Self-adaptive control method and system for energy consumption optimization of computing power equipment and program product

Abstract

The invention relates to the technical field of energy consumption control, and discloses a computing power equipment energy consumption optimization self-adaptive control method and system and a program product, and the method comprises the steps: obtaining an initial operation environment parameter set of computing power equipment; based on the initial operation environment parameter set, calculating through a data fusion algorithm to obtain a target environment evaluation value; based on the initial operation environment parameter set and the target environment evaluation value, determining a temperature control mode and an anti-interference protection mode of the computing power equipment; controlling a temperature adjusting device to adjust the temperature of the computing power equipment by using the temperature control mode to obtain a temperature adjusting result of the computing power equipment; an anti-interference protection mode is utilized to control the protection device to adjust external disturbance of the computing power equipment, and an interference adjustment result of the computing power equipment is obtained; and controlling the energy consumption and redundancy management module to adjust the energy consumption and redundancy of the computing power equipment by using the temperature adjustment result and the interference adjustment result to obtain an energy consumption optimization self-adaptive control result of the computing power equipment, thereby realizing self-adaptive energy consumption optimization.

Description

Technical Field

[0001] This invention relates to the field of energy consumption control technology, specifically to a method, system, and program product for adaptive control of energy consumption optimization of computing equipment. Background Technology

[0002] With the rapid development of artificial intelligence, big data, and high-performance computing technologies, computing power equipment has become a crucial infrastructure driving intelligent manufacturing, autonomous driving, smart energy, rail transportation, and national defense security. AI computing power equipment typically integrates multiple high-power AI chips, accelerator cards, and high-speed storage units. During operation, it generates a significant amount of heat, and its compact internal structure and high heat density place extremely high demands on temperature control stability and response speed. However, in complex scenarios such as industrial, field, and special operations, the environment in which the equipment operates often experiences drastic temperature changes, high humidity, air pressure fluctuations, high dust density, frequent vibrations, and strong electromagnetic interference, making it difficult for traditional heat dissipation and temperature control systems to guarantee long-term stable operation.

[0003] The existing technology has the following main problems: it lacks a multi-source environmental perception and comprehensive judgment mechanism, making it unable to accurately identify complex working conditions; heat dissipation and heating control are driven by fixed thresholds, resulting in slow response speed and inaccurate adjustment; the system has high energy consumption and lacks intelligent energy efficiency optimization and redundancy protection; and it is not adequately protected against non-temperature risks such as vibration and electromagnetic interference, affecting the reliability and safety of the equipment. Summary of the Invention

[0004] This invention provides a method, system, and program product for optimizing the energy consumption of computing devices to solve the problem that existing technologies make it difficult for traditional heat dissipation and temperature control systems to ensure long-term stable operation.

[0005] In a first aspect, the present invention provides an adaptive control method for optimizing energy consumption in computing devices, for use in a main control unit, wherein the main control unit is connected to a temperature regulation device, a protection device, and an energy consumption and redundancy management module; the method includes: The process involves: acquiring the initial operating environment parameter set of the computing power equipment; calculating the target environment evaluation value based on the initial operating environment parameter set using a data fusion algorithm; determining the temperature control mode and anti-interference protection mode of the computing power equipment based on the initial operating environment parameter set and the target environment evaluation value; using the temperature control mode, controlling the temperature regulation device to adjust the temperature of the computing power equipment, and obtaining the temperature regulation result of the computing power equipment; using the anti-interference protection mode, controlling the protection device to adjust the external disturbances of the computing power equipment, and obtaining the interference regulation result of the computing power equipment; and using the temperature regulation result and the interference regulation result, controlling the energy consumption and redundancy management module to adjust the energy consumption and redundancy of the computing power equipment, and obtaining the energy consumption optimization adaptive control result of the computing power equipment.

