A temperature regulation method and system based on power consumption prediction

Abstract

The application discloses a temperature regulation method and system based on power consumption prediction, relates to the technical field of device heat dissipation control, and comprises a power consumption detection module, an AI prediction module and a device control module; the power consumption detection module collects power consumption data of a target device in real time; the AI prediction module receives the power consumption data, predicts the temperature change trend of the target device at a future preset time point based on a temperature prediction model according to the power consumption change trend; the device control module adjusts the operating parameters of a heat dissipation device in advance according to the temperature prediction result before the temperature actually rises; the application further comprises an alarm and early warning module which gives an early warning when the predicted temperature exceeds a threshold value; the power consumption data comprises current, voltage and power parameters, and the sampling frequency is not lower than 10 Hz; the temperature prediction model is a deep learning model which is trained by historical power consumption and temperature data and is based on the power consumption change amount; the application can overcome the inherent delay of traditional feedback control and improve the active protection capability and control quality of a heat dissipation system.

Description

Technical Field

[0001] This invention relates to the field of equipment heat dissipation control technology, and in particular to a temperature regulation method and system based on power consumption prediction. Background Technology

[0002] With the increasing integration of electronic devices and the continuous improvement of computing performance, the power density of core heat-generating components inside these devices has increased significantly. Thermal management has become a key technical aspect for ensuring stable system operation and extending service life. Currently, mainstream temperature control strategies generally adopt a feedback regulation mode based on real-time monitoring. This involves collecting temperature signals through temperature sensors placed near heat sources or key system nodes. When the detected value exceeds a preset temperature threshold, the operating level of heat dissipation equipment (such as fans, air conditioning circulation systems, or liquid cooling pipelines) is activated or increased to achieve cooling. This solution is simple in structure and easy to implement, and therefore widely used in the thermal design of various existing electronic devices.

[0003] However, the aforementioned "detection-feedback" control mechanism is inherently limited by the hysteresis characteristic of the physical process of heat conduction. Specifically, after heat is generated from the core heat-generating unit (such as a central processing unit chip or a graphics processing unit chip), it must pass through multiple stages of heat conduction paths, including chip packaging, thermal interface materials, heat sink substrate, and air / coolant, before reaching the installation location of the temperature sensor. This transfer process inevitably introduces a delay of several seconds or even tens of seconds. During this delay, if the device's workload experiences a sudden increase, the core chip temperature will rise rapidly due to the instantaneous power surge. However, due to the hysteresis of the sensor feedback signal, the cooling system cannot respond synchronously. By the time the temperature information is sensed, the device is already in a short-term overheating state. At this time, the cooling system is forced to operate at full load power to suppress the temperature rise. This not only leads to a sharp increase in system energy consumption but also easily causes large fluctuations in device temperature, adversely affecting the thermomechanical reliability of chip solder joints, substrate materials, and surrounding components.

[0004] Furthermore, existing solutions rely solely on passive decision-making based on current or historical measured temperatures, lacking the ability to anticipate and predict impending load changes and their resulting temperature rise trends. Consequently, they cannot proactively implement control measures before temperatures exceed limits. This deficiency is particularly pronounced in applications such as high-performance computing servers, data center clusters, and aerospace electronic equipment where load dynamics change frequently and reliability requirements are stringent, making it difficult to meet the system's comprehensive demands for real-time heat dissipation response and stable temperature control. To address these issues, it is necessary to propose a temperature control strategy that can predict the direction of temperature evolution based on device power consumption trends and adjust heat dissipation capacity accordingly. This would overcome the inherent delay of traditional feedback control and improve the active protection capability and control quality of the heat dissipation system. Summary of the Invention

[0005] In order to at least solve one of the above-mentioned technical problems, the present invention aims to provide a temperature control method and system based on power consumption prediction, which overcomes the inherent delay of traditional feedback control and improves the active protection capability and control quality of the heat dissipation system.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] A temperature control system based on power consumption prediction includes a power consumption detection module, an AI prediction module, and a device control module;

[0008] The power consumption detection module is used to collect power consumption data of the target device in real time;

[0009] The AI ​​prediction module is connected to the power consumption detection module and is used to receive the power consumption data and predict the temperature change trend of the target device at a future preset time point based on the pre-trained temperature prediction model and the power consumption change trend.

