehensive technology within computer science that attempts to understand the essence of intelligence and produce new intelligent machines that can react in a way similar to human intelligence. AI studies the design principles and implementation methods of various intelligent machines, enabling them to possess perception, reasoning, and decision-making capabilities.

[0117] Artificial intelligence (AI) is a comprehensive discipline encompassing a wide range of fields, including both hardware and software technologies. Fundamental AI technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interactive systems, and mechatronics. AI software technologies primarily include computer vision, speech processing, natural language processing, and machine learning / deep learning.

[0118] For example, the policy determination model proposed in this application embodiment can be a model related to machine learning or deep learning. For example, the policy determination model can be any one of Convolutional Neural Networks (CNN), De-Convolutional Networks (DN), Deep Neural Networks (DNN), Deep Convolutional Inverse Graphics Networks (DCIGN), Region-based Convolutional Networks (RCNN), Faster Region-based Convolutional Networks (Faster RCNN), and Bidirectional Encoder Representations from Transformers (BERT) models, etc.

[0119] In some embodiments, the task feedback information includes temperature feedback information, temperature gradient information, task execution frequency information, and task execution quality information.

[0120] The step "Calculate thermal risk information and performance loss information based on the task scheduling feedback information" includes:

[0121] The calculation of thermal risk information and performance loss information based on the task scheduling feedback information includes:

[0122] The thermal risk information is calculated based on the temperature feedback information and the temperature gradient information;

[0123] The task execution quality information is calculated based on the task execution frequency information and the task execution quality information.

[0124] In some embodiments, task execution quality information may refer to the task's delay time or a suspended flag, and temperature feedback information may include a reference temperature, etc.

[0125] In some embodiments, the mathematical expression for the objective function F(x) can be defined as follows:

[0126] min = · +(1- )·

[0127] Among them, constraints , , , . This represents the "quality" of the current scheduling strategy. The smaller the value, the better the strategy. Range of values Used to adjust the emphasis on "temperature" and "performance". When the temperature approaches a dangerous value, Automatically increase (e.g., from 0.5 to 0.8) to force the system to prioritize cooling. This is a thermal risk subfunction that quantifies the threat posed by the current temperature distribution to system safety. It is typically calculated based on temperature over-limit penalties and hotspot gradients. The performance loss subfunction quantifies the system performance cost sacrificed for heat de-heating (such as task latency and throughput reduction due to frequency reduction). It is a decision variable vector, which includes specific control actions such as whether to migrate the L2 / L3 task (0 / 1 variables) and the DVFS frequency level.

[0128] in, , : indicates that only temperatures above the reference temperature are calculated. The portion thereof is used as a penalty. : Indicates the maximum temperature gradient on the chip surface. : Normalization coefficient, ensuring that the two terms are on the same order of magnitude.

[0129] in, , and It is a normalization coefficient, ensuring that the two terms are on the same order of magnitude. L3 mission The delay time or the suspended flag. L2 mission core The current operating frequency. If the system suspends a large number of background tasks (L3) or reduces the frequency of computing cores (L2) to cool down, A larger value indicates a significant performance loss.

[0130] In some embodiments, the temperature of each core can be collected in real time. Task queue status, current frequency Based on the current operating conditions (such as the previously mentioned "transient shock" or "steady-state cruise"), the initial weights are retrieved from the strategy library. Sum of coefficients Substitute the current state into the formula. Due to decision variables If the task is discrete (e.g., task transfer only has "yes / no"), the optimal solution can be found using exhaustive search (for a small number of cores) or genetic algorithms (for large-scale systems). The optimal decision vector obtained by solving This translates into specific control commands (e.g., "Move task A to core 3", "Reduce core 1 frequency to 800MHz").

[0131] 104. Based on the task scheduling strategy, schedule and process each task to be processed.

[0132] For example, the task scheduling policy is converted into specific control instructions (e.g., "move task A to core 3", "reduce the frequency of core 1 to 800MHz"), and then each task to be processed is scheduled based on the control instructions.

