Industrial computer memory and storage collaborative cooling system

Abstract

The invention relates to the technical field of industrial computer heat dissipation, in particular to an industrial computer memory and storage collaborative heat dissipation system which is used for solving the problems that in the prior art, temperature distribution modeling and state evaluation cannot be achieved, thermal state categories cannot be divided by integrating a temperature threshold value and an operation state, and the equipment health judgment accuracy is reduced. According to the invention, a temperature control monitoring module constructs a thermal field model by using operation state and thermal behavior data, divides hardware units, establishes a heat conduction path, realizes temperature distribution modeling and state evaluation, calculates thermal power density through standardized parameters, and solves steady-state temperature by using a numerical method in combination with boundary conditions and a cooling model. Structured temperature information is output, a temperature threshold value and an operation state are integrated to divide thermal state categories, and the equipment health judgment accuracy is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial computer heat dissipation, and more particularly to an industrial computer memory and storage collaborative heat dissipation system. Background Art

[0002] As the performance of industrial computers improves, the power consumption of memory and storage devices has increased significantly. Traditional independent cooling solutions have difficulty achieving efficient heat dissipation due to uneven heat source distribution and lack of intelligent control mechanisms. Not only do they consume a lot of energy, they are also prone to local overheating and hardware risks, seriously restricting the stability and energy efficiency of the system.

[0003] Patent application CN111984098A discloses a computer system heat dissipation method, applied to the BMC in a target computer system. This method uses a busbar pre-installed on a PCBA to detect the system's temperature and adjusts the fan speed based on the acquired temperature information, thereby achieving heat dissipation control for the computer system. Compared to existing technologies, this method obtains temperature values ​​by detecting the current flowing through the PCBA, effectively avoiding the influence of the PCBA board temperature on the measurement results. It can more accurately reflect the actual system temperature, thereby achieving more precise heat dissipation control. However, the above-mentioned reference patent achieves precise heat dissipation by setting a busbar detection system temperature on the PCBA and dynamically adjusting the fan speed by the BMC, effectively avoiding the temperature interference of the PCBA itself and improving the detection accuracy and heat dissipation efficiency. However, it cannot achieve temperature distribution modeling and status evaluation, cannot classify thermal status categories based on temperature thresholds and operating status, reduces the accuracy of equipment health judgment, cannot predict and mark potential high-temperature units, cannot dynamically adjust heat distribution through multiple mechanisms, cannot iteratively optimize and generate non-inferior heat dissipation strategies under temperature, power consumption and performance constraints, cannot combine lifecycle management and model feedback mechanism to achieve dynamic updates, and cannot form a complete heat dissipation control closed loop.

[0004] To this end, we propose an industrial computer memory and storage collaborative cooling system to address the above problems. Summary of the Invention

[0005] The purpose of the present invention is to provide an industrial computer memory and storage collaborative heat dissipation system, which solves the problems that the existing technology cannot realize temperature distribution modeling and status evaluation, cannot comprehensively divide thermal status categories according to temperature thresholds and operating status, reduces the accuracy of equipment health judgment, cannot predict and mark potential high-temperature units, cannot dynamically adjust heat distribution through multiple mechanisms, cannot iteratively optimize and generate non-inferior heat dissipation strategies under temperature, power consumption and performance constraints, cannot combine lifecycle management and model feedback mechanism to achieve dynamic updates, and cannot form a complete heat dissipation control closed loop.

[0006] The purpose of the present invention is achieved through the following technical solutions: An industrial computer memory and storage collaborative cooling system, applied to an industrial automation management platform, includes: The data acquisition module is used to collect the operating status and thermal behavior data of the memory and storage unit in real time, and classify and pre-process the collected operating status and thermal behavior data; The temperature control monitoring module builds a thermal field model of the memory and storage units based on pre-processed operating status data and thermal behavior data, and outputs the temperature distribution of the heat accumulation area and the corresponding thermal status identification information; The hot zone prediction and intervention module builds a high-heat zone prediction model based on operating status data and temperature trend information. It combines task scheduling strategies, data migration mechanisms, and memory access path optimization to dynamically adjust heat source distribution. The adaptive strategy generation module, based on a multi-objective evolutionary algorithm, combines strategy lifecycle management with a model collaborative iteration mechanism to generate and update cooling control strategies that adapt to different hardware configurations.

[0007] As a preferred embodiment of the present invention, the process of the temperature control monitoring module constructing the thermal field model of the memory and storage unit includes: Obtain pre-processed operational status data and thermal behavior data. The operational status data includes IOPS, throughput, ECC single-bit error count, and total number of bad blocks. The thermal behavior data includes core temperature, ambient temperature, and fan speed. Constructing the hardware space topology: 3D modeling of the memory modules and storage devices in industrial computers according to their actual physical layout in the motherboard or chassis; Each device is further divided into multiple hardware units and used as the basic nodes of the thermal field model; Establish a thermal network structure based on hardware units to form a complete heat conduction path diagram within the system; Establish heat source mapping relationship: Based on the collected operating status data, a thermal power density model is established for each hardware unit: ; where Q i is the thermal power per unit time of the i-th hardware unit, All are empirical calibration coefficients. IOPS, throughput, ECC, and bad block counts are pre-processed and standardized operating status data. Set up thermal boundary conditions and cooling the model: The heat conduction boundary conditions are set according to the ambient temperature and fan speed, and the heat convection coefficient h and wind speed function v (RPM) are introduced to establish the cooling capacity model: , where q sr is the heat flux per unit area, A is the heat conduction surface area, T xp is the chip surface temperature, T hj is the ambient temperature; Solve for the steady-state distribution of the thermal field: Substitute the heat source distribution and cooling boundary conditions into the heat diffusion governing equation: , where k is the thermal conductivity of the material, T is the temperature field, and Q is the power density of the heat source; The numerical method is used to solve the heat diffusion control equation to obtain the steady-state temperature distribution T of each hardware unit. i .