[0010] The device control module is connected to the AI ​​prediction module and is used to send control commands to the heat dissipation device in advance, based on the temperature prediction results output by the AI ​​prediction module, before the actual temperature of the target device rises, and to adjust the operating parameters of the heat dissipation device.

[0011] Preferably, it also includes an alarm and early warning module connected to the AI ​​prediction module, used to issue an early warning signal when the predicted temperature is about to exceed a preset safety threshold.

[0012] Preferably, the power consumption data collected by the power consumption detection module includes current, voltage, and power parameters.

[0013] Preferably, the sampling frequency of the power consumption detection module is not less than 10Hz.

[0014] Preferably, the temperature prediction model is a deep learning model, trained using historical power consumption data and corresponding historical temperature data as training samples.

[0015] Preferably, the temperature prediction model performs temperature prediction calculations based on the power consumption change, the power consumption change rate, and the thermal capacity parameters of the target device.

[0016] A temperature control method based on power consumption prediction includes the following steps:

[0017] S1: Real-time acquisition of power consumption data of the target device;

[0018] S2: Input the power consumption data into a pre-trained AI temperature prediction model to predict the temperature change trend of the target device at a future preset time point;

[0019] S3: Based on the temperature prediction results, adjust the heat dissipation equipment in advance before the actual temperature of the target equipment rises.

[0020] Preferably, in S1, the power consumption data collected includes current, voltage and power parameters, and the sampling frequency is not less than 10Hz.

[0021] Preferably, the AI ​​temperature prediction model is a deep learning model, which is trained using historical power consumption data and corresponding historical temperature data as training samples, power consumption data as input features, and future temperature values ​​as prediction targets.

[0022] Preferably, the AI ​​temperature prediction model performs temperature prediction calculations based on the amount of power consumption change, the rate of power consumption change, and the thermal capacity parameters of the target device.

[0023] The present invention has the following beneficial effects:

[0024] I. Eliminating the Lag in Heat Dissipation Control: The core benefit of this invention lies in overcoming the lag defect of traditional temperature control modes at the principle level. By setting a power consumption detection module to collect the current, voltage, and power parameters of the target device in real time, and using an AI prediction module to predict the temperature change trend at a preset time point based on the power consumption change trend, the control basis is transformed from passive temperature measurement values ​​to active power consumption trend prediction. The device control module can send control commands to the heat dissipation device in advance before the actual temperature of the target device rises, thereby establishing a synchronous matching relationship between heat generation and heat dissipation, fundamentally avoiding short-term overheating caused by the delay in the heat conduction path, and matching the heat dissipation response speed with the load change rate.

[0025] II. Reducing the operating energy consumption of the heat dissipation system; This invention effectively reduces the overall energy consumption of the system by adjusting the heat dissipation equipment in advance. Because the temperature prediction results provide advance notice of future temperature rise, the heat dissipation equipment can adjust its operating parameters gradually and smoothly, without being forced to operate at maximum power during sudden temperature increases. This advance intervention control strategy eliminates the sudden start-up and shutdown of the heat dissipation equipment, making the operation of the heat dissipation system more stable and efficient, thereby significantly reducing the additional energy loss caused by emergency full-speed operation.

[0026] Third, it reduces the fluctuation range of equipment temperature; this invention can effectively suppress drastic temperature oscillations in the target equipment and improve the stability of temperature control. The equipment control module adjusts the heat dissipation capacity in advance based on the predicted temperature change trend, so that the heat dissipation action is synchronized with the heat generation trend, avoiding the process of the temperature rising sharply and then being pulled down forcefully. This smooth regulation process maintains the equipment temperature within a relatively stable range, reducing the thermomechanical stress impact on chip solder joints, substrate materials and surrounding components caused by repeated thermal expansion and contraction, which has a direct positive significance for extending the service life of the equipment.