[0006] The energy consumption optimization adaptive control method for computing equipment provided by this invention helps avoid misjudgment based on a single parameter by acquiring the initial operating environment parameter set of the computing equipment, thus improving the comprehensiveness of environmental state identification. Furthermore, by processing the initial operating environment parameter set through a data fusion algorithm, it achieves millisecond-level environmental response while outputting a target environmental assessment value reflecting the predicted trend of environmental changes, providing a basis for early activation of temperature control / protection strategies and improving the accuracy of operating condition identification. Furthermore, by determining the temperature control mode and anti-interference protection mode of the computing equipment through multi-dimensional judgment, it avoids defects such as temperature overshoot and frequent start-stop in a single mode; simultaneously covering dual-dimensional control of temperature and interference, it can balance temperature control accuracy and protection reliability, adapting to complex and extreme environments. Furthermore, by adjusting the temperature of the computing equipment through the temperature control mode, it can achieve stable control of chip junction temperature in high-temperature environments, while avoiding coolant freezing and start-up delays in low-temperature environments, ensuring stable operation of the equipment across the entire temperature range. Furthermore, by using the anti-interference protection mode to control the protection device to adjust external disturbances, it can actively reduce the impact of vibration resonance on device lifespan, improve electromagnetic interference immunity, reduce calculation errors and system crash risk, and lower the operation interruption rate. Furthermore, by utilizing temperature and interference regulation results to control energy consumption and redundancy management modules, energy redundancy is adjusted, enabling on-demand allocation of heat dissipation power while ensuring continuous system operation during failures, thus balancing energy saving and reliability. Therefore, by implementing this invention, adaptive adjustment of the operating status of computing equipment and energy consumption optimization are achieved.

[0007] In one optional implementation, based on the initial set of operating environment parameters, a target environment evaluation value is obtained through a data fusion algorithm, including: The initial operating environment parameter set is time-aligned to obtain the first operating environment parameter set. Robust preprocessing is then performed on the first operating environment parameter set to obtain the target operating environment parameter set and multiple channel health values. Time-domain and frequency-domain features are extracted from the target operating environment parameter set to obtain a composite feature set. Based on the composite feature set and the multiple channel health values, the actual weight values ​​of the multiple channels are determined. Based on the composite feature set and the actual weight values ​​of the multiple channels, a two-layer fusion model based on weighted evidence fusion and state-space prediction is used to obtain the target environment evaluation value.

[0008] The energy consumption optimization adaptive control method for computing devices provided by this invention eliminates the data asynchrony problem caused by differences in sampling frequencies of multi-source sensors by aligning the initial operating environment parameter set with time, thus avoiding distortion of fusion results caused by timing deviations. Furthermore, robust preprocessing dynamically reduces the weight of faulty channels, ensuring the fusion results remain stable and usable even when sensors degrade, avoiding control failures caused by sensor anomalies in traditional systems, and improving data anti-interference capabilities. Furthermore, feature extraction enhances the identification of complex operating conditions. Furthermore, by combining composite feature sets and multiple channel health values ​​to determine the actual channel weights, the problem of excessively high failure rates in traditional fixed-weight key channels is avoided, achieving dynamic adaptation of sensor weights and ensuring that high-reliability channel data dominates the fusion results, further improving the accuracy and anti-interference capabilities of environmental assessment values. Furthermore, combining weighted evidence fusion and state-space prediction not only solves the problem of unstable judgments under conflicting evidence but also enables short-term trend prediction, helping to provide early warning of environmental deterioration and providing a basis for proactive control, further shortening response lag.

[0009] In one optional implementation, based on the initial operating environment parameter set and the target environment assessment value, a temperature control mode and an anti-interference protection mode are determined, including: When the ambient temperature value in the initial operating environment parameter set is greater than the target environment assessment value, the temperature control mode is determined to be the heat dissipation control mode; when the ambient temperature value is less than the target environment assessment value, the temperature control mode is determined to be the heating control mode; when the vibration intensity value in the initial operating environment parameter set is greater than the preset threshold, the anti-interference protection mode is determined to be the vibration protection mode; when the electromagnetic field intensity in the initial operating environment parameter set does not meet the preset conditions, the anti-interference protection mode is determined to be the electromagnetic protection mode.