[0008] As a preferred embodiment of the present invention, the process of the temperature control monitoring module outputting the temperature distribution of the heat accumulation area and the corresponding thermal state identification information includes: Temperature distribution output in heat accumulation areas: The steady-state temperature distribution obtained by numerically solving the heat diffusion equation is used to construct the temperature field map inside the system; Based on the hardware unit division and hardware physical layout, the system is divided into multiple heat accumulation areas; Thermal status identification information generation: The system classifies the status of each heat accumulation area according to the preset temperature threshold and generates the corresponding thermal status identification; Thermal status indicators include the following categories: Normal state: the area temperature is lower than the set temperature upper limit and the load is in a stable state; Critical state: The zone temperature is higher than 90% of the set temperature upper limit but does not exceed the upper limit, and there is load fluctuation; Overheating: The zone temperature exceeds the set upper temperature limit, or the increase in the ECC error count within a unit monitoring cycle exceeds the preset threshold; Failure warning state: The area temperature is continuously higher than the set temperature upper limit for a preset period of time, and the growth rate of the number of new bad blocks in a unit monitoring cycle exceeds the set threshold; State output method: The system organizes the temperature field calculation results and thermal state identification information into a structured data format and outputs them.

[0009] As a preferred embodiment of the present invention, the process of constructing a high-heat zone prediction model by the heat zone prediction intervention module includes: Obtain pre-processed operating status data and temperature trend information, where the temperature trend information is the steady-state temperature distribution data of each hardware unit under continuous load conditions; Based on the relationship between the pre-processed operating status data and the historical temperature data, a thermal power density model for each hardware unit is established. The specific steps are as follows: S1: Analyze the correlation between operating parameters and temperature changes; S2: Extract the thermal response characteristics of each hardware unit under the same load; S3: Use thermal power density function to represent the thermal power density of each hardware unit; S4: construct heat intensity distribution map; Based on the thermal power density and heat conduction characteristics, a prediction model for temperature changes in the future time period is established. The specific steps are as follows: T1: Analyze the time series characteristics of historical temperature data; T2: extraction temperature change rate and fluctuation range; T3: Calculate future temperature changes based on current thermal power density and thermal diffusion capacity; T4: Set the prediction time window to identify hardware units whose temperatures will exceed the threshold in the future; The system marks each hardware unit as a candidate for a hotspot based on the following four conditions: The current temperature reaches or exceeds the set threshold; The temperature change trend shows an increase at multiple consecutive time points; The thermal power density is higher than the overall system average; The heat source concentration is higher than the set threshold; The system outputs the following information: the high-temperature zone number, the temperature value of the corresponding hardware unit, and the status identification.

[0010] As a preferred embodiment of the present invention, the process of the hot zone prediction intervention module combining task scheduling strategy, data migration mechanism and memory access path optimization to perform dynamic adjustment of heat source distribution includes: To optimize the task scheduling strategy, the system performs the following operations based on the current temperature of each hardware unit and the task thermal contribution value output by the thermal trend prediction model: Assign new tasks to hardware units whose current temperature is lower than the set temperature threshold; According to the task heat contribution value, a target execution unit is selected from the hardware units whose temperature increment is expected to be less than a set increment threshold within a preset time period in the future; Migrate tasks running on hardware units with temperatures higher than a set temperature threshold to available hardware units with temperatures lower than the threshold and with the smallest thermal load increment. Perform periodic migration for tasks that run for longer than the set duration. Dynamically adjust CPU and I / O scheduling weights; The task scheduler receives the temperature prediction data and heat contribution value output from the hot zone prediction module; Data migration mechanism: The system performs the following operations based on the data read and write frequency and temperature status of each hardware unit: Real-time monitoring of data reading and writing frequency per unit time; When the number of accesses per unit time of a data block exceeds the set access frequency threshold, it is migrated from the current hardware unit to an adjacent hardware unit whose current temperature is lower than the set temperature threshold; Creating a data copy in a hardware unit whose current temperature is lower than a set temperature threshold; When the data writing frequency exceeds the set writing frequency threshold, the writing operation to the area is restricted; When the temperature of the hardware unit is continuously higher than the set temperature threshold for more than the set duration, the data in it will be migrated out; To optimize memory access paths, the system performs the following operations based on the access pressure and temperature status of each memory controller channel: Dynamically monitor the access request pressure and current temperature of each memory channel; Dynamically adjust the mapping path of access requests based on access pressure and temperature status; When the current temperature of the memory area is higher than the set temperature threshold, set access redirection rules; For tasks that meet the preset task attributes, the memory area with the lowest heat load on the current access path is selected for access; Dynamically adjust memory address mapping rules based on memory access patterns and current temperature status.

[0011] As a preferred embodiment of the present invention, the process of the adaptive strategy generation module using a multi-objective evolutionary algorithm to evolve includes: The system must meet the following three constraints when generating a cooling control strategy: Control the temperature value of the hardware unit to be within the set range; The total power consumption of the control system does not exceed the set upper limit; Maintaining mission execution performance not below the set lower limit; Application of Multi-Objective Evolutionary Algorithm: The multi-objective evolutionary algorithm simulates the biological evolution process to find a set of non-inferior solutions that meet multiple constraints in the solution space. For the heat dissipation control strategy, the following objective function is set: Temperature control target: Calculate the difference between the current hardware unit temperature value and the set safety temperature upper limit; Power consumption control target: Calculate the difference between the current system power consumption and the maximum allowable power consumption; Performance maintenance target: Calculate the difference between the minimum acceptable performance and the current task execution performance; The multi-objective evolutionary algorithm evolves according to the following steps: Z1: Select individuals with better performance from the current population; Z2: Combine the parameters of two or more individuals to generate a new candidate strategy; Z3: Perturb the parameters of some candidate strategies and introduce new solutions; Z4: Calculate the value of each candidate strategy on each objective function; Z5: Retain non-inferior solutions to form the population of the next generation.