[0027] Fourth, it possesses proactive safety early warning capabilities; this invention integrates an alarm and early warning module on the basis of a predictive mechanism, constructing a proactive safety protection capability. The alarm and early warning module continuously receives temperature prediction results output by the AI ​​prediction module. When the predicted temperature is about to exceed a preset safety threshold, it issues an early warning signal, enabling maintenance personnel to have sufficient emergency response time before the temperature actually exceeds the limit. This function shifts the trigger point for safety protection from post-event alarm to pre-event trend warning, effectively reducing the risk of equipment failure or damage due to overheating, and has significant value in high-reliability scenarios such as servers and data centers.

[0028] V. Strong adaptability to complex power consumption changes; This invention employs a deep learning model as the temperature prediction model, using historical power consumption data and corresponding historical temperature data as training samples. It can learn the complex nonlinear mapping relationship between power consumption changes and temperature response from a large amount of operational data. Compared to methods relying on simplified thermodynamic formulas, this model can more accurately capture the unique thermal dynamic characteristics of the target equipment under different load conditions and operating states, making the temperature prediction results closer to reality. This ensures reliable prediction accuracy and control effectiveness in application scenarios with frequent dynamic load changes.

[0029] VI. Overall Synergistic Effect of Solution Implementation; This invention, through the organic combination of a power consumption detection module, an AI prediction module, an equipment control module, and an alarm and early warning module, forms a complete closed-loop link from data acquisition to trend prediction, and then to early control and early warning. The high-frequency sampling data provided by the power consumption detection module ensures the timeliness of trend analysis, while the predictive information provided by the AI ​​prediction module provides a unified and forward-looking decision-making basis for control and early warning. The collaborative operation of each module jointly constructs an intelligent temperature control architecture with the ability to proactively perceive and intervene in the thermal state of equipment, achieving an overall improvement in the real-time performance, accuracy, stability, and safety of heat dissipation control. Attached Figure Description

[0030] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0031] Figure 1 This is a schematic diagram of the component connection relationship in an embodiment of the present invention. Detailed Implementation

[0032] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0033] like Figure 1 As shown, a temperature control system based on power consumption prediction includes a power consumption detection module, an AI prediction module, and a device control module; the power consumption detection module is used to collect power consumption data of the target device in real time.

[0034] The AI ​​prediction module, connected to the power consumption detection module, is used to receive power consumption data and, based on a pre-trained temperature prediction model, predict the temperature change trend of the target device at a future preset time point according to the power consumption change trend.

[0035] The equipment control module, connected to the AI ​​prediction module, is used to send control commands to the heat dissipation equipment in advance, based on the temperature prediction results output by the AI ​​prediction module, before the actual temperature of the target equipment rises, and to adjust the operating parameters of the heat dissipation equipment.

[0036] like Figure 1 As shown, it also includes an alarm and early warning module, which is connected to the AI ​​prediction module and is used to issue an early warning signal when the predicted temperature is about to exceed the preset safety threshold.

[0037] like Figure 1 As shown, this embodiment establishes a predictive mapping relationship between power consumption data and future temperature changes, using power consumption trends as a leading indicator for temperature control, replacing the traditional approach of passively waiting for actual temperature measurements, thereby enabling advance control of the heat dissipation equipment. The following sections describe the technical features in segments.

[0038] The power consumption detection module collects power consumption data of the target device in real time. Its principle is to use electrical parameter monitoring as a direct means of sensing the heating intensity of the heat source. The power consumption change of the heating element is instantly reflected in the current, voltage and power parameters. This process is almost without delay, thus providing the system with the earliest heating status information, bypassing the time required for heat conduction to the temperature sensor, and eliminating feedback time lag at the source.

[0039] The AI ​​prediction module receives power consumption data and, based on a pre-trained temperature prediction model, predicts the temperature change trend of the target device at a preset future time point based on the power consumption change trend. The principle behind this step is to use a deep learning model to learn the complex nonlinear relationship between power consumption changes and temperature response during the target device's historical operation. The model comprehensively considers the amount of power consumption change, the rate of power consumption change, and the device's own thermal capacity parameters, transforming the current real-time power consumption sequence into a predicted temperature value for the future. This process is equivalent to calculating and deducing the trajectory of the temperature before it actually changes.

