Multi-mode sensing network cooling management system and method based on AI

Abstract

The invention relates to the technical field of heat dissipation management, in particular to a multi-modal sensing network cooling management system and method based on AI, and the system comprises a task heat load classification module, a fluid inertia compensation regulation and control module, a heat dissipation contribution degree evaluation module, a cooling strategy sharing optimization module and a cluster heat management federation calculation module. According to the method, by analyzing the ratio of the task calculation period to the power consumption cumulant and subdividing the task thermal load, cooling parameter adjustment better meets the actual thermal load requirement, targeted cooling measures can be implemented according to different task types, the heat dissipation efficiency is improved, and by adjusting the cooling strategy in real time, the cooling efficiency is improved. The cooling strategy is optimized based on the heat dissipation contribution degree, the overall energy efficiency and reliability of the server are effectively improved, optimization is achieved in a single server through federated calculation of cluster heat management, optimal configuration of cooling parameters is achieved in a whole server cluster through a federated learning mechanism, and the reliability of the server is improved. And the overall performance and stability are ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of heat dissipation management, and particularly to an AI-based multi-modal sensing network cooling management system and method. Background Art

[0002] The technical field of heat dissipation management involves controlling the heat generated during the operation of electronic devices, computing systems, industrial equipment, etc., to maintain the stable operation of the equipment and improve energy efficiency. This technical field covers passive and active heat dissipation methods and is optimized by combining principles such as fluid dynamics, heat transfer, and materials science. In recent years, this field has gradually incorporated technologies such as artificial intelligence, intelligent sensing, and dynamic regulation to achieve precise temperature management, energy consumption optimization, and equipment reliability improvement. Especially in high-density computing environments, heat dissipation management plays a crucial role in energy efficiency and equipment lifespan.

[0003] The multi-modal sensing network cooling management system is an intelligent cooling control system based on the fusion of various sensing data, dedicated to the liquid cooling of servers. This system utilizes multi-dimensional sensor data such as temperature, flow rate, and pressure, analyzes the coolant state through intelligent algorithms, and realizes real-time adjustment of cooling parameters to optimize heat dissipation efficiency, reduce energy consumption, and improve server stability. It is mainly applied to data centers, cloud computing servers, and high-performance computing environments, aiming to enhance the adaptive regulation ability of the liquid cooling system and ensure the long-term efficient operation of the server while reducing energy consumption.

[0004] Traditional management systems rely on unified cooling strategies and lack adaptive adjustments for various heat load situations, resulting in overcooling when the heat load is low and difficulty in timely response when the heat load suddenly increases, affecting the energy efficiency and reliability of the equipment. For example, in a data center, due to the inability to accurately dynamically evaluate the heat loads of each server, some servers may overheat due to insufficient cooling, while other servers will consume excessive cooling resources under low load. In the cooling management of server clusters in the prior art, there is a lack of effective data sharing and collaboration mechanisms, making the adjustment of the overall heat dissipation strategy unable to be optimized in real time according to the actual operating status of each server, affecting the overall heat dissipation efficiency and energy consumption. Summary of the Invention

[0005] The purpose of the present invention is to solve the drawbacks existing in the prior art, and to propose an AI-based multi-modal sensing network cooling management system and method.

[0006] To achieve the above purpose, the present invention adopts the following technical scheme: an AI-based multi-modal sensing network cooling management system, the system includes:

[0007] The task heat load classification module obtains multi-modal sensor network data, extracts the task calculation cycle of the server calculation task and the power consumption accumulation amount per unit time, collects the chip temperature change trend, calculates the task heat accumulation rate, and calculates the task heat load information;

[0008] The fluid inertia compensation and regulation module calculates the fluid kinetic energy and the flow inertia influence amount based on the task heat load information, and adjusts the coolant flow rate according to the task category to obtain an optimized cooling flow rate value;

[0009] The heat dissipation contribution evaluation module calculates the temperature drop rate per unit time based on the optimized cooling flow rate value, and calculates the server heat dissipation contribution in combination with the energy consumption change before and after the adjustment;

[0010] The cooling strategy sharing and optimization module adjusts the weights of the server cooling strategy sharing model based on the server heat dissipation contribution, screens the servers with high contribution as the main reference nodes, and adjusts the calculation logic of the sharing strategy to obtain the optimized sharing strategy parameters;

[0011] The cluster thermal management federated computing module constructs a cooling parameter federated learning computing matrix based on the optimized sharing strategy parameters, calculates the adaptability of the server cooling execution parameters within the cluster, and adjusts the cooling strategy sharing parameters to obtain the server cluster cooling control information.

[0012] The improvements of the present invention are that the task heat load information is specifically the heat accumulation ratio of the calculation cycle, the power consumption accumulation amount per unit time, and the chip temperature change rate; the optimized cooling flow rate value includes the fluid kinetic energy adjustment coefficient, the flow inertia compensation amount, and the target coolant flow rate; the server heat dissipation contribution specifically refers to the temperature drop rate per unit time, the cooling energy consumption ratio, and the energy consumption change before and after the heat dissipation adjustment; the optimized sharing strategy parameters include the contribution weight adjustment value, the main reference node screening parameter, and the sharing strategy calculation correction item; the server cluster cooling control information is specifically the cooling parameter federated computing matrix, the server cooling adaptability, and the cooling strategy sharing parameter.

[0013] The improvements of the present invention are that the task heat load classification module includes:

[0014] The task calculation feature extraction sub-module obtains multi-modal sensor network data, extracts the task calculation cycle of the server calculation task, calculates the power consumption accumulation amount per unit time, monitors the chip temperature change trend, and calculates the chip temperature change rate according to the time series data to obtain a task calculation feature data set;

[0015] The task heat accumulation rate calculation sub-module calls the task calculation feature data set and uses the formula:

[0016]

[0017] Calculate to obtain the task heat accumulation rate and get the task heat accumulation rate value;

[0018] Among them, R h represents the task heat accumulation rate, E tot represents the power consumption accumulation amount per unit time, T comp represents the task calculation period, T chip,i represents the chip temperature at the i-th moment, Δt represents the temperature sampling interval time, and n represents the time series length;

[0019] The task classification determination sub-module calls the task heat accumulation rate value, sets a classification threshold, determines the task type based on the ratio, classifies it into short-term high-heat tasks, long-term medium-heat tasks, and low-heat tasks, and establishes task heat load information.