[0012] As a preferred embodiment of the present invention, the process of the adaptive strategy generation module generating a heat dissipation control strategy by combining strategy lifecycle management and model collaborative iteration mechanism includes: Policy lifecycle management: The initial policy is generated by a multi-objective evolutionary algorithm, with parameters determined based on the current hardware configuration and operating environment; The system collects the following data for strategy evaluation: hardware unit temperature values, system power consumption, and task execution performance; When the existing strategy cannot meet the objective function, the strategy update process is initiated. The update methods include: adjusting the objective function weight, introducing new variables, and re-running the evolutionary algorithm to generate a new strategy; Model collaborative iteration mechanism: Use historical operating data to train the initial heat dissipation control model. The input includes hardware configuration information, ambient temperature, and task load characteristics. The output is a set of candidate strategies and their corresponding constraint satisfaction. During operation, the system continuously collects data and feeds it back to the model to adjust policy parameters. The feedback data includes: current temperature distribution, power consumption fluctuations, and performance index changes; When the operating environment changes significantly, model retraining is performed. The retraining process includes: collecting new operating data, constructing new training samples, retraining the model, and updating the parameter settings in the strategy generation module.

[0013] As a preferred embodiment of the present invention, the process of the adaptive strategy generation module generating and updating the heat dissipation control strategy adapted to different hardware configurations includes: Adaptation mechanism for different hardware configurations: The heat dissipation control strategy needs to be adapted to different hardware configurations. The adaptation process is as follows: Hardware feature identification: read hardware identifiers, temperature sensor numbers, and power consumption limit parameters; Strategy matching: Select the adaptation strategy from the strategy library based on hardware characteristics; Strategy adjustment: If there is no perfect matching strategy, a new strategy is generated based on the evolutionary algorithm; Policy deployment: Apply the selected or generated policy to the target device; The generated cooling control policy contains the following operations: Temperature regulation: adjust the fan speed or cooling device working status according to the current temperature value; Power consumption allocation: Control the power consumption of each hardware unit so that its total does not exceed the set upper limit; Task scheduling: Dynamically adjust task allocation based on hardware load to balance temperature and performance requirements; Hardware start and stop: Determine whether to shut down non-critical hardware modules to reduce heat based on the operating status.

[0014] Compared with the prior art, the advantages of the present invention are: In this invention, a thermal field model is constructed using operating status and thermal behavior data through a temperature control monitoring module, hardware units are divided and heat conduction paths are established, temperature distribution modeling and status assessment are achieved, thermal power density is calculated through standardized parameters, boundary conditions and cooling models are combined, and a numerical method is used to solve the steady-state temperature. Structured temperature information is output, and thermal status categories are divided according to the temperature threshold and operating status, thereby improving the accuracy of equipment health judgment. In this invention, the hot zone prediction and intervention module analyzes the hardware operating status and temperature trends, predicts and marks potential high-temperature units, and dynamically adjusts the heat distribution using three mechanisms: task scheduling, data migration, and memory access optimization. Tasks are prioritized in low-temperature areas, hot data is migrated to hardware units with lower temperatures, and memory access is dynamically redirected to avoid local overheating. In the present invention, the adaptive strategy generation module adopts a multi-objective evolutionary algorithm to iteratively optimize and generate a non-inferior heat dissipation strategy under the constraints of temperature, power consumption and performance. It combines lifecycle management and model feedback mechanism to achieve dynamic updates, supports multiple hardware configuration adaptation, covers temperature regulation, power consumption allocation, task scheduling and hardware start and stop, and forms a complete heat dissipation control closed loop. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 This is a system block diagram of Embodiment 1 of the present invention; Figure 2 A flow chart of the steps for establishing a thermal power density model for each hardware unit in the present invention; Figure 3 A flow chart of the steps of establishing a temperature change prediction model for a future time period in the present invention; Figure 4 This is a system block diagram of Example 2 of the present invention. DETAILED DESCRIPTION

[0016] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making creative work shall fall within the scope of protection of the present invention.

[0017] Example 1: Figure 1 、 Figure 2 and Figure 3 As shown, the present invention proposes an industrial computer memory and storage collaborative heat dissipation system, which is applied to an industrial automation management platform, including: The data acquisition module is used to collect the operating status and thermal behavior data of the memory and storage units in real time, classify and preprocess the collected operating status and thermal behavior data, perform standardization processing, outlier identification and correction, time series feature extraction and data compression on the operating status data, and perform sensor error correction, local area temperature interpolation, temperature range mapping, overheating event annotation and ambient temperature difference compensation on the thermal behavior data; The data acquisition module supports multi-source heterogeneous data collection, acquires the operating status and thermal behavior data of memory and storage units in real time, and performs classification and pre-processing to improve the effectiveness of subsequent analysis; performs standardization, outlier correction, time series feature extraction and compression on operating status data to improve consistency and transmission efficiency; performs error correction, temperature interpolation, range mapping, temperature difference compensation and overheating annotation on thermal behavior data to enhance data accuracy and early warning capabilities; all results are output in a structured format to facilitate system calls and integration, providing reliable data support for thermal management.