[0010] The energy consumption optimization adaptive control method for computing equipment provided by this invention precisely triggers a high-temperature heat dissipation strategy when the ambient temperature exceeds the target environmental assessment value. This solves the heat dissipation lag caused by traditional fixed temperature thresholds and lack of consideration for multiple environmental parameters, preventing chip frequency reduction or damage due to overheating and ensuring stable computing power output under high load. Furthermore, when the ambient temperature is below the target environmental assessment value, it precisely triggers a heating control mode, avoiding the problems of liquid freezing and startup delays in traditional low-temperature environments. This ensures the safe startup of computing equipment in extremely cold environments and prevents device performance degradation caused by low temperatures. Furthermore, when the vibration intensity exceeds a preset threshold, a vibration protection mode actively responds to vibration risks, avoiding the problems of traditional passive vibration reduction, which cannot cope with strong vibrations, leading to solder joint fatigue and shortened equipment lifespan. This reduces mechanical resonance damage to internal structures and extends equipment maintenance cycles. Furthermore, when the electromagnetic field strength does not meet preset conditions, an electromagnetic protection mode is used for protection, which can block electromagnetic interference in real time, reducing calculation errors and system crash probability, and improving operational stability in electromagnetic environments.

[0011] In one optional implementation, the temperature regulation device includes a heat dissipation execution module; using a temperature control mode, the temperature regulation device is controlled to regulate the temperature of the computing power device to obtain the temperature regulation result of the computing power device, including: When the temperature control mode is the heat dissipation control mode, multiple working modes of the heat dissipation control mode are obtained, including air cooling control mode, liquid cooling control mode, and combined air cooling and liquid cooling control mode. When the computing power device meets the first operating state, the air cooling control mode is used to control the heat dissipation execution module to adjust the temperature of the computing power device. When the air cooling control mode cannot meet the temperature adjustment requirements of the computing power device, the liquid cooling control mode is used to control the heat dissipation execution module to adjust the temperature of the computing power device. When the computing power device does not meet the first operating state, the combined air cooling and liquid cooling control mode is used to control the heat dissipation execution module to adjust the temperature of the computing power device, and the temperature adjustment result of the computing power device is obtained.

[0012] The energy consumption optimization adaptive control method for computing devices provided by this invention prioritizes low-energy air cooling when the computing device meets the first operating state, reducing energy waste from the constant operation of traditional liquid cooling, lowering device power consumption, simplifying control logic, and improving energy efficiency under low load. Furthermore, for medium-to-high load scenarios where air cooling capacity is insufficient, liquid cooling control mode is used to regulate temperature, avoiding temperature control lag caused by relying solely on air cooling, ensuring the chip junction temperature remains stable within a safe range, and preventing performance degradation due to overheating. Moreover, when the computing device does not meet the first operating state, temperature is regulated through a combined air cooling and liquid cooling control mode, balancing heat dissipation and energy consumption, avoiding insufficient heat dissipation or excessive energy consumption in a single mode, and improving operational stability under extreme thermal environments. Therefore, by implementing this invention, multiple operating modes of heat dissipation control provide adaptable solutions for different thermal load scenarios, avoiding the problems of insufficient air cooling capacity under high temperature and high load and excessive liquid cooling energy consumption under low load, thus achieving a balance between heat dissipation efficiency and energy saving.