[0133] This application proposes a multi-task-based thermal management method, which includes: real-time analysis of the spacecraft's pending task flow to obtain task analysis results, wherein the pending task flow includes at least one pending task; based on the task analysis results, classifying the at least one pending task to obtain multiple task types, wherein the multiple task types include heat-sensitive tasks, thermoelastic tasks, and heat-tolerant tasks; monitoring the sensor parameters of the spacecraft system, and when the sensor parameters are detected to be inconsistent with the sensor thresholds, determining the corresponding task scheduling strategy for each pending task based on the sensor parameters and the task type of each pending task, wherein the sensors include temperature parameters and / or power parameters; and scheduling each pending task based on the task scheduling strategy. This solution, through graded triggering (pre-adjustment, active migration, emergency protection), prevents the system from frequently fluctuating near thresholds, ensuring stable operation. Under the same heat dissipation conditions, because the L1 task is always protected, the availability of the system's core functions is significantly higher than traditional solutions. Furthermore, spatial migration balances hotspots, reducing physical aging of chips caused by uneven thermal stress. The operating condition feature template library contains at least 10 pre-set typical operating condition templates to match the current operating state of the system. Through the feature-template-policy mapping mechanism, the system can quickly identify complex thermal environments and call the optimal solution, avoiding the lag of traditional PID control.

[0134] To better implement the multi-task-based thermal management method provided in this application, one embodiment also provides a multi-task-based thermal management device, which can be integrated into an electronic device. The meanings of the terms used are the same as in the multi-task-based thermal management method described above, and specific implementation details can be found in the description of the method embodiments.

[0135] In one embodiment, a multi-tasking-based thermal management device is provided, which can be specifically integrated into an electronic device, such as... Figure 5 As shown, the multi-task-based thermal management device includes: a parsing unit 301, a partitioning unit 302, a determination unit 303, and a construction unit 304, as detailed below:

[0136] The parsing unit 301 is used to perform real-time parsing of the spacecraft's pending task flow to obtain the task parsing result, wherein the pending task flow includes at least one pending task.

[0137] The partitioning unit 302 is used to partition the at least one task to be processed based on the task parsing result to obtain multiple task types, wherein the multiple task types include heat-sensitive tasks, heat-elastic tasks and heat-tolerant tasks.

[0138] The determination unit 303 is used to monitor the sensor parameters of the spacecraft system. When the sensor parameters are found to be inconsistent with the sensor threshold, the unit determines the corresponding task scheduling strategy for each task to be processed based on the sensor parameters and the task type of each task to be processed. The sensors include temperature parameters and / or power parameters.

[0139] The scheduling unit 304 is used to schedule each task to be processed based on the task scheduling strategy.

[0140] In some embodiments, the partitioning unit 302 includes:

[0141] The acquisition subunit is used to acquire, based on the task parsing result, the task time information, task hardware association information, task business logic information and task priority corresponding to each task to be processed.

[0142] The sub-unit is used to divide the at least one task to be processed according to at least one of the task time information, task hardware association information, task corresponding business logic information and task priority, so as to obtain multiple task types.

[0143] In some embodiments, the determining unit 303 includes:

[0144] A comparison subunit is used to compare the temperature parameter with at least one preset threshold range;

[0145] A determination subunit is used to determine the corresponding task scheduling strategy for each task to be processed based on the comparison results of the temperature parameter and the at least one threshold interval and the task type of the task to be processed.

[0146] In some embodiments, the determining subunit includes:

[0147] The first execution module is used to perform time-power coupling control when the temperature parameter is within a first threshold range. The time-power coupling control includes performing dynamic voltage frequency adjustment on the thermal elastic task, reducing the operating frequency of the thermal elastic task, and pausing the background calculation of the thermal tolerance task.

[0148] The second execution module is used to perform space power coupling control when the temperature parameter is in the second threshold range. The space power coupling control includes migrating the thermally tolerant task from the high-temperature core to the low-temperature core and down-clocking or delaying the thermal elastic task.