[0018] The temperature control monitoring module builds a thermal field model of the memory and storage units based on pre-processed operating status data and thermal behavior data, and outputs the temperature distribution of the heat accumulation area and the corresponding thermal status identification information; The process of the temperature control monitoring module building the thermal field model of the memory and storage unit includes: Obtain pre-processed operational status data and thermal behavior data. The operational status data includes IOPS, throughput, ECC single-bit error count, and total number of bad blocks. The thermal behavior data includes core temperature, ambient temperature, and fan speed. Constructing hardware space topology (thermal grid partitioning): 3D spatial modeling of memory modules and storage devices (SSD, HDD, NVMe) in industrial computers based on their actual physical layout in the motherboard or chassis; Each device is further divided into multiple hardware units (memory master chip, cache particles, DRAM particles), and used as the basic nodes of the thermal field model; Establish a thermal network structure based on hardware units to form a complete heat conduction path diagram within the system; Establish heat source mapping relationship (thermal power density calculation): Based on the collected operating status data, a thermal power density model is established for each hardware unit: ; where Q iis the thermal power per unit time of the i-th hardware unit, These are empirical calibration coefficients used to convert business load into thermal power. IOPS, throughput, ECC, and bad block counts are all pre-processed standardized operating status data. i is the number of input and output operations completed by the i-th hardware unit in unit time, throughput i is the total amount of data transmitted by the i-th hardware unit per unit time, ECC i is the number of error correction codes triggered by data errors in the i-th hardware unit per unit time, the number of bad blocks i is the number of storage blocks that are unavailable or marked as damaged in the i-th non-volatile storage hardware unit; Set up thermal boundary conditions and cooling the model: The heat conduction boundary conditions are set according to the ambient temperature and fan speed, and the heat convection coefficient h and wind speed function v (RPM) are introduced to establish the cooling capacity model: , where q sr is the heat flux per unit area, A is the heat conduction surface area, T xp is the chip surface temperature, T hj is the ambient temperature; The finite element method is used to model and solve the steady-state distribution of the thermal field, and the heat source distribution and cooling boundary conditions are substituted into the heat diffusion control equation: , where k is the thermal conductivity of the material, T is the temperature field, and Q is the power density of the heat source. The steady-state temperature distribution of each thermal unit is obtained by numerically solving the problem using the finite element discretization method; The numerical method is used to solve the heat diffusion control equation to obtain the steady-state temperature distribution T of each hardware unit. i The finite difference method discretizes the spatial domain into multiple grid points and calculates the temperature value of each grid point. The finite element method divides the spatial domain into small units and calculates the temperature of each unit. The process of the temperature control monitoring module outputting the temperature distribution of the heat accumulation area and the corresponding thermal status identification information includes: Temperature distribution output in heat-collecting areas: The steady-state temperature distribution obtained by numerically solving the heat diffusion equation is used to construct a temperature field map within the system. The temperature field map uses two-dimensional or three-dimensional form to display the temperature values ​​of each hardware unit; Temperature visualization is as follows: Heat map: color depth indicates temperature; Isotherm map: line density reflects the temperature change trend; Based on the hardware unit division and hardware physical layout, the system is divided into multiple heat aggregation areas. Each heat aggregation area contains several hardware units to present local temperature characteristics. Temperature values ​​and status information are marked in the temperature graph with colors or symbols, providing basic data support for subsequent monitoring and analysis; Thermal status identification information generation: The system classifies the status of each heat accumulation area according to the preset temperature threshold and generates the corresponding thermal status identification; The temperature threshold is set according to the hardware specifications, including upper and lower limits. If the temperature exceeds the threshold range during device operation, the corresponding status flag is triggered; Thermal status indicators include the following categories: Normal state: the area temperature is lower than the set temperature upper limit and the load is in a stable state; Critical state: The zone temperature is higher than 90% of the set temperature upper limit but does not exceed the upper limit, and there is load fluctuation; Overheating: The zone temperature exceeds the set upper temperature limit, or the increase in the ECC error count within a unit monitoring cycle exceeds the preset threshold; Failure warning state: The area temperature is continuously higher than the set temperature upper limit for a preset period of time, and the growth rate of the number of new bad blocks in a unit monitoring cycle exceeds the set threshold; The set temperature limit refers to the maximum allowable operating temperature set based on hardware specifications, heat dissipation capabilities, and operational stability requirements. The bad block growth rate refers to the number of new bad blocks added within a unit monitoring cycle and is used to assess the aging and reliability of storage media. The preset duration refers to a time threshold set by the system to determine whether the high temperature state has entered a long-term abnormal stage, such as 1 minute or 5 minutes. The ECC error count increment refers to the change in the number of data corrections required using error correction code within a unit monitoring cycle. Thermal status judgment not only relies on temperature data, but also combines operating status parameters to comprehensively evaluate the health of the equipment; State output mode: The system organizes the temperature field calculation results and thermal state identification information into a structured data format for output; The output includes: the temperature value of each hardware unit, the corresponding thermal status identification and fault warning level; Supported data output formats include JSON and XML, which facilitates integration with external monitoring systems, alarm systems or control platforms; The temperature control and monitoring module constructs a thermal field model based on operating status and thermal behavior data to realize temperature distribution modeling and thermal status assessment at the hardware unit level; divides the thermal grid through three-dimensional space modeling, and establishes a heat conduction path diagram in combination with the hardware layout to provide a structural basis for refined thermal analysis; uses standardized operating parameters to calculate the thermal power density of each unit, establishes a heat source mapping relationship, and solves the steady-state temperature distribution in combination with boundary conditions and cooling models. It supports finite difference method or finite element method for numerical calculation to adapt to scenarios of different complexity; the system outputs structured temperature distribution information, generates thermal maps, isotherm maps and other visualization forms, and intuitively displays local heat accumulation areas; comprehensively judges the thermal status based on preset temperature thresholds and operating status parameters, and divides it into four categories: normal, critical, overheating, and failure warning, to improve the accuracy of equipment health assessment.