[0013] In one optional embodiment, the temperature regulation device further includes an extreme cold heating module; using a temperature control mode, the temperature regulation device is controlled to regulate the temperature of the computing power equipment to obtain the temperature regulation result of the computing power equipment, and the device further includes: When the temperature control mode is heating control mode, the temperature difference is determined based on the ambient temperature and the operating temperature of the computing device. When the temperature difference meets the preset temperature difference condition, the extreme cold heating module is controlled to adjust the temperature of the computing device based on the preset heating temperature curve, and the temperature adjustment result of the computing device is obtained. When the temperature difference does not meet the preset temperature difference condition but the ambient temperature meets the preset temperature condition, the extreme cold heating module is controlled to adjust the temperature of the computing device until the temperature difference meets the preset temperature difference condition, and the temperature adjustment result is obtained.

[0014] The energy consumption optimization and adaptive control method for computing devices provided by this invention avoids thermal stress damage caused by traditional blind heating by precisely quantifying the temperature difference between the low-temperature environment and the device's requirements, thus helping to ensure the accuracy of temperature regulation. Furthermore, when the temperature difference meets the preset temperature difference condition, a preset heating temperature curve adjustment can achieve stable heating in low-temperature environments, avoiding damage to device structures caused by sudden heating, ensuring stable operation of the device under low-temperature conditions, and reducing ineffective heating energy consumption, thereby improving energy efficiency in low-temperature environments. Furthermore, when the temperature difference does not meet the preset temperature difference condition but the ambient temperature meets the preset temperature condition, continuous preheating can ensure that the device reaches a safe start-up temperature, avoiding hardware failures caused by low-temperature startup and improving device availability in extremely cold scenarios.

[0015] In one optional implementation, the protection device includes a vibration protection module and an electromagnetic protection module; utilizing an anti-interference protection mode, the protection device is controlled to adjust external disturbances to the computing power equipment, resulting in interference adjustment of the computing power equipment, including: When the anti-interference protection mode is vibration protection mode, the vibration protection module is controlled to reduce the operating frequency of the computing power equipment or adjust the tasks assigned to the computing power equipment, and the interference adjustment results of the computing power equipment are obtained. When the anti-interference protection mode is electromagnetic protection mode, the control electromagnetic protection module processes the target communication line or signal filtering parameters of the computing power equipment and obtains the interference adjustment result of the computing power equipment.

[0016] The energy consumption optimization adaptive control method for computing equipment provided by this invention can proactively avoid the risk of mechanical resonance by reducing the operating frequency or adjusting task allocation in vibration protection mode. This reduces the stress fatigue of components caused by resonance effects that traditional passive vibration damping cannot eliminate, thus extending the equipment's lifespan. Simultaneously, task adjustment ensures uninterrupted core computing, balancing protection and computing power output. Furthermore, by processing target communication lines or signal filtering parameters in electromagnetic protection mode, it can dynamically block electromagnetic interference transmission paths, avoiding signal distortion caused by sudden strong electromagnetic interference that traditional fixed shielding cannot handle. This reduces computational errors and the probability of system crashes, improving data transmission stability in strong electromagnetic environments.

[0017] In one optional implementation, the energy consumption and redundancy management module is used to adjust the energy consumption and redundancy of the computing device using temperature regulation results and disturbance regulation results, to obtain the energy consumption optimization adaptive control result of the computing device, including: Based on the temperature regulation results and interference regulation results, the cooling efficiency of the computing equipment is determined; based on the cooling efficiency, the energy consumption and redundancy management module adjusts the air cooling and liquid cooling ratio of the computing equipment, and obtains the power utilization rate and heat dissipation path performance of the computing equipment; when the power utilization rate or heat dissipation path performance does not meet the preset conditions, the energy consumption and redundancy management module switches to the backup circuit; the backup circuit is used to control the operation of the computing equipment, and the energy consumption optimization adaptive control results are obtained.