[0020] The improvement of the present invention is that the fluid inertia compensation regulation module includes:

[0021] The fluid kinetic energy calculation sub-module calculates the specific kinetic energy of the fluid per unit mass based on the task heat load information, calls the current coolant flow rate of the server and the specific heat capacity of the fluid per unit mass, calculates the energy change amount during the coolant temperature rise process by calling the temperature gradient of the heat exchange plate, and combines the task heat load coefficient to calculate the fluid kinetic energy;

[0022] The inertia influence analysis sub-module calls the fluid kinetic energy, calculates the flow inertia of the coolant in the pipeline, calculates the momentum change rate of the fluid according to the coolant density, flow rate and fluid cross-sectional area, and combines the heat exchange plate temperature gradient, using the formula:

[0023]

[0024] Calculate the influence of the flow inertia on the heat exchange process to obtain the flow inertia influence amount;

[0025] Among them, I represents the flow inertia influence amount, m represents the mass of the coolant per unit volume, v represents the coolant flow rate, ρ represents the coolant density, A represents the cross-sectional area of the coolant pipeline, ΔT p represents the heat exchange plate temperature gradient, c p represents the specific heat capacity of the coolant;

[0026] The cooling flow rate regulation sub-module calls the flow inertia influence amount, adjusts the coolant flow rate according to the task category, including increasing the coolant flow rate of the server for short-term high-heat tasks, periodically adjusting the coolant temperature of the server for long-term medium-heat tasks, and reducing the flow rate of the server for low-heat tasks to obtain the optimized cooling flow rate value.

[0027] The improvement of the present invention is that the heat dissipation contribution degree evaluation module includes:

[0028] Based on the optimized value of the cooling flow rate, the temperature change rate calculation sub-module obtains the chip temperature data before and after the adjustment of the server cooling system, calculates the temperature change rate before and after the adjustment, and obtains the chip temperature change rate;

[0029] The cooling energy consumption ratio calculation sub-module calls the chip temperature change rate, obtains the server power consumption data before and after the adjustment of the server cooling system, and uses the formula:

[0030]

[0031] Calculate the ratio of the temperature drop rate per unit time to the cooling energy consumption to obtain the cooling energy consumption ratio;

[0032] Among them, R E represents the cooling energy consumption ratio, ΔT t represents the temperature change rate at time t, E t represents the cooling energy consumption at time t, N represents the total number of data samplings, max(ΔT t ) represents the maximum temperature drop rate within the sampling period, and max(E t ) represents the maximum cooling energy consumption within the sampling period;

[0033] The heat dissipation contribution degree calculation sub-module calls the cooling energy consumption ratio, obtains the energy consumption data before and after the heat dissipation adjustment of the server, and according to the reduced amount of energy consumption after the adjustment, uses the formula:

[0034]

[0035] Calculate to obtain the server heat dissipation contribution degree;

[0036] Among them, C represents the server heat dissipation contribution degree, Q before represents the energy consumption of the cooling system before adjustment, Q after represents the energy consumption of the cooling system after adjustment, and R E represents the cooling energy consumption ratio.

[0037] The improvement of the present invention is that the cooling strategy sharing optimization module includes:

[0038] Based on the server heat dissipation contribution degree, the server contribution degree weighting sub-module obtains the contribution degree data of each server in the cluster, calculates the proportion of the contribution degree of each server, normalizes the contribution degree value of each server, and calculates the influence weight of the cooling strategy of each server to obtain the server contribution degree weight value;

[0039] The shared strategy master node screening sub-module calls the server contribution degree weight value, sets a contribution degree weight threshold, and screens the servers higher than the threshold as the main reference nodes, using the formula:

[0040]

[0041] Calculate the influence degree of the reference node on the overall cooling strategy, and obtain the influence value of the main reference node of the sharing strategy;

[0042] Among them, W s represents the influence value of the main reference node of the sharing strategy, C j represents the heat dissipation contribution degree of the jth server, P j represents the load power consumption of the jth server, M represents the number of main reference servers selected, represents the average contribution degree of the selected servers;

[0043] The sharing strategy optimization calculation sub-module calls the influence value of the main reference node of the sharing strategy, calculates the sharing strategy influence matrix based on the contribution degree of the main reference node, adjusts the weight of the cooling strategy sharing model, optimizes the calculation logic of the sharing strategy, and generates the sharing strategy optimization parameters.

[0044] The improvement of the present invention is that the cluster thermal management federated computing module includes:

[0045] The cooling execution data statistics sub-module obtains the cooling execution data of each server in the server cluster based on the sharing strategy optimization parameters, records the coolant flow rate, heat exchange plate temperature gradient, server power consumption, and temperature drop rate during the execution of the cooling strategy of each server, and establishes a server cooling execution data set;

[0046] The federated computing matrix construction sub-module calls the server cooling execution data set, constructs a cooling parameter federated learning computing matrix, and uses the formula;

[0047]

[0048] Calculate the server cooling adaptability;

[0049] Among them, A d represents the server cooling adaptability, N A represents the total number of servers participating in the calculation in the server cluster, P A,f represents the power consumption of the fth server, T A,f represents the temperature drop rate of server f, represents the average value of the power consumption in the server cluster, represents the average value of the temperature drop rate in the server cluster, represents the standard deviation of the power consumption in the server cluster, represents the standard deviation of the temperature drop rate in the server cluster, and e is the base of the natural logarithm;

[0050] The shared parameter optimization sub-module calls the server cooling adaptability, adjusts the shared parameters of the server cooling strategy, sets the optimization rules for the cluster cooling strategy, and obtains the server cluster cooling control information.

[0051] An AI-based multi-modal sensing network cooling management method, which is executed based on the above-mentioned AI-based multi-modal sensing network cooling management system, includes the following steps:

[0052] S1: Obtain multi-modal sensing network data, extract the task calculation cycle and the power consumption accumulation amount per unit time of the server computing task, calculate the task heat accumulation rate, and calculate the task heat load information;

[0053] S2: The fluid inertia compensation regulation module calculates the fluid kinetic energy and the flow inertia influence amount based on the task heat load information, adjusts the coolant flow rate, and obtains the optimized cooling flow rate value;

[0054] S3: The heat dissipation contribution evaluation module calculates the temperature drop rate per unit time based on the optimized cooling flow rate value, and calculates the server heat dissipation contribution in combination with the energy consumption change before and after adjustment;

[0055] S4: The cooling strategy sharing optimization module selects the servers with high contribution as the main reference nodes based on the server heat dissipation contribution, adjusts the calculation logic of the sharing strategy, and obtains the optimized sharing strategy parameters;

[0056] S5: The cluster thermal management federated calculation module constructs a cooling parameter federated learning calculation matrix based on the optimized sharing strategy parameters, calculates the adaptability of the server cooling execution parameters within the cluster, adjusts the shared parameters of the cooling strategy, and obtains the server cluster cooling control information.