[0149] The third execution module is used to perform emergency protection control when the temperature parameter exceeds the third threshold. The emergency protection control includes suspending or terminating the heat-tolerant task, reducing the frequency of the thermal elastic task to the lowest level, keeping the heat-sensitive task running at full speed, or triggering hardware-level clock gating protection.

[0150] In some embodiments, the thermal management device proposed in this application further includes:

[0151] The acquisition unit is used to acquire task scheduling feedback information during the scheduling process of the task to be processed;

[0152] The adjustment unit is used to adjust the strategy determination model according to the task scheduling feedback information to obtain the adjusted strategy determination model.

[0153] An execution unit is used to update the task scheduling strategy corresponding to each pending task by using the adjusted strategy to determine the model.

[0154] In some embodiments, the adjustment unit includes:

[0155] The calculation subunit is used to calculate thermal risk information and performance loss information based on the task scheduling feedback information;

[0156] A sub-unit is constructed to construct target loss information based on the thermal risk information and the performance loss information;

[0157] The adjustment subunit is used to adjust the policy determination model based on the target loss information to obtain the adjusted policy determination model.

[0158] In some embodiments, the computing subunit includes:

[0159] The first calculation subunit is used to calculate the thermal risk information based on the temperature feedback information and the temperature gradient information;

[0160] The second calculation subunit is used to calculate the task execution quality information based on the task running frequency information and the task execution quality information.

[0161] In practice, each of the above units can be implemented as an independent entity or can be arbitrarily combined to be implemented as the same or several entities. For the specific implementation of each of the above units, please refer to the previous method embodiments, which will not be repeated here.

[0162] The aforementioned multi-task-based thermal management device can suppress thermal shock to spacecraft.

[0163] This application also provides an electronic device, which may include a terminal or a server. For example, the electronic device may serve as a multi-tasking thermal management terminal, such as a mobile phone, tablet computer, etc.; or it may serve as a server, such as a multi-tasking thermal management server. Figure 6 As shown, it illustrates a structural schematic diagram of the terminal involved in an embodiment of this application. Specifically:

[0164] The electronic device may include components such as a processor 401 with one or more processing cores, a memory 402 with one or more computer-readable storage media, a power supply 403, and an input unit 404. Those skilled in the art will understand that... Figure 6 The electronic device structure shown does not constitute a limitation on the electronic device and may include more or fewer components than shown, or combine certain components, or have different component arrangements. Wherein:

[0165] The processor 401 is the control center of the electronic device. It connects various parts of the electronic device via various interfaces and lines. By running or executing software programs and / or modules stored in the memory 402, and by calling data stored in the memory 402, it performs various functions and processes data, thereby providing overall monitoring of the electronic device. Optionally, the processor 401 may include one or more processing cores; preferably, the processor 401 may integrate an application processor and a modem processor, wherein the application processor mainly handles the operating system, user page, and applications, and the modem processor mainly handles wireless communication. It is understood that the modem processor may not be integrated into the processor 401.

[0166] The memory 402 can be used to store software programs and modules. The processor 401 executes various functional applications and data processing by running the software programs and modules stored in the memory 402. The memory 402 may mainly include a program storage area and a data storage area. The program storage area may store the operating system, application programs required for at least one function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the electronic device, etc. In addition, the memory 402 may include high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other volatile solid-state storage device. Accordingly, the memory 402 may also include a memory controller to provide the processor 401 with access to the memory 402.

[0167] The electronic device also includes a power supply 403 that supplies power to the various components. Preferably, the power supply 403 can be logically connected to the processor 401 through a power management system, thereby enabling functions such as charging, discharging, and power consumption management through the power management system. The power supply 403 may also include one or more DC or AC power supplies, recharging systems, power fault detection circuits, power converters or inverters, power status indicators, and other arbitrary components.