[0018] The adaptive control method for optimizing energy consumption of computing equipment provided by this invention determines the cooling efficiency of computing equipment by combining temperature regulation results and disturbance regulation results, quantifying the actual performance of the heat dissipation system and helping to improve the accuracy of energy consumption management. Furthermore, by adjusting the air-cooling and liquid-cooling ratio based on cooling efficiency, heat dissipation power is allocated on demand, minimizing energy consumption while ensuring heat dissipation effect, thus reducing energy costs in data centers or industrial scenarios. Furthermore, by switching to a backup circuit when power utilization or heat dissipation path is abnormal, the equipment continues to operate without interruption when the heat dissipation system fails, improving the reliability and fault tolerance of the equipment in extreme environments. Furthermore, by utilizing the backup circuit to obtain energy consumption optimization results, the energy consumption optimization state is maintained even in fault scenarios, balancing reliability and energy saving, and achieving energy efficiency balance under fault conditions.

[0019] In an optional implementation, the method further includes: modifying the energy consumption optimization adaptive control process of the computing device based on the temperature adjustment results, the disturbance adjustment results, and the energy consumption optimization adaptive control results.

[0020] The energy consumption optimization adaptive control method for computing equipment provided by this invention corrects the energy consumption optimization adaptive control process by combining temperature, interference and energy consumption optimization adaptive control results, thereby realizing the self-learning optimization of the control strategy. This enables the equipment to maintain high control accuracy and thermal balance efficiency over a long period of time, while reducing manual maintenance costs and extending the equipment's life cycle value.

[0021] Secondly, the present invention provides an adaptive control system for optimizing the energy consumption of computing devices, the system comprising: The system comprises a main control unit, an environmental sensing module, a temperature regulation device, a protection device, an energy consumption and redundancy management module, and a power management unit. The main control unit is connected to the environmental sensing module, the temperature regulation device, the protection device, the energy consumption and redundancy management module, and the power management unit. The power management unit is also connected to the energy consumption and redundancy management module. The temperature regulation device includes a heat dissipation execution module and an extreme cold heating module. The protection device includes a vibration protection module and an electromagnetic protection module. The environmental sensing module is used to acquire the operating environment parameter set of the computing device and send the operating environment parameter set to the main control unit. The main control unit is used to execute the energy consumption optimization adaptive control method of the first aspect or any corresponding embodiment described above.

[0022] The energy consumption optimization adaptive control system for computing equipment provided by this invention provides multi-source, real-time environmental data input through an environmental perception module system, breaking through the limitations of traditional single-point sensing and incomplete data. This ensures the comprehensiveness and accuracy of subsequent control decisions and improves the system's ability to perceive complex environments. Furthermore, by executing an energy consumption optimization adaptive control method through the main control unit, adaptive adjustment and energy consumption optimization of the computing equipment's operating status are achieved.

[0023] Thirdly, the present invention provides a computer program product, including computer instructions, which are used to cause a computer to execute the computing power device energy consumption optimization adaptive control method of the first aspect or any corresponding embodiment described above. Attached Figure Description

[0024] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0025] Figure 1 This is a schematic diagram of the adaptive control system for optimizing the energy consumption of computing equipment according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating the adaptive control method for optimizing the energy consumption of computing devices according to an embodiment of the present invention. Figure 3 This is a schematic diagram of the hardware structure of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0027] It is understood that before using the technical solutions disclosed in the various embodiments of the present invention, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in the present invention and their authorization should be obtained in accordance with relevant laws and regulations through appropriate means.

[0028] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0029] As an optional application scenario of this invention, such as Figure 1 As shown, a computing device energy consumption optimization adaptive control system 1 is provided, including: a main control unit 2, an environmental sensing module 3, a temperature regulation device 4, a protection device 5, an energy consumption and redundancy management module 6, and a power management unit 7.

[0030] Among them, the main control unit 2 is connected to the environmental sensing module 3, the temperature regulation device 4, the protection device 5, the energy consumption and redundancy management module 6, and the power management unit 7 respectively; the power management unit 7 is also connected to the energy consumption and redundancy management module 6.

[0031] In an optional embodiment, the temperature regulating device 4 includes a heat dissipation execution module 401 and an extreme cold heating module 402; the protection device 5 includes a vibration protection module 501 and an electromagnetic protection module 502.