[0019] The hot zone prediction and intervention module builds a high-heat zone prediction model based on operating status data and temperature trend information. It combines task scheduling strategies, data migration mechanisms, and memory access path optimization to dynamically adjust heat source distribution. The process of building a high-heat zone prediction model in the heat zone prediction and intervention module includes: Obtain pre-processed operating status data and temperature trend information, where the temperature trend information is the steady-state temperature distribution data of each hardware unit under continuous load conditions; Based on the relationship between the pre-processed operating status data and the historical temperature data, a thermal power density model for each hardware unit is established. The specific steps are as follows: S1: Analyze the correlation between operating parameters and temperature changes. Use statistical or machine learning methods to analyze the mapping relationship between operating parameters (such as task frequency, access bandwidth, and error rate) and temperature changes. S2: Extract the thermal response characteristics of each hardware unit under the same load, and record the temperature rise, stabilization time, fluctuation range and other characteristics of multiple hardware units under the same load conditions; S3: Use the thermal power density function to represent the thermal power density of each hardware unit. The heat generated per unit volume per unit time is used as the thermal power density and is calculated using an empirical formula or fitting method. S4: Build a heat intensity distribution map. Based on the heat power density distribution and hardware layout, generate a heat intensity distribution map reflecting the current heat flux density spatial distribution. Based on the thermal power density and heat conduction characteristics, a prediction model for temperature changes in the future time period is established. The specific steps are as follows: T1: Analyze the time series characteristics of historical temperature data to identify the trend, periodicity and suddenness of temperature changes; T2: Extract the temperature change rate and fluctuation range, record the speed of temperature increase or decrease, and the fluctuation range under different loads; T3: Calculate future temperature changes based on current thermal power density and thermal diffusion capacity. Combined with thermal power density, material thermal conductivity, and heat dissipation design parameters, estimate temperature changes within the next 5 to 10 minutes. T4: Set a prediction time window to identify hardware units whose temperatures will exceed the threshold in the future. Define a prediction time window (for example, the next 5 minutes) to determine which hardware units' temperatures will exceed the safety upper limit. The system marks each hardware unit as a candidate for a hotspot based on the following four conditions: The current temperature reaches or exceeds the set threshold; The temperature change trend shows an increase at multiple consecutive time points; The thermal power density is higher than the overall system average; The heat source concentration is higher than the set threshold; If a hardware unit meets any one or more of the conditions, it is marked as a high-heat zone candidate; The system outputs the following information: the high-temperature zone number, the temperature value of the corresponding hardware unit, and the status identification; The output data format supports structured representation and can be used for subsequent intervention module calls; The hot zone prediction and intervention module combines task scheduling strategies, data migration mechanisms, and memory access path optimization to dynamically adjust heat source distribution. The process includes: To optimize the task scheduling strategy, the system performs the following operations based on the current temperature of each hardware unit and the task thermal contribution value output by the thermal trend prediction model: Assign new tasks to hardware units whose current temperature is lower than the set temperature threshold; According to the task heat contribution value, a target execution unit is selected from the hardware units whose temperature increment is expected to be less than a set increment threshold within a preset time period in the future; Migrate tasks running on hardware units with temperatures higher than a set temperature threshold to available hardware units with temperatures lower than the threshold and with the smallest thermal load increment. Perform periodic migration for tasks that run for longer than the set duration. Dynamically adjust CPU and I / O scheduling weights; The task scheduler receives the temperature prediction data and heat contribution value output from the hot zone prediction module; Data migration mechanism: The system performs the following operations based on the data read and write frequency and temperature status of each hardware unit: Real-time monitoring of data reading and writing frequency per unit time; When the number of accesses per unit time of a data block exceeds the set access frequency threshold, it is migrated from the current hardware unit to an adjacent hardware unit whose current temperature is lower than the set temperature threshold; Creating a data copy in a hardware unit whose current temperature is lower than a set temperature threshold; When the data writing frequency exceeds the set writing frequency threshold, the writing operation to the area is restricted; When the temperature of the hardware unit is continuously higher than the set temperature threshold for more than the set duration, the data in it will be migrated out; To optimize memory access paths, the system performs the following operations based on the access pressure and temperature status of each memory controller channel: Dynamically monitor the access request pressure and current temperature of each memory channel; Dynamically adjust the mapping path of access requests based on access pressure and temperature status; When the current temperature of the memory area is higher than the set temperature threshold, set access redirection rules; For tasks that meet the preset task attributes, the memory area with the lowest heat load on the current access path is selected for access; Dynamically adjust memory address mapping rules based on memory access patterns and current temperature status; The set temperature threshold is a reference temperature value set by the system based on the hardware specifications, which is used to determine whether the hardware is in normal working condition. The set increase threshold represents the maximum allowable value of temperature change per unit time. The set running time is the judgment basis for the task running time, for example, 30 minutes. The set access frequency threshold is the upper limit of the number of accesses per unit time. The set write frequency threshold is the upper limit of the write operations per unit time. The set duration represents the time threshold for the high temperature state to continue, for example, 5 minutes. The preset task attribute refers to the task meeting any of the following conditions: the running time exceeds the set running time, the heat contribution value exceeds the set increase threshold, and the priority is lower than the set level. The heat contribution value represents the heat impact generated by the task on the hardware unit during execution, and is estimated by the thermal power model trained with historical operation data; The memory access path is selected based on the current access density, historical access records, and temperature conditions. The goal is to avoid local overheating caused by local access concentration. The system uses three mechanisms, task scheduling, data migration, and memory access path optimization, to work together to periodically adjust task and data distribution based on thermal monitoring data, thus achieving dynamic control of heat source distribution. Each mechanism operates independently and is controlled by a unified thermal feedback mechanism. The three share status information to form a closed-loop control structure, ensuring that the thermal management strategy can continuously adapt to changes in the system's operating status. The thermal zone prediction and intervention module builds a high-temperature zone prediction model by analyzing operating status and temperature trend data, thereby evaluating and marking the future thermal status of hardware units. Based on thermal power density calculation and temperature change modeling, it identifies potential high-temperature risk units and dynamically adjusts the heat source distribution by combining task scheduling, data migration and memory access path optimization. Task scheduling is prioritized to low-temperature areas to support task migration and concurrency control. Data migration optimizes distribution based on heat and temperature status to balance load and performance. Memory access paths are dynamically redirected to avoid local overheating. The three mechanisms work together to form a closed-loop feedback loop, continuously adapting to changes in system operation and improving overall thermal management efficiency and system stability.