[0057] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0058] In the present invention, by analyzing the ratio of the task calculation cycle to the power consumption accumulation amount, the task heat load is subdivided, so that the adjustment of the cooling parameters is more in line with the actual heat load requirements, and targeted cooling measures can be implemented according to different task types, improving the heat dissipation efficiency. By adjusting the cooling strategy in real time and optimizing the cooling strategy based on the heat dissipation contribution, the overall energy efficiency and reliability of the server are effectively improved. Through the federated calculation of cluster thermal management, not only is the optimization achieved within a single server, but also the optimal configuration of the cooling parameters is achieved in the entire server cluster through the federated learning mechanism, ensuring the overall performance and stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 It is the system flow chart of the present invention;

[0060] Figure 2 This is the flowchart of the task heat load classification module of the present invention;

[0061] Figure 3 This is the flowchart of the fluid inertia compensation regulation module of the present invention;

[0062] Figure 4 This is the flowchart of the heat dissipation contribution evaluation module of the present invention;

[0063] Figure 5 This is the flowchart of the cooling strategy sharing and optimization module of the present invention;

[0064] Figure 6 This is the flowchart of the cluster thermal management federated computing module of the present invention. Detailed implementation manners

[0065] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0066] In the description of the present invention, it should be understood that the orientation or positional relationship indicated by the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. is the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation to the present invention. In addition, in the description of the present invention, "a plurality of" means two or more, unless otherwise specifically defined.

[0067] Please refer to Figure 1 , the present invention provides a technical solution: an AI-based multi-modal sensing network cooling management system, the system includes:

[0068] The task heat load classification module obtains multi-modal sensing network data, extracts the task calculation cycle and the power consumption accumulation amount per unit time of the server calculation task, collects the chip temperature change trend, extracts the task heat load characteristics, calculates the task heat accumulation rate according to the ratio of the task calculation cycle to the power consumption accumulation amount, sets a classification threshold according to the ratio, classifies the tasks into short-term high-heat tasks, long-term medium-heat tasks, and low-heat tasks, and calculates the task heat load information according to the classification results;

[0069] The fluid inertia compensation regulation module calculates the fluid kinetic energy and the flow inertia influence amount based on the task heat load information, invokes the current coolant flow rate, the heat exchange plate temperature gradient, and the specific heat capacity per unit mass of the fluid of the server, and adjusts the coolant flow rate according to the task category to obtain an optimized cooling flow rate value;

[0070] Based on the optimized cooling flow rate value, the heat dissipation contribution evaluation module obtains the chip temperature change rate and the server power consumption before and after the adjustment of the server cooling system, calculates the temperature drop rate per unit time, and combines the energy consumption changes before and after the server heat dissipation adjustment to calculate the server heat dissipation contribution;

[0071] Based on the server heat dissipation contribution, the cooling strategy sharing optimization module obtains the heat dissipation contribution data of the servers in the cluster, assigns weights according to the contribution size, adjusts the weights of the server cooling strategy sharing model, screens the servers with high contribution as the main reference nodes, and adjusts the calculation logic of the sharing strategy to obtain the optimized sharing strategy parameters;

[0072] Based on the optimized sharing strategy parameters, the cluster thermal management federated computing module obtains the cooling execution data of each server in the server cluster, constructs a federated learning calculation matrix for cooling parameters, calculates the adaptability of the server cooling execution parameters in the cluster, and adjusts the server cooling strategy sharing parameters to obtain the server cluster cooling control information.

[0073] The task heat load information specifically includes the heat accumulation ratio in the calculation period, the cumulative power consumption per unit time, and the chip temperature change rate. The optimized cooling flow rate value includes the fluid kinetic energy adjustment coefficient, the flow inertia compensation amount, and the target coolant flow rate. The server heat dissipation contribution specifically refers to the temperature drop rate per unit time, the cooling energy consumption ratio, and the energy consumption change before and after the heat dissipation adjustment. The optimized sharing strategy parameters include the contribution weight adjustment value, the main reference node screening parameter, and the sharing strategy calculation correction term. The server cluster cooling control information specifically includes the federated calculation matrix for cooling parameters, the server cooling adaptability, and the cooling strategy sharing parameters.

[0074] Please refer to Figure 2 , the task heat load classification module includes:

[0075] The task calculation feature extraction sub-module obtains the multi-modal sensor network data, extracts the task calculation period of the server calculation task, calculates the cumulative power consumption per unit time, monitors the chip temperature change trend, and calculates the chip temperature change rate according to the time series data to obtain the task calculation feature data set;

[0076] The task calculation feature extraction sub-module receives data from the multi-modal sensor network, including but not limited to the real-time power consumption data collected by the power consumption sensor, the temperature data obtained by the chip internal temperature sensor, and the task scheduling log on the server side. First, analyze the execution process of the server calculation task, read the task start and end times, and calculate the task calculation period T comp as the time difference between the two. For example, if the start time of a task is 12:00:00 and the end time is 12:00:30, then T comp= 30 s. Meanwhile, accumulate the power consumption per unit time within this calculation period to obtain E tot , and obtain the total power consumption by accumulating the power consumption measurement values per second. For example, if the power consumptions of a certain task within 30 s are [8.5 W, 8.7 W, 8.6 W, …, 9.0 W], then E tot =

[0077] ∑P i ×Δt, where Δt = 1 s, and finally obtain E tot = 258.4 J. In addition, monitor the changing trend of the chip temperature, read the temperature values in the time series, and calculate the chip temperature change rate. For the temperature measurement points {T chip,0 , T chip,1 ,..., T chip,n}, calculate the temperature change rate at adjacent moments For example, for the temperature sequence [50.2 °C, 50.5 °C, 51.0 °C, 51.3 °C], the change rates are calculated as [0.3 °C / s, 0.5 °C / s, 0.3 °C / s]. Based on the comprehensive calculation results, form a task calculation feature data set, including data items such as the task calculation period, the accumulated power consumption, and the temperature change rate, and store them in the database. During this process, the measurement standard of the accumulated power consumption per unit time E tot is based on the designed power consumption range of the server-side chip. Generally, 10% to 90% of the peak power consumption of the chip is selected as a reasonable measurement interval. Taking a common computing chip with a TDP (Thermal Design Power) of 100 W as an example, the power consumption range during actual operation is 10 W to 90 W. A power consumption lower than 10 W may indicate an idle state, and a power consumption higher than 90 W may indicate an abnormal load. Therefore, if the measured power consumption data is lower than 10 W, it may belong to a low-load or idle state. If it is higher than 90 W, there may be an overload risk, and it should be comprehensively analyzed in combination with the calculation period.