[0168] The electronic device may also include an input unit 404, which can be used to receive input digital or character information, and generate keyboard, mouse, joystick, optical or trackball signal inputs related to user settings and function control.

[0169] Although not shown, the electronic device may also include a display unit, etc., which will not be described in detail here. Specifically, in this embodiment, the processor 401 in the electronic device loads the executable files corresponding to the processes of one or more applications into the memory 402 according to the following instructions, and the processor 401 runs the applications stored in the memory 402 to realize various functions, as follows:

[0170] The task flow to be processed of the spacecraft is analyzed in real time to obtain the task analysis result, wherein the task flow to be processed includes at least one task to be processed;

[0171] Based on the task parsing results, the at least one task to be processed is divided into multiple task types, including heat-sensitive tasks, thermally elastic tasks, and heat-tolerant tasks.

[0172] The system monitors the sensor parameters of the spacecraft system. When the sensor parameters are found to be inconsistent with the sensor threshold, the system determines the corresponding task scheduling strategy for each task based on the sensor parameters and the task type of each task to be processed. The sensors include temperature parameters and / or power parameters.

[0173] Based on the task scheduling strategy, each task to be processed is scheduled.

[0174] For details on the implementation of each of the above operations, please refer to the previous examples, which will not be repeated here.

[0175] According to one aspect of this application, a computer program product or computer program is provided, comprising computer instructions stored in a computer-readable storage medium. A processor of an electronic device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the electronic device to perform the methods provided in the various optional implementations of the above embodiments.

[0176] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be performed by a computer program, or by a computer program controlling related hardware. The computer program can be stored in a computer-readable storage medium and loaded and executed by a processor.

[0177] Therefore, embodiments of this application also provide a storage medium storing a computer program that can be loaded by a processor to execute the steps of any of the multi-task-based thermal management methods provided in embodiments of this application. For example, the computer program can execute the following steps:

[0178] The task flow to be processed of the spacecraft is analyzed in real time to obtain the task analysis result, wherein the task flow to be processed includes at least one task to be processed;

[0179] Based on the task parsing results, the at least one task to be processed is divided into multiple task types, including heat-sensitive tasks, thermally elastic tasks, and heat-tolerant tasks.

[0180] The system monitors the sensor parameters of the spacecraft system. When the sensor parameters are found to be inconsistent with the sensor threshold, the system determines the corresponding task scheduling strategy for each task based on the sensor parameters and the task type of each task to be processed. The sensors include temperature parameters and / or power parameters.

[0181] Based on the task scheduling strategy, each task to be processed is scheduled.

[0182] For details on the implementation of each of the above operations, please refer to the previous examples, which will not be repeated here.

[0183] Since the computer program stored in the storage medium can execute the steps of any of the multi-task-based thermal management methods provided in the embodiments of this application, the beneficial effects that any of the multi-task-based thermal management methods provided in the embodiments of this application can achieve can be realized, as detailed in the preceding embodiments, and will not be repeated here.

[0184] The foregoing has provided a detailed description of a multi-task-based thermal management method, apparatus, electronic device, and storage medium provided in the embodiments of this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the embodiments above are only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A multi-task based thermal management method, characterized by, include: The task flow to be processed of the spacecraft is analyzed in real time to obtain the task analysis result, wherein the task flow to be processed includes at least one task to be processed; Based on the task parsing results, the at least one task to be processed is divided into multiple task types, including heat-sensitive tasks, thermally elastic tasks, and heat-tolerant tasks. The system monitors the sensor parameters of the spacecraft system. When the sensor parameters are found to be inconsistent with the sensor threshold, the system determines the corresponding task scheduling strategy for each task based on the sensor parameters and the task type of each task to be processed. The sensors include temperature parameters and / or power parameters. Based on the task scheduling strategy, each task to be processed is scheduled.

2. The method of claim 1, wherein, Based on the task parsing results, the at least one task to be processed is divided into multiple task types, including: Based on the task parsing results, obtain the task time information, task hardware association information, task business logic information, and task priority for each task to be processed. Based on at least one of the task time information, task hardware association information, task corresponding business logic information, and task priority, the at least one task to be processed is divided into multiple task types.