[0020] Embodiment 2: The technical solution of this embodiment of the present invention differs from that of Embodiment 1 in that: like Figure 4 As shown in the figure, the adaptive strategy generation module is based on a multi-objective evolutionary algorithm, combined with strategy lifecycle management and model collaborative iteration mechanism to generate and update heat dissipation control strategies that adapt to different hardware configurations; The adaptive strategy generation module uses a multi-objective evolutionary algorithm to evolve the process including: The system must meet the following three constraints when generating a cooling control strategy: Control the temperature value of the hardware unit to be within the set range; The total power consumption of the control system does not exceed the set upper limit; Maintaining mission execution performance not below the set lower limit; Since there are conflicts among these three constraints, a multi-objective evolutionary algorithm is used to search for a strategy combination that satisfies all constraints in the solution space; Application of Multi-Objective Evolutionary Algorithm: The multi-objective evolutionary algorithm simulates the biological evolution process to find a set of non-inferior solutions that meet multiple constraints in the solution space. For the heat dissipation control strategy, the following objective function is set: Temperature control target: Calculate the difference between the current hardware unit temperature value and the set safety temperature upper limit; Power consumption control target: Calculate the difference between the current system power consumption and the maximum allowable power consumption; Performance maintenance target: Calculate the difference between the minimum acceptable performance and the current task execution performance; The smaller the objective function value, the better the strategy performs on the corresponding constraints; The multi-objective evolutionary algorithm evolves according to the following steps: Z1: Select individuals with better performance from the current population; Z2: Combine the parameters of two or more individuals to generate a new candidate strategy; Z3: Perturb the parameters of some candidate strategies and introduce new solutions; Z4: Calculate the value of each candidate strategy on each objective function; Z5: retain non-inferior solutions to form the population of the next generation; The process continues to run until the set maximum number of iterations or convergence condition is reached; The adaptive strategy generation module combines strategy lifecycle management with the model collaborative iteration mechanism to generate a heat dissipation control strategy. The process includes: Policy lifecycle management: The initial policy is generated by a multi-objective evolutionary algorithm, with parameters determined based on the current hardware configuration and operating environment; The system collects the following data for strategy evaluation: hardware unit temperature, system power consumption, and task execution performance. Based on the collected data, it determines whether the current strategy meets the objective function requirements. When the existing strategy cannot meet the objective function, the strategy update process is initiated. The update methods include: adjusting the objective function weight, introducing new variables, and re-running the evolutionary algorithm to generate a new strategy; Model collaborative iteration mechanism: To improve the adaptability of the strategy to different operating conditions, a model collaborative iteration mechanism is established. The initial heat dissipation control model is trained using historical operating data. The input includes hardware configuration information, ambient temperature, and task load characteristics. The output is a set of candidate strategies and their corresponding constraint satisfaction. During operation, the system continuously collects data and feeds it back to the model to adjust policy parameters. The feedback data includes: current temperature distribution, power consumption fluctuations, and performance index changes. The model updates its internal parameters based on the feedback results. When the operating environment changes significantly, the model is retrained. The retraining process includes: collecting new operating data, constructing new training samples, retraining the model, and updating the parameter settings in the strategy generation module; The process by which the adaptive strategy generation module generates and updates the heat dissipation control strategy adapted to different hardware configurations includes: Adaptation mechanism for different hardware configurations: The heat dissipation control strategy needs to be adapted to different hardware configurations. The adaptation process is as follows: Hardware feature identification: read hardware identifiers, temperature sensor numbers, and power consumption limit parameters; Strategy matching: Select the adaptation strategy from the strategy library based on hardware characteristics; Strategy adjustment: If there is no perfect matching strategy, a new strategy is generated based on the evolutionary algorithm; Policy deployment: Apply the selected or generated policy to the target device; The generated cooling control policy contains the following operations: Temperature regulation: adjust the fan speed or cooling device working status according to the current temperature value; Power consumption allocation: Control the power consumption of each hardware unit so that its total does not exceed the set upper limit; Task scheduling: Dynamically adjust task allocation based on hardware load to balance temperature and performance requirements; Hardware start and stop: decide whether to shut down non-critical hardware modules to reduce heat generation based on the operating status; The adaptive strategy generation module is based on a multi-objective evolutionary algorithm and searches for the optimal cooling strategy under the triple constraints of temperature, power consumption, and performance. It continuously optimizes strategy combinations through evolutionary steps such as selection, crossover, and mutation to generate a set of non-inferior solutions. It combines strategy lifecycle management to implement strategy evaluation, updating, and dynamic adjustment. Relying on the model collaborative iteration mechanism, it uses real-time data feedback to optimize model parameters and improve strategy adaptability. It supports automatic matching and strategy generation for different hardware configurations. The adaptation process covers identification, matching, adjustment, and deployment to ensure that the strategy is portable and scalable. The generated strategy covers temperature regulation, power consumption control, task scheduling, and hardware start and stop, forming a complete cooling control closed loop to ensure stable system operation.