[0078] The task heat accumulation rate calculation sub-module calls the task calculation feature data set and uses the formula:

[0079]

[0080] Perform operations to obtain the task heat accumulation rate and get the task heat accumulation rate value;

[0081] Among them, R h represents the task heat accumulation rate, E tot represents the accumulated power consumption per unit time, T comp represents the task calculation period, T chip,i represents the chip temperature at the i-th moment, Δt represents the temperature sampling interval time, and n represents the length of the time series;

[0082] The task heat accumulation rate calculation sub-module receives the task calculation feature data set and based on the formula:

[0083]

[0084] Perform calculations. Calculate the power consumption per unit time, i.e., such as E tot = 258.4 J, T comp = 30 s, then Secondly, calculate the sum of squares of the temperature change rate, e.g.:

[0085]

[0086] Take the square root to get Calculate R h = 8.613 + 0.656 = 9.269, and store the result in the database. When calculating R h the influence factor of the sum of squares part depends on the temperature fluctuation range. If the temperature change rate is large, it indicates that the calculation task causes a sharp rise in the chip temperature, and there may be a heat dissipation bottleneck in the system. Therefore, for the temperature change rate calculation part, a correction coefficient α can be set to balance the impact of short-term severe fluctuations on the overall calculation. Usually, the value ranges from 0.5 to 2, and the setting of α is based on the chip operating temperature range and the cooling system response time. For example, for a general computing chip, the normal operating temperature range is 40°C to 85°C. If the temperature fluctuation exceeds 5°C / s, then α is taken as 1.5 to increase the influence weight of the temperature change on R h If the temperature changes slowly, α is taken as 0.8 to reduce the impact of short-term small temperature fluctuations. For this calculation, let α = 1.2, then R h The corrected calculation is as follows:

[0087] R h = 8.613 + 1.2×0.656 = 9.398;

[0088] Finally, R h = 9.398, and this result is stored and used for task classification determination.

[0089] The task classification determination sub-module calls the task heat accumulation rate value, sets the classification threshold, determines the task type based on the ratio, divides it into short-term high-heat tasks, long-term medium-heat tasks, and low-heat tasks, and establishes task heat load information;

[0090] The task classification determination sub-module calls the task heat accumulation rate value R h and classifies the task according to the classification threshold. The task classification threshold is set as follows:

[0091] Short-term high-heat tasks: R h > 10;

[0092] Long-term medium-heat tasks: 5 ≤ R h ≤ 10;

[0093] Low heat mission: R h <5;

[0094] The classification threshold is set based on the chip's heat dissipation capability and the type of computing task. Usually, high heat tasks (R h >10) corresponds to intensive computing loads such as high-performance computing and AI training. Such tasks have high requirements for the cooling system. When scheduling tasks, they should be assigned to computing nodes with active cooling capabilities first. Long-term medium-heat tasks (5≤R h ≤10) are usually ordinary computing tasks, such as database query, video rendering, etc., and the chip temperature is relatively stable. Low-heat tasks (R h <5) are mostly low-power operations such as I / O processing and data access. This classification threshold is adjusted according to chip test data. For example, in the test environment of Intel Xeon processor, the R corresponding to different task loads is h Typical values ​​are: Short-term high-heat task R h About 12-15, long-term medium heat task R h About 6-9, low heat task R h About 2-4, so the classification threshold is set as above.

[0095] According to the calculation results R h =9.398, the task is judged as a long-term medium heat task and recorded in the task heat load information database.

[0096] Table 1: Example of mission heat accumulation rate calculation

[0097]

[0098] As shown in Table 1, the heat accumulation rate calculation process of task number 1 complies with the set formula, and the calculated R h =9.398, and the classification result is a long-term medium-heat mission.

[0099] See also Figure 3 , the fluid inertia compensation control module includes:

[0100] The fluid kinetic energy calculation submodule, based on the task heat load information, calls the server's current coolant flow rate and the specific heat capacity per unit mass of the fluid to calculate the specific kinetic energy per unit mass of the fluid, calls the temperature gradient of the heat exchange plate to calculate the energy change during the coolant temperature rise process, and calculates the fluid kinetic energy in combination with the task heat load coefficient;

[0101] The fluid kinetic energy calculation submodule calls the server's current coolant flow rate and the specific heat capacity per unit mass of the fluid based on the task heat load information. First, read the current coolant flow rate v. For example, if the flow rate in the coolant pipeline is set to 2.5m / s, the specific heat capacity c is called at the same time. pAs a known physical property of the coolant, for example, 4.18 kJ / (kg·K) (taking water as the coolant as an example), calculate the specific kinetic energy of the fluid per unit mass, specifically calculate the kinetic energy where the mass m of the fluid per unit mass can be directly taken as 1 kg, and the calculation shows that:

[0102]

[0103] Next, call the temperature gradient ΔT of the heat exchange plate p , if the measured temperature gradient of the heat exchange plate is 15 K, calculate the energy change absorbed by the coolant during the temperature rise process, and use Q = mc p ΔT p for calculation, where m is set to 1 kg, then:

[0104] Q = 1 × 4.18 × 15 = 62.7 J;

[0105] Combine with the task heat load coefficient α h for adjustment. This heat load coefficient is used to correct the actual kinetic energy contribution of the coolant, and it is set according to the average calculation load ratio of the server. For example, when the server is in a full-load calculation state, α h takes 1, and when the server load is lower than 30%, α h takes 0.7. In the load range between 30% - 100%, this coefficient is calculated according to a linear relationship with the load. For example, if the current server load rate is 85%, then calculate Substitute into the calculation to get:

[0106] KE′ = 3.125 + 0.85 × 62.7 = 56.43 J;

[0107] Obtain the fluid kinetic energy of 56.43 J and store the calculation result in the database.

[0108] The inertial influence analysis sub-module calls the fluid kinetic energy to calculate the flow inertia of the coolant in the pipeline, calculates the rate of change of momentum of the fluid according to the coolant density, flow velocity and fluid cross-sectional area, and combines with the temperature gradient of the heat exchange plate, using the formula:

[0109]

[0110] Calculate the influence of the flow inertia on the heat exchange process to obtain the flow inertia influence amount;

[0111] where I represents the flow inertia influence amount, m represents the mass of the coolant per unit volume, v represents the coolant flow velocity, ρ represents the coolant density, A represents the cross-sectional area of the coolant pipeline, ΔT p represents the temperature gradient of the heat exchange plate, and c p represents the specific heat capacity of the coolant;

[0112] The inertial influence analysis sub-module calls the fluid kinetic energy to calculate the flow inertia of the coolant in the pipeline. First, calculate the rate of change of the coolant's momentum, based on the coolant density ρ, flow velocity v, and fluid cross-sectional area A. For example, assume the density of water ρ = 1000 kg / m³, the pipeline cross-sectional area A = 0.005 m², and the flow velocity v = 2.5 m / s.

[0113] Combined with the temperature gradient ΔT of the heat exchange plate p = 15 K, use the formula:

[0114]

[0115] Substitute the calculated parameters:

[0116]

[0117] Among them, ΔT p As the temperature gradient of the heat exchange plate, its value is set according to the temperature difference design criterion of the cooling system. This criterion is determined by the maximum operating temperature of the server and the coolant inlet temperature. For example, if the maximum allowable temperature of the server is set to 80 °C and the coolant inlet temperature is usually maintained below 65 °C, then a reasonable temperature gradient of the heat exchange plate should satisfy 10 K ≤ ΔT p ≤ 20 K. Within this range, ΔT p Can be adjusted according to the heat generation situation of specific tasks. For example, when the server is running high-load tasks, ΔT p Is set close to 20 K, and vice versa, set close to 10 K. In this example, 15 K is taken as the actual temperature gradient. Finally, calculate the flow inertia influence amount I = 48.01 and store the data in the database.