[0040] Based on the temperature prediction results output by the AI ​​prediction module, the equipment control module sends control commands to the heat dissipation equipment in advance, adjusting its operating parameters before the actual temperature rise of the target equipment. The principle is to shift the timing of the heat dissipation system's action from when the temperature exceeds the limit to when the prediction indicates an impending temperature rise, ensuring that the deployment of heat dissipation capacity begins synchronously with the increase in heat generation, thus avoiding a drastic temperature spike. The adjustment process can employ a gradual approach, adjusting fan speed or cooling medium flow rate for smoother operation of the heat dissipation equipment.

[0041] The alarm and early warning module is connected to the AI ​​prediction module. When the temperature is predicted to exceed the preset safety threshold, an early warning signal is issued. The principle is to continuously compare the predicted temperature value with the safety limit. Once the prediction curve shows that the future temperature will exceed the safety boundary, an alarm is issued at the current moment. This changes the trigger point for safety protection from a reactive alarm to a proactive warning, giving maintenance personnel time to intervene in emergencies.

[0042] The power consumption detection module collects power consumption data including current, voltage, and power parameters. Its principle lies in comprehensively characterizing the real-time power consumption state of the target device through synchronous monitoring of multi-dimensional electrical parameters. The real-time product of current and voltage yields the instantaneous power. These three parameters together constitute the basic elements of power consumption data, enabling the system to capture changes in heat intensity from the source, providing direct and lag-free input for subsequent trend analysis and temperature prediction.

[0043] The power consumption detection module has a sampling frequency of no less than 10Hz. Its principle is to sample power consumption data at a rate of no less than ten times per second, ensuring the capture of rapid transient changes in power consumption. This sampling frequency allows for the complete recording of even sub-second power consumption pulses or sudden load surges in the target device. This provides the AI ​​prediction module with sufficiently dense data points to identify power consumption trends, avoiding the omission of crucial change details due to excessively large sampling intervals, thus ensuring real-time predictive response capabilities.

[0044] The temperature prediction model employs a deep learning model, trained using historical power consumption data and corresponding historical temperature data as training samples. Its principle lies in leveraging the powerful fitting ability of deep learning networks to complex nonlinear mapping relationships, automatically learning the correspondence between the target device's unique power consumption variation patterns and temperature response from a large amount of offline historical data. Compared to manually setting simplified thermodynamic formulas with fixed parameters, this method can more accurately capture the temperature evolution pattern formed by the combined effects of power consumption changes, change rates, and the device's own heat capacity parameters under different operating conditions. This allows the model to output more realistic temperature prediction results when faced with real-time power consumption input after deployment.

[0045] The temperature prediction model calculates temperature forecasts based on changes in power consumption, the rate of change in power consumption, and the heat capacity parameters of the target device. Its principle lies in using these physical quantities as key decision factors in the model's inference process. Changes in power consumption reflect the absolute increase or decrease in heat intensity, the rate of change in power consumption reflects the severity of sudden load changes, and the heat capacity parameter determines the temperature rise of the target device after absorbing the same amount of heat. These three factors work together to enable the model to consider both the dynamic characteristics of the input and the physical and thermal properties of the device when calculating future temperature trends, thus arriving at predictions that conform to fundamental thermodynamic laws. This mechanism effectively improves the predictive model's generalization ability and physical plausibility when facing unseen load patterns.

[0046] The working principle and effects of the technical solution you provided are explained below:

[0047] The technical solution first collects power consumption data such as current, voltage, and power of the target device in real time, sampling at a high frequency of no less than 10 Hz. This ensures that the system can capture subtle instantaneous changes in power consumption, providing a sufficient and accurate data foundation for subsequent predictions. The effect is that high-resolution power consumption data is a prerequisite for temperature trend prediction, effectively avoiding prediction delays or biases caused by data sparsity.