[0021] The above are only preferred specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes based on the technical solutions and improved concepts of the present invention within the technical scope disclosed by the present invention, and they should be covered by the scope of protection of the present invention.

Claims

1. An industrial computer memory and storage collaborative heat dissipation system, applied to an industrial automation management platform, characterized in that: include: The data acquisition module is used to collect the operating status and thermal behavior data of the memory and storage unit in real time, and classify and pre-process the collected operating status and thermal behavior data; The temperature control monitoring module builds a thermal field model of the memory and storage units based on pre-processed operating status data and thermal behavior data, and outputs the temperature distribution of the heat accumulation area and the corresponding thermal status identification information; The hot zone prediction and intervention module builds a high-heat zone prediction model based on operating status data and temperature trend information. It combines task scheduling strategies, data migration mechanisms, and memory access path optimization to dynamically adjust heat source distribution. The adaptive strategy generation module, based on a multi-objective evolutionary algorithm, combines strategy lifecycle management with a model collaborative iteration mechanism to generate and update cooling control strategies that adapt to different hardware configurations.

2. The industrial computer memory and storage collaborative heat dissipation system according to claim 1, characterized in that: The process of constructing the thermal field model of the memory and storage unit by the temperature control monitoring module includes: Obtain pre-processed operational status data and thermal behavior data. The operational status data includes IOPS, throughput, ECC single-bit error count, and total number of bad blocks. The thermal behavior data includes core temperature, ambient temperature, and fan speed. Constructing the hardware space topology: 3D modeling of the memory modules and storage devices in industrial computers according to their actual physical layout in the motherboard or chassis; Each device is further divided into multiple hardware units and used as the basic nodes of the thermal field model; Establish a thermal network structure based on hardware units to form a complete heat conduction path diagram within the system; Establish heat source mapping relationship: Based on the collected operating status data, a thermal power density model is established for each hardware unit: ； where Q i is the thermal power per unit time of the i-th hardware unit, All are empirical calibration coefficients. IOPS, throughput, ECC, and bad block counts are pre-processed and standardized operating status data. Set up thermal boundary conditions and cooling the model: The heat conduction boundary conditions are set according to the ambient temperature and fan speed, and the heat convection coefficient h and wind speed function v (RPM) are introduced to establish the cooling capacity model: , where q sr is the heat flux per unit area, A is the heat conduction surface area, T xp is the chip surface temperature, T hj is the ambient temperature; Solve for the steady-state distribution of the thermal field: Substitute the heat source distribution and cooling boundary conditions into the heat diffusion governing equation: , where k is the thermal conductivity of the material, T is the temperature field, and Q is the power density of the heat source; The numerical method is used to solve the heat diffusion control equation to obtain the steady-state temperature distribution T of each hardware unit. i .

3. The industrial computer memory and storage collaborative heat dissipation system according to claim 2, characterized in that: The process of the temperature control monitoring module outputting the temperature distribution of the heat accumulation area and the corresponding thermal state identification information includes: Temperature distribution output in heat accumulation areas: The steady-state temperature distribution obtained by numerically solving the heat diffusion equation is used to construct the temperature field map inside the system; Based on the hardware unit division and hardware physical layout, the system is divided into multiple heat accumulation areas; Thermal status identification information generation: The system classifies the status of each heat accumulation area according to the preset temperature threshold and generates the corresponding thermal status identification; Thermal status indicators include the following categories: Normal state: the area temperature is lower than the set temperature upper limit and the load is in a stable state; Critical state: The zone temperature is higher than 90% of the set temperature upper limit but does not exceed the upper limit, and there is load fluctuation; Overheating: The zone temperature exceeds the set upper temperature limit, or the increase in the ECC error count within a unit monitoring cycle exceeds the preset threshold; Failure warning state: The area temperature is continuously higher than the set temperature upper limit for a preset period of time, and the growth rate of the number of new bad blocks in a unit monitoring cycle exceeds the set threshold; State output method: The system organizes the temperature field calculation results and thermal state identification information into a structured data format and outputs them.

4. The industrial computer memory and storage collaborative heat dissipation system according to claim 1, characterized in that: The process of constructing a high-heat zone prediction model by the heat zone prediction intervention module includes: Obtain pre-processed operating status data and temperature trend information, where the temperature trend information is the steady-state temperature distribution data of each hardware unit under continuous load conditions; Based on the relationship between the pre-processed operating status data and the historical temperature data, a thermal power density model for each hardware unit is established. The specific steps are as follows: S1: Analyze the correlation between operating parameters and temperature changes; S2: Extract the thermal response characteristics of each hardware unit under the same load; S3: Use thermal power density function to represent the thermal power density of each hardware unit; S4: construct heat intensity distribution map; Based on the thermal power density and heat conduction characteristics, a prediction model for temperature changes in the future time period is established. The specific steps are as follows: T1: Analyze the time series characteristics of historical temperature data; T2: extraction temperature change rate and fluctuation range; T3: Calculate future temperature changes based on current thermal power density and thermal diffusion capacity; T4: Set the prediction time window to identify hardware units whose temperatures will exceed the threshold in the future; The system marks each hardware unit as a candidate for a hotspot based on the following four conditions: The current temperature reaches or exceeds the set threshold; The temperature change trend shows an increase at multiple consecutive time points; The thermal power density is higher than the overall system average; The heat source concentration is higher than the set threshold; The system outputs the following information: the high-temperature zone number, the temperature value of the corresponding hardware unit, and the status identification.