[0118] The cooling flow rate regulation sub-module calls the flow inertia influence amount and adjusts the coolant flow rate according to the task category, including increasing the coolant flow rate for servers with short-term high-heat tasks, periodically adjusting the coolant temperature for servers with long-term medium-heat tasks, and reducing the flow rate for servers with low-heat tasks to obtain the optimized cooling flow rate value;

[0119] The cooling flow rate regulation sub-module calls the flow inertia influence amount and adjusts the coolant flow rate according to the task category. First, set the flow rate regulation rules according to the task classification results. The classification criteria are as follows: Short-term high-heat tasks (R h > 10): Increase the coolant flow rate v' = v + 1 m / s. Long-term medium-heat tasks (5 ≤ R h ≤ 10): Periodically adjust the coolant temperature, such as adjusting ±2 K every 10 minutes. Low-heat tasks (R h < 5): Reduce the coolant flow rate v' = v - 0.5 m / s. Assume this task is classified as a long-term medium-heat task, then periodically adjust the coolant temperature, adjusting ±2 K every 10 minutes. Assume the current initial coolant temperature is T cIf it is 20°C, the coolant temperature in the next cycle is adjusted to T c ′ = 20 ± 2°C, that is, 18°C or 22°C. During the cooling flow rate regulation process, the adjustment amount of the coolant flow rate sets the adjustment threshold according to the inertia influence amount I. The threshold setting range is based on the response time of the cooling system to the flow rate change, generally set in the range of 5 ≤ I ≤ 50. When I > 50, the flow rate adjustment amplitude increases by 20%. When I < 5, the flow rate adjustment amplitude decreases by 10%. In this example, I = 48.01, close to 50, so the flow rate adjustment threshold takes the maximum adjustment range, that is, the periodic temperature adjustment is maintained within ±2K without changing the flow rate.

[0120] Table 2: Example of Inertia Influence Calculation

[0121]

[0122] As shown in Table 2, the calculation result of the flow inertia influence of task number 1 is 48.01, which is consistent with the calculation formula and is finally used for cooling flow rate regulation. According to the flow rate adjustment threshold, the coolant temperature adjustment amplitude of this task in the next cycle is set to ±2K.

[0123] Please refer to Figure 4 , the heat dissipation contribution evaluation module includes:

[0124] The temperature change rate calculation sub-module obtains the chip temperature data before and after the adjustment of the server cooling system based on the optimized cooling flow rate value, calculates the temperature change rate before and after the adjustment, and obtains the chip temperature change rate;

[0125] The temperature change rate calculation sub-module obtains the chip temperature data before and after the adjustment of the server cooling system based on the optimized cooling flow rate value. First, read the chip temperature data T before before the adjustment and the chip temperature data T after after the adjustment. The acquisition time interval is set to Δt = 5s to ensure the stability of data sampling. For example, the measured values of the chip temperature before the adjustment are [65.2°C, 64.8°C, 64.5°C, 64.0°C], and the measured values of the temperature after the adjustment are [63.0°C, 62.5°C, 62.2°C, 62.0°C]. Then calculate the temperature change rate before and after the adjustment using:

[0126]

[0127] Calculation before adjustment:

[0128]

[0129]

[0130] Calculation after adjustment:

[0131]

[0132] The chip temperature change rate is calculated. The average temperature change rate before adjustment is -0.08 °C / s, and after adjustment is -0.066 °C / s. Set the reference value T of the chip temperature change rate rate,thresh According to the operating temperature standard of the data center server, generally the temperature change rate is controlled between [-0.1 °C / s, -0.05 °C / s]. If the temperature change rate is lower than -0.1 °C / s, it may cause the coolant cooling rate to be too fast, affecting the temperature stability of the server. If the temperature change rate is higher than -0.05 °C / s, the cooling efficiency may be relatively low. Therefore, set the reference value T rate,thresh = -0.08 °C / s, which meets the temperature change rate before and after adjustment within this range.

[0133] The cooling energy consumption ratio calculation sub-module calls the chip temperature change rate, obtains the server power consumption data before and after the adjustment of the server cooling system, and uses the formula:

[0134]

[0135] Calculate the ratio of the temperature drop rate per unit time to the cooling energy consumption to obtain the cooling energy consumption ratio;

[0136] Among them, R E represents the cooling energy consumption ratio, ΔT t represents the temperature change rate at time t, E t represents the cooling energy consumption at time t, N represents the total number of data samplings, max(ΔT t ) represents the maximum temperature drop rate within the sampling period, max(E t ) represents the maximum cooling energy consumption within the sampling period;

[0137] The cooling energy consumption ratio calculation sub-module calls the chip temperature change rate, obtains the server power consumption data before and after the adjustment of the server cooling system, sets the total number of data samplings N = 3, and the collected cooling energy consumption data E t is set to be before adjustment

[0138] 30W, 32W, 31W], and after adjustment [29W, 30W, 29.5W], calculate:

[0139]

[0140] max(ΔT t ) = -0.06 °C / s;

[0141] max(E t ) = 32W;

[0142] Substitute into the formula:

[0143]

[0144] Obtain the cooling energy consumption ratio R E = 0.000048, set the cooling energy consumption ratio weight W E According to the energy efficiency ratio of the server cooling system, generally, the energy efficiency ratio of the data center cooling system ranges from 0.00002 to 0.0001. Therefore, set the cooling energy consumption ratio weight W E = 0.000048, this value is used to calculate the energy efficiency of the cooling system. If R E is lower than 0.00002, the cooling system efficiency is too low. If it is higher than 0.0001, it may cause the cooling system to operate excessively and affect the power consumption balance.

[0145] The heat dissipation contribution calculation sub-module calls the cooling energy consumption ratio, obtains the energy consumption data of the server before and after heat dissipation adjustment, and according to the reduction amount of energy consumption after adjustment, uses the formula:

[0146]

[0147] Calculate the heat dissipation contribution of the server;

[0148] Among them, C represents the heat dissipation contribution of the server, Q before represents the energy consumption of the cooling system before adjustment, Q after represents the energy consumption of the cooling system after adjustment, R E represents the cooling energy consumption ratio;

[0149] The heat dissipation contribution calculation sub-module calls the cooling energy consumption ratio, obtains the energy consumption data of the server before and after heat dissipation adjustment, and sets the energy consumption Q of the cooling system before adjustment before = 5000J, the energy consumption Q of the cooling system after adjustment after = 4800J, calculate:

[0150]

[0151] Obtain the heat dissipation contribution C of the server = 0.00000192, set the heat dissipation contribution coefficient K C According to the cooling system optimization goal, the general setting range is between 0.000001 and 0.00001. If the heat dissipation contribution coefficient is lower than 0.000001, it means that the cooling optimization effect is weak. If it is higher than 0.00001, it may cause the energy consumption optimization to be not obvious. Therefore, set K C = 0.00000192, meeting the reasonable heat dissipation contribution evaluation standard.