[0048] Next, the scheme inputs the collected power consumption data into a pre-trained AI temperature prediction model. This model is a deep learning model, whose training process uses historical power consumption data as input features and the actual temperature value at the corresponding future time as the prediction target, enabling the model to learn the complex mapping relationship between power consumption changes and temperature response. In actual prediction, the model focuses on calculating based on the current amount and rate of change of power consumption, as well as the inherent heat capacity parameters of the device, thereby predicting the temperature change trend of the target device at a specific future time. The effect of this process is that the deep learning model can handle nonlinear relationships and integrate the influence of multiple factors. Compared with traditional threshold judgment, it can identify early signs of an impending temperature rise earlier and more accurately, achieving true advanced prediction.

[0049] Finally, based on the above predictions, the system sends control commands to the heat dissipation equipment in advance, before the target device's temperature actually rises. The core effect of this feedforward control method is that it completely changes the passive and lagging characteristics of traditional temperature feedback control; by preempting the time difference in temperature rise, it enables the heat dissipation equipment to respond and adjust its operating parameters in advance, thereby effectively suppressing the temperature peak within a preset safe range. This achieves a shift from passive cooling to active temperature management, significantly improving the stability and response efficiency of the system's temperature control.

[0050] The above are merely specific embodiments of the present invention, but the technical features of the present invention are not limited thereto. Any simple changes, equivalent substitutions, or modifications made based on the present invention to solve essentially the same technical problems and achieve essentially the same technical effects are all covered within the protection scope of the present invention.

Claims

1. A temperature control system based on power consumption prediction, characterized in that, It includes a power consumption detection module, an AI prediction module, and a device control module; The power consumption detection module is used to collect power consumption data of the target device in real time; The AI ​​prediction module is connected to the power consumption detection module and is used to receive the power consumption data and predict the temperature change trend of the target device at a future preset time point based on the pre-trained temperature prediction model and the power consumption change trend. The device control module is connected to the AI ​​prediction module and is used to send control commands to the heat dissipation device in advance, based on the temperature prediction results output by the AI ​​prediction module, before the actual temperature of the target device rises, to adjust the operating parameters of the heat dissipation device.

2. The system according to claim 1, characterized in that, It also includes an alarm and early warning module, which is connected to the AI ​​prediction module and is used to issue an early warning signal when the predicted temperature is about to exceed a preset safety threshold.

3. The system according to claim 1, characterized in that, The power consumption data collected by the power consumption detection module includes current, voltage, and power parameters.

4. The system according to claim 1, characterized in that, The sampling frequency of the power consumption detection module is not less than 10Hz.

5. The system according to claim 1, characterized in that, The temperature prediction model is a deep learning model, trained using historical power consumption data and corresponding historical temperature data as training samples.

6. The system according to claim 5, characterized in that, The temperature prediction model calculates the temperature based on the change in power consumption, the rate of change in power consumption, and the thermal capacity parameters of the target device.

7. A temperature control method based on power consumption prediction, based on the temperature control system based on power consumption prediction as described in any one of claims 1 to 6, characterized in that, Includes the following steps: S1: Real-time acquisition of power consumption data of the target device; S2: Input the power consumption data into a pre-trained AI temperature prediction model to predict the temperature change trend of the target device at a future preset time point; S3: Based on the temperature prediction results, adjust the heat dissipation equipment in advance before the actual temperature of the target equipment rises.

8. The temperature control method according to claim 7, characterized in that, In S1, the collected power consumption data includes current, voltage, and power parameters, and the sampling frequency is not less than 10Hz.

9. The temperature control method according to claim 7, characterized in that, The AI ​​temperature prediction model is a deep learning model, which is trained using historical power consumption data and corresponding historical temperature data as training samples, power consumption data as input features, and future temperature values ​​as prediction targets.

10. The temperature control method according to claim 9, characterized in that, The AI ​​temperature prediction model calculates temperature prediction based on the amount of power consumption change, the rate of power consumption change, and the thermal capacity parameters of the target device.

Images