5. The industrial computer memory and storage collaborative heat dissipation system according to claim 4, characterized in that: The process of dynamically adjusting the heat source distribution by the hot zone prediction intervention module in combination with task scheduling strategy, data migration mechanism and memory access path optimization includes: To optimize the task scheduling strategy, the system performs the following operations based on the current temperature of each hardware unit and the task thermal contribution value output by the thermal trend prediction model: Assign new tasks to hardware units whose current temperature is lower than the set temperature threshold; According to the task heat contribution value, a target execution unit is selected from the hardware units whose temperature increment is expected to be less than a set increment threshold within a preset time period in the future; Migrate tasks running on hardware units with temperatures higher than a set temperature threshold to available hardware units with temperatures lower than the threshold and with the smallest thermal load increment. Perform periodic migration for tasks that run for longer than the set duration. Dynamically adjust CPU and I / O scheduling weights; The task scheduler receives the temperature prediction data and heat contribution value output from the hot zone prediction module; Data migration mechanism: The system performs the following operations based on the data read and write frequency and temperature status of each hardware unit: Real-time monitoring of data reading and writing frequency per unit time; When the number of accesses per unit time of a data block exceeds the set access frequency threshold, it is migrated from the current hardware unit to an adjacent hardware unit whose current temperature is lower than the set temperature threshold; Creating a data copy in a hardware unit whose current temperature is lower than a set temperature threshold; When the data writing frequency exceeds the set writing frequency threshold, the writing operation to the area is restricted; When the temperature of the hardware unit is continuously higher than the set temperature threshold for more than the set duration, the data in it will be migrated out; To optimize memory access paths, the system performs the following operations based on the access pressure and temperature status of each memory controller channel: Dynamically monitor the access request pressure and current temperature of each memory channel; Dynamically adjust the mapping path of access requests based on access pressure and temperature status; When the current temperature of the memory area is higher than the set temperature threshold, set access redirection rules; For tasks that meet the preset task attributes, the memory area with the lowest heat load on the current access path is selected for access; Dynamically adjust memory address mapping rules based on memory access patterns and current temperature status.

6. The industrial computer memory and storage collaborative heat dissipation system according to claim 1, characterized in that: The process of the adaptive strategy generation module evolving using a multi-objective evolutionary algorithm includes: The system must meet the following three constraints when generating a cooling control strategy: Control the temperature value of the hardware unit to be within the set range; The total power consumption of the control system does not exceed the set upper limit; Maintaining mission execution performance not below the set lower limit; Application of Multi-Objective Evolutionary Algorithm: The multi-objective evolutionary algorithm simulates the biological evolution process to find a set of non-inferior solutions that meet multiple constraints in the solution space. For the heat dissipation control strategy, the following objective function is set: Temperature control target: Calculate the difference between the current hardware unit temperature value and the set safety temperature upper limit; Power consumption control target: Calculate the difference between the current system power consumption and the maximum allowable power consumption; Performance maintenance target: Calculate the difference between the minimum acceptable performance and the current task execution performance; The multi-objective evolutionary algorithm evolves according to the following steps: Z1: Select individuals with better performance from the current population; Z2: Combine the parameters of two or more individuals to generate a new candidate strategy; Z3: Perturb the parameters of some candidate strategies and introduce new solutions; Z4: Calculate the value of each candidate strategy on each objective function; Z5: Retain non-inferior solutions to form the population of the next generation.

7. The industrial computer memory and storage collaborative heat dissipation system according to claim 6, characterized in that: The process of generating a heat dissipation control strategy by combining the strategy lifecycle management and the model collaborative iteration mechanism by the adaptive strategy generation module includes: Policy lifecycle management: The initial policy is generated by a multi-objective evolutionary algorithm, with parameters determined based on the current hardware configuration and operating environment; The system collects the following data for strategy evaluation: hardware unit temperature values, system power consumption, and task execution performance; When the existing strategy cannot meet the objective function, the strategy update process is initiated. The update methods include: adjusting the objective function weight, introducing new variables, and re-running the evolutionary algorithm to generate a new strategy; Model collaborative iteration mechanism: Use historical operating data to train the initial heat dissipation control model. The input includes hardware configuration information, ambient temperature, and task load characteristics. The output is a set of candidate strategies and their corresponding constraint satisfaction. During operation, the system continuously collects data and feeds it back to the model to adjust policy parameters. The feedback data includes: current temperature distribution, power consumption fluctuations, and performance index changes; When the operating environment changes significantly, model retraining is performed. The retraining process includes: collecting new operating data, constructing new training samples, retraining the model, and updating the parameter settings in the strategy generation module.

8. The industrial computer memory and storage collaborative heat dissipation system according to claim 7, characterized in that: The process of the adaptive strategy generation module generating and updating the heat dissipation control strategy adapted to different hardware configurations includes: Adaptation mechanism for different hardware configurations: The heat dissipation control strategy needs to be adapted to different hardware configurations. The adaptation process is as follows: Hardware feature identification: read hardware identifiers, temperature sensor numbers, and power consumption limit parameters; Strategy matching: Select the adaptation strategy from the strategy library based on hardware characteristics; Strategy adjustment: If there is no perfect matching strategy, a new strategy is generated based on the evolutionary algorithm; Policy deployment: Apply the selected or generated policy to the target device; The generated cooling control policy contains the following operations: Temperature regulation: adjust the fan speed or cooling device working status according to the current temperature value; Power consumption allocation: Control the power consumption of each hardware unit so that its total does not exceed the set upper limit; Task scheduling: Dynamically adjust task allocation based on hardware load to balance temperature and performance requirements; Hardware start and stop: Determine whether to shut down non-critical hardware modules to reduce heat based on the operating status.

Images