[0032] In existing technologies, AI computing devices generally employ fixed-parameter air cooling or liquid cooling methods. Air cooling systems rely on forced convection by fans to remove heat; they are simple in structure but highly susceptible to ambient temperature and airflow resistance. Liquid cooling systems use coolant circulating in pipes for heat exchange; they are more efficient but complex to control and consume more energy. Both types of cooling methods mostly use single-mode control and cannot adaptively adjust to environmental changes and load fluctuations. When the equipment is in a high-temperature, high-humidity environment, traditional air cooling is easily affected by decreased air density and the accumulation of moisture and heat, resulting in a significant decrease in heat dissipation. Under low-temperature conditions, the coolant in the liquid cooling system may increase in viscosity or even freeze due to excessively low temperatures, leading to startup delays or heat dissipation failure. Furthermore, because these systems mostly rely on fixed threshold control, their response is lag-dependent and their adjustment precision is limited, making them prone to problems such as temperature overshoot, frequent start-ups and shutdowns, and energy waste.

[0033] Furthermore, in industrial environments characterized by high vibration and strong electromagnetic interference, computing equipment faces new operational risks. On the one hand, mechanical vibration can cause stress fatigue in internal solder joints and components, shortening equipment lifespan; on the other hand, electromagnetic interference may affect the stable transmission of chip logic signals, causing calculation errors or even system crashes. Existing protective measures are mostly passive isolation, such as installing rubber damping pads or metal shielding covers, but these solutions lack intelligent response capabilities and cannot dynamically adjust according to the intensity of external interference, resulting in limited protective effectiveness. Especially during high-intensity operating cycles, traditional systems struggle to balance temperature control accuracy, energy efficiency, and protection reliability.

[0034] In terms of energy management, current computing devices generally employ fixed-frequency or fixed-power operating modes for their temperature control systems, failing to dynamically allocate power based on real-time chip load and ambient temperature differences. When the system is under low load or in a cold environment, the heat dissipation device still operates at high power, leading to energy waste; while in high-load scenarios, insufficient energy supply may cause overheating and performance degradation. Due to the lack of a coordinated control mechanism between environment and power consumption, existing technologies present a significant contradiction between achieving energy conservation and safe operation.

[0035] According to an embodiment of the present invention, an embodiment of an adaptive control method for optimizing the energy consumption of computing devices is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0036] This embodiment provides a computing power equipment energy consumption optimization adaptive control method, which can be used in the main control unit 2 of the aforementioned computing power equipment energy consumption optimization adaptive control system 1. Figure 2 This is a flowchart of an adaptive control method for optimizing the energy consumption of computing devices according to an embodiment of the present invention, such as... Figure 2As shown, the process includes the following steps: Step S201: Obtain the initial operating environment parameter set of the computing power device.

[0037] In one optional embodiment, the computing device refers to a hardware device that integrates multiple high-power AI chips, accelerator cards, and high-speed storage units to support artificial intelligence computing tasks and is adapted to complex and extreme environments such as industrial, field, and special operations. It may include artificial intelligence servers, edge computing terminals, high-performance industrial control equipment, etc.

[0038] In one optional embodiment, the initial operating environment parameter set represents the set of basic parameters collected by the multi-source sensing unit of the environment sensing module 3 during the initial startup and operation of the computing device or the initial stage of periodic monitoring. These parameters reflect the external environment and key internal state of the device and may include information such as ambient temperature, humidity, air pressure, vibration intensity, and electromagnetic field intensity.

[0039] In one optional embodiment, multiple different sensors are deployed at key locations inside the computing device (such as near the chip, in the heat dissipation duct, or on the device casing) and at external environmental monitoring points. These sensors may include temperature sensors, humidity sensors, air pressure sensors, vibration sensors, electromagnetic sensors, etc.

[0040] Furthermore, data collection can be performed using a dual-mode approach, combining periodic