[0152] Table 3: Cooling energy consumption ratio calculation example

[0153]

[0154] As shown in Table 3, after cooling adjustment, the temperature change rate decreases, the cooling energy consumption reduces, and the calculated cooling energy consumption ratio is 0.000048 and is used for the calculation of the heat dissipation contribution degree, where the cooling energy consumption ratio weight W E and the heat dissipation contribution degree coefficient K C are set according to the cooling system evaluation criteria to ensure that the calculation results meet the cooling optimization requirements of the data center.

[0155] Please refer to Figure 5 , the cooling strategy sharing and optimization module includes:

[0156] Based on the server heat dissipation contribution degree, the server contribution degree weighting sub-module obtains the contribution degree data of each server in the cluster, calculates the contribution degree ratio of each server, normalizes the contribution degree value of each server, calculates the cooling strategy influence weight of each server, and obtains the server contribution degree weight value;

[0157] Based on the server heat dissipation contribution degree, the server contribution degree weighting sub-module obtains the contribution degree data of each server in the cluster. First, it calls the heat dissipation contribution degree C j and the load power consumption P j of all servers in the cluster. For example, if the cluster contains M = 4 servers, and their heat dissipation contribution degrees are [0.00000192, 0.00000210, 0.00000175, 0.00000225] respectively, and their load power consumptions are [300W, 320W, 280W, 350W], calculate the contribution degree ratio of each server:

[0158]

[0159] Calculated as:

[0160]

[0161] The contribution degree ratio of each server:

[0162]

[0163] Normalize the contribution degree value of each server, calculate the cooling strategy influence weight of the server, obtain the server contribution degree weight value, and store it in the database. The setting basis of the contribution degree weight value lies in the relative influence relationship between the server heat dissipation contribution degree and the load power consumption. This value fluctuates with the actual load of the server. The calculation method adopts the normalization method to ensure that the sum of the contribution degree weights of all servers is 1. The normalization adopts the maximum-minimum normalization method, that is

[0164]

[0165] Calculated based on the server contribution data:

[0166] max(C) = 0.00000225;

[0167] min(C) = 0.00000175;

[0168]

[0169] Obtain the normalized contribution weight values of each server to ensure that the values are within the range and not absolutely affected by the contribution value of a single server.

[0170] The shared policy master node screening sub-module calls the server contribution weight values, sets the contribution weight threshold, screens the servers higher than the threshold as the main reference nodes, and uses the formula:

[0171]

[0172] Calculate the influence degree of the reference nodes on the overall cooling policy to obtain the influence value of the shared policy main reference nodes;

[0173] Among them, W s represents the influence value of the shared policy main reference nodes, C j represents the heat dissipation contribution of the j-th server, P j represents the load power consumption of the j-th server, M represents the number of main reference servers selected, represents the average contribution of the selected servers;

[0174] The shared policy master node screening sub-module calls the server contribution weight values, sets the contribution weight threshold, screens the servers higher than the threshold as the main reference nodes. Suppose the contribution weight threshold is set to 0.25, and servers 2 (0.2619) and 4 (0.2803) are selected as the main reference nodes, and use the formula:

[0175]

[0176] Calculate:

[0177]

[0178] Calculate the contribution variance:

[0179]

[0180] Take the square root:

[0181]

[0182] Calculate the final influence value:

[0183]

[0184] Obtain the influence value W of the shared policy main reference node s = 2.02×10 -6 ;

[0185] The basis for setting the contribution degree weight threshold lies in the distribution characteristics of the heat dissipation contribution degree within the server cluster. Specifically, the mean plus standard deviation method of the overall contribution degree is used for setting, that is

[0186]

[0187] where σ is the standard deviation. According to the aforementioned calculation, the standard deviation σ = 5.96×10 -7 , finally

[0188] θ = 0.000002005 + 5.96×10 -7 = 0.000002601;

[0189] Select the servers with a contribution degree higher than this threshold as the main reference nodes.

[0190] The shared policy optimization calculation sub-module calls the influence value of the shared policy main reference node, calculates the shared policy influence matrix based on the contribution degree of the main reference node, adjusts the weights of the cooling policy sharing model, optimizes the calculation logic of the shared policy, and generates the optimized parameters of the shared policy;

[0191] The shared policy optimization calculation sub-module calls the influence value of the shared policy main reference node, calculates the shared policy influence matrix based on the contribution degree of the main reference node, and sets the element M of the influence matrix ij According to the influence relationship of server i on server j, such as:

[0192]

[0193] Adjust the weights of the cooling policy sharing model, adopt the optimized parameter matrix W, and set:

[0194]

[0195] The weight matrix W in the shared policy optimization calculation is obtained by normalizing the calculation based on the contribution degree of the main reference node. Each weight value ensures that the row and column sums are 1 and is dynamically adjusted with the contribution degree of the main reference node. The specific values are derived from the influence ratio of the main node on the overall cooling policy.

[0196] Please refer to Figure 6 , the cluster thermal management federated calculation module includes:

[0197] The cooling execution data statistics sub-module optimizes parameters based on the sharing policy, obtains the cooling execution data of each server in the server cluster, records the coolant flow rate, heat exchange plate temperature gradient, server power consumption, and temperature drop rate during the execution of the cooling policy for each server, and establishes a server cooling execution data set;

[0198] The cooling execution data statistics sub-module optimizes parameters based on the sharing policy, obtains the cooling execution data of each server in the server cluster. First, it collects the coolant flow rate, heat exchange plate temperature gradient, server power consumption, and temperature drop rate of each server in the server cluster, and sets a sampling interval of every 5s to ensure the continuity and accuracy of the data. For example, the coolant flow rate record of a certain server within 60s is [2.5, 2.6, 2.4, 2.7, 2.5] m / s, the heat exchange plate temperature gradient record is [12, 13, 11, 14, 12] K, the server power consumption record is [320, 315, 318, 322, 319] W, and the temperature drop rate record is [-0.10, -0.12, -0.08, -0.14, -0.11] °C / s. It integrates the data of all servers to establish a server cooling execution data set and stores it in the database. The sampling period of the coolant flow rate is set according to the server heat dissipation contribution weight value, which is calculated from the heat dissipation contribution of each server and transformed into a relative weight in the range of 0 - 1 through normalization calculation, and then used for the allocation of the coolant flow rate. Set the server heat dissipation contribution weight value W C The calculation formula is as follows:

[0199]

[0200] Set the server heat dissipation contributions in the cluster to be

[0201] [0.00000192, 0.00000210, 0.00000175, 0.00000225], and calculate its normalized weight:

[0202]

[0203] This weight value determines the server cooling policy allocation. For example, server 4 has the highest contribution, and its coolant flow rate allocation weight is higher than that of other servers, which is ultimately used to optimize the allocation of cooling resources.