3. The method of claim 1, wherein, When the sensor parameters include temperature parameters, the step of determining the corresponding task scheduling strategy for each task to be processed based on the sensor parameters and the task type of each task to be processed includes: The temperature parameter is compared with at least one preset threshold range; Based on the comparison results of the temperature parameter and the at least one threshold range, as well as the task type of the task to be processed, a corresponding task scheduling strategy is determined for each task to be processed.

4. The method of claim 3, wherein, The step of determining the corresponding task scheduling strategy for each task based on the comparison results of the temperature parameter and the at least one threshold interval, and the task type of the task to be processed, includes: When the temperature parameter is within the first threshold range, time-power coupling control is executed, wherein the time-power coupling control includes performing dynamic voltage frequency adjustment on the thermal elastic task, reducing the operating frequency of the thermal elastic task, and pausing the background calculation of the thermal tolerance task. When the temperature parameter is in the second threshold range, space power coupling control is executed, wherein the space power coupling control includes migrating the thermally tolerant task from the high-temperature core to the low-temperature core, and down-clocking or delaying the scheduling of the thermally elastic task. When the temperature parameter exceeds the third threshold, emergency protection control is executed. The emergency protection control includes suspending or terminating the heat-tolerant task, reducing the frequency of the thermal elastic task to the lowest level, keeping the heat-sensitive task running at full speed, or triggering hardware-level clock gating protection.

5. The method according to any one of claims 1 to 4, characterized in that, The step of determining the corresponding task scheduling strategy for each pending task includes: The strategy determination model is used to determine the corresponding task scheduling strategy for each task to be processed. The method further includes: During the scheduling of the tasks to be processed, task scheduling feedback information is collected; Based on the task scheduling feedback information, the strategy determination model is adjusted to obtain the adjusted strategy determination model; The adjusted strategy is used to determine the updated task scheduling strategy for each pending task.

6. The method of claim 5, wherein, The step of adjusting the strategy determination model based on the task scheduling feedback information to obtain the adjusted strategy determination model includes: Calculate thermal risk information and performance loss information based on the task scheduling feedback information; Based on the thermal risk information and the performance loss information, target loss information is constructed; The strategy determination model is adjusted based on the target loss information to obtain the adjusted strategy determination model.

7. The method according to claim 6, characterized in that, The task feedback information includes temperature feedback information, temperature gradient information, task execution frequency information, and task execution quality information; The calculation of thermal risk information and performance loss information based on the task scheduling feedback information includes: The thermal risk information is calculated based on the temperature feedback information and the temperature gradient information; The task execution quality information is calculated based on the task execution frequency information and the task execution quality information.

8. A thermal management device, characterized in that, include: The parsing unit is used to perform real-time parsing of the spacecraft's pending task flow to obtain the task parsing result, wherein the pending task flow includes at least one pending task. A partitioning unit is used to partition the at least one task to be processed based on the task parsing result to obtain multiple task types, wherein the multiple task types include heat-sensitive tasks, heat-elastic tasks, and heat-tolerant tasks. A determination unit is used to monitor sensor parameters of the spacecraft system. When the sensor parameters are found to be inconsistent with the sensor threshold, the unit determines the corresponding task scheduling strategy for each task to be processed based on the sensor parameters and the task type of each task to be processed. The sensors include temperature parameters and / or power parameters. The scheduling unit is used to schedule each task to be processed based on the task scheduling strategy.

9. An electronic device, comprising: It includes a memory and a processor; the memory stores an application program, and the processor runs the application program within the memory to perform the operations in the multitasking-based thermal management method according to any one of claims 1 to 7.

10. A readable storage medium, characterized in that, The computer-readable storage medium stores a plurality of instructions adapted for loading by a processor to perform the steps of the multitasking-based thermal management method according to any one of claims 1 to 7.

Images