[0204] The federated computing matrix construction sub-module calls the server cooling execution data set to construct a cooling parameter federated learning computing matrix, using the formula;

[0205]

[0206] Calculate the server cooling adaptability;

[0207] Among them, Ad Represents the server cooling adaptation degree, N A Represents the total number of servers participating in the calculation in the server cluster, P A,f Represents the power consumption of the f-th server, T A,f Represents the temperature drop rate of server f Represents the mean value of the power consumption in the server cluster, μ TA Represents the mean value of the temperature drop rate in the server cluster Represents the standard deviation of the power consumption in the server cluster Represents the standard deviation of the temperature drop rate in the server cluster, where e is the base of the natural logarithm

[0208] The federated computing matrix construction sub-module calls the server cooling execution data set to construct a cooling parameter federated learning computing matrix. First, calculate the mean value of the power consumption in the server cluster And the standard deviation Suppose there are N A = 4 servers in the cluster, and their power consumption records are [320, 310, 330, 325] W, and the temperature drop rate records are [-0.10, -0.09, -0.12, -0.11] °C / s, then:

[0209]

[0210] Calculate the mean value μ of the temperature drop rate TA And the standard deviation σ TA :

[0211]

[0212] Substitute into the formula:

[0213]

[0214] Calculate one by one:

[0215]

[0216]

[0217] Similarly, calculate the values of other servers and find the mean value. Finally, calculate the server cooling adaptation degree A d = 0.689. The mean value of the power consumption And the standard deviation As the reference value, it represents the fluctuation range of the server power consumption. Its standard deviation reflects the degree of dispersion of the power consumption distribution. A higher Indicates that the power consumption differences among the servers in the cluster are relatively large, and a lower Indicates that the power consumption tends to be stable. This reference value is used to measure the rationality of the cooling adaptation degree calculation

[0218] The shared parameter optimization sub-module calls the server cooling adaptability, adjusts the shared parameters of the server cooling strategy, sets the optimization rules for the cluster cooling strategy, and obtains the server cluster cooling control information;

[0219] The shared parameter optimization sub-module calls the server cooling adaptability, adjusts the shared parameters of the server cooling strategy, sets the optimization rules for the cluster cooling strategy, and sets the shared parameter matrix S of the server cooling strategy according to the weight of each server cooling adaptability:

[0220]

[0221] The adjusted cooling strategy control parameter S′ is calculated as follows:

[0222]

[0223] The weight values in the shared parameter matrix S are calculated based on the server cooling adaptability, and its optimization rules are based on the server cooling adaptability A d is scaled, and the scaling factor A d = 0.689 is used as a measure of cooling adaptability to proportionally adjust the weight matrix to reasonably allocate cooling resources. Finally, the server cluster cooling control information is obtained and stored in the database.

[0224] The AI-based multi-modal sensor network cooling management method includes the following steps:

[0225] S1: Obtain multi-modal sensor network data, extract the task calculation cycle and the power consumption accumulation per unit time of the server calculation task, collect the chip temperature change trend, calculate the task heat accumulation rate, and calculate the task heat load information;

[0226] S2: The fluid inertia compensation regulation module calculates the fluid kinetic energy and the flow inertia influence amount based on the task heat load information, and adjusts the coolant flow rate according to the task category to obtain the optimized cooling flow rate value;

[0227] S3: The heat dissipation contribution evaluation module calculates the temperature drop rate per unit time based on the optimized cooling flow rate value, and calculates the server heat dissipation contribution in combination with the energy consumption change before and after adjustment;

[0228] S4: The cooling strategy sharing optimization module adjusts the weights of the server cooling strategy sharing model based on the server heat dissipation contribution, selects the servers with high contribution as the main reference nodes, and adjusts the shared strategy calculation logic to obtain the optimized shared strategy parameters;

[0229] S5: The cluster thermal management federated computing module optimizes parameters based on the shared policy, constructs a cooling parameter federated learning computing matrix, calculates the adaptability of the server cooling execution parameters within the cluster, adjusts the shared parameters of the cooling policy, and obtains the server cluster cooling control information.

[0230] The above are only the preferred embodiments of the present invention, and do not limit the present invention in other forms. Any person skilled in the art may use the disclosed technical content to make changes or modifications into equivalent embodiments with equivalent changes and apply them to other fields. However, as long as it does not depart from the technical solution content of the present invention, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention still fall within the protection scope of the technical solution of the present invention.

Claims

1. AI-based multimodal sensor network cooling management system, characterized by: The system comprises: The task heat load classification module obtains multimodal sensor network data, extracts the task calculation cycle and the accumulated power consumption per unit time of the server computing task, collects the chip temperature change trend, calculates the task heat accumulation rate, and calculates the task heat load information; The fluid inertia compensation control module calculates the fluid kinetic energy and flow inertia influence based on the task heat load information, adjusts the coolant flow rate according to the task category, and obtains the cooling flow rate optimization value; The heat dissipation contribution evaluation module calculates the temperature drop rate per unit time based on the cooling flow rate optimization value, and calculates the heat dissipation contribution of the server in combination with the energy consumption change before and after the adjustment; The cooling strategy sharing optimization module adjusts the weight of the server cooling strategy sharing model based on the server heat dissipation contribution, selects servers with high contribution as main reference nodes, adjusts the sharing strategy calculation logic, and obtains sharing strategy optimization parameters; The cluster thermal management federated computing module optimizes the parameters based on the shared strategy, constructs a cooling parameter federated learning computing matrix, calculates the adaptability of the server cooling execution parameters within the cluster, adjusts the cooling strategy shared parameters, and obtains the server cluster cooling control information.

2. The AI-based multimodal sensor network cooling management system according to claim 1, characterized in that: The task thermal load information specifically includes the calculation cycle heat accumulation ratio, the accumulated power consumption per unit time and the chip temperature change rate. The cooling flow rate optimization value includes the fluid kinetic energy adjustment coefficient, the flow inertia compensation amount and the coolant target flow rate. The server heat dissipation contribution specifically refers to the temperature drop rate per unit time, the cooling energy consumption ratio and the energy consumption change before and after the heat dissipation adjustment. The sharing strategy optimization parameters include the contribution weight adjustment value, the main reference node screening parameters and the sharing strategy calculation correction items. The server cluster cooling control information specifically includes the cooling parameter federation calculation matrix, the server cooling adaptability and the cooling strategy sharing parameters.

3. The AI-based multimodal sensor network cooling management system according to claim 1, characterized in that: The task heat load classification module includes: The task computing feature extraction submodule obtains multimodal sensor network data, extracts the task computing cycle of the server computing task, calculates the cumulative power consumption per unit time, monitors the chip temperature change trend, calculates the chip temperature change rate based on the time series data, and obtains the task computing feature data set; The task heat accumulation rate calculation submodule calls the task calculation feature data set and uses the formula: Calculate and obtain the task heat accumulation rate to obtain the task heat accumulation rate value; Among them, R h represents the mission heat accumulation rate, E tot Represents the cumulative power consumption per unit time, T comp Represents the task calculation period, T chip,i represents the chip temperature at the i-th moment, Δt represents the temperature sampling interval, and n represents the length of the time series; The task classification determination submodule calls the task heat accumulation rate value, sets the classification threshold, determines the task type according to the ratio, divides it into short-term high-heat tasks, long-term medium-heat tasks, and low-heat tasks, and establishes task heat load information.

4. The AI-based multimodal sensor network cooling management system according to claim 1, characterized in that: The fluid inertia compensation control module includes: The fluid kinetic energy calculation submodule, based on the task heat load information, calls the server's current coolant flow rate and the specific heat capacity per unit mass of the fluid to calculate the specific kinetic energy per unit mass of the fluid, calls the temperature gradient of the heat exchange plate to calculate the energy change during the coolant temperature rise process, and calculates the fluid kinetic energy in combination with the task heat load coefficient; The inertia influence analysis submodule calls the fluid kinetic energy, calculates the flow inertia of the coolant in the pipeline, calculates the momentum change rate of the fluid according to the coolant density, flow velocity and fluid cross-sectional area, and combines the temperature gradient of the heat exchange plate to use the formula: Calculate the influence of flow inertia on the heat transfer process and obtain the flow inertia influence amount; Where I represents the flow inertia influence, m represents the coolant mass per unit volume, v represents the coolant flow rate, ρ represents the coolant density, A represents the coolant pipe cross-sectional area, ΔT p represents the temperature gradient of the heat exchange plate, c p represents the specific heat capacity of the coolant; The cooling flow rate control submodule calls the flow inertia influence amount and adjusts the coolant flow rate according to the task category, including increasing the coolant flow rate for short-term high-heat task servers, periodically adjusting the coolant temperature for long-term medium-heat task servers, and reducing the flow rate for low-heat task servers to obtain the cooling flow rate optimization value.

5. The AI-based multimodal sensor network cooling management system according to claim 1, characterized in that: The heat dissipation contribution evaluation module includes: The temperature change rate calculation submodule obtains the chip temperature data before and after the server cooling system is adjusted based on the cooling flow rate optimization value, calculates the temperature change rate before and after the adjustment, and obtains the chip temperature change rate; The cooling energy consumption ratio calculation submodule calls the chip temperature change rate to obtain the server power consumption data before and after the server cooling system is adjusted, using the formula: Calculate the ratio of the temperature drop rate per unit time to the cooling energy consumption to obtain the cooling energy consumption ratio; Among them, R E Represents the cooling energy consumption ratio, ΔT t represents the rate of temperature change at time t, E t represents the cooling energy consumption at time t, N represents the total number of data samples, max(ΔT t ) represents the maximum temperature drop rate during the sampling period, max(E t ) represents the maximum cooling energy consumption during the sampling period; The heat dissipation contribution calculation submodule calls the cooling energy consumption ratio to obtain the energy consumption data before and after the server heat dissipation adjustment, and uses the formula according to the energy consumption reduction after adjustment: Calculate the heat dissipation contribution of the server; Among them, C represents the server heat dissipation contribution, Q before represents the energy consumption of the cooling system before adjustment, Q after represents the energy consumption of the cooling system after adjustment, R E Represents the cooling energy consumption ratio.

6. The AI-based multimodal sensor network cooling management system according to claim 1, characterized in that: The cooling strategy sharing optimization module includes: The server contribution weighting submodule obtains the contribution data of each server in the cluster based on the server heat dissipation contribution, calculates the contribution ratio of each server, normalizes the contribution value of each server, calculates the cooling strategy influence weight of each server, and obtains the server contribution weight value; The shared strategy master node screening submodule calls the server contribution weight value, sets the contribution weight threshold, and screens servers above the threshold as the main reference nodes, using the formula: Calculate the influence of the reference node on the overall cooling strategy and obtain the influence value of the main reference node of the shared strategy; Among them, W s represents the impact value of the main reference node of the shared strategy, C j represents the cooling contribution of the jth server, P j represents the load power consumption of the jth server, M represents the number of selected main reference servers, Represents the average contribution of the filtering server; The sharing strategy optimization calculation submodule calls the influence value of the main reference node of the sharing strategy, calculates the sharing strategy influence matrix according to the contribution of the main reference node, adjusts the weight of the cooling strategy sharing model, optimizes the sharing strategy calculation logic, and generates sharing strategy optimization parameters.

7. The AI-based multimodal sensor network cooling management system according to claim 1, characterized in that: The cluster thermal management federated computing module includes: The cooling execution data statistics submodule obtains the cooling execution data of each server in the server cluster based on the shared strategy optimization parameters, records the coolant flow rate, heat exchange plate temperature gradient, server power consumption and temperature drop rate during the execution of the cooling strategy of each server, and establishes a server cooling execution data set; The federated computing matrix construction submodule calls the server cooling execution data set to construct a cooling parameter federated learning computing matrix using the formula; Calculate server cooling suitability; Among them, A d Represents the server cooling adaptability, N A represents the total number of servers participating in the calculation in the server cluster, P A,f represents the power consumption of the fth server, T A,f represents the temperature drop rate of server f, Represents the average power consumption in the server cluster, Represents the average temperature drop rate in the server cluster, represents the standard deviation of power consumption in the server cluster, Represents the standard deviation of the temperature drop rate in the server cluster, and e is the base of the natural logarithm; The shared parameter optimization submodule calls the server cooling adaptability, adjusts the server cooling strategy shared parameters, sets cluster cooling strategy optimization rules, and obtains server cluster cooling control information.

8. The AI-based multimodal sensor network cooling management method is characterized by: The AI-based multimodal sensor network cooling management system according to any one of claims 1 to 7 comprises the following steps: S1: Obtain multimodal sensor network data, extract the task calculation cycle and power consumption accumulation per unit time of the server computing task, calculate the task heat accumulation rate, and calculate the task heat load information; S2: The fluid inertia compensation control module calculates the fluid kinetic energy and flow inertia influence based on the task heat load information, adjusts the coolant flow rate, and obtains the cooling flow rate optimization value; S3: The heat dissipation contribution evaluation module calculates the temperature drop rate per unit time based on the cooling flow rate optimization value, and calculates the heat dissipation contribution of the server in combination with the energy consumption change before and after the adjustment; S4: The cooling strategy sharing optimization module selects servers with high contribution as main reference nodes based on the heat dissipation contribution of the servers, adjusts the sharing strategy calculation logic, and obtains sharing strategy optimization parameters; S5: The cluster thermal management federated computing module optimizes the parameters based on the shared strategy, constructs a cooling parameter federated learning computing matrix, calculates the adaptability of the server cooling execution parameters within the cluster, adjusts the cooling strategy shared parameters, and obtains the server cluster cooling control information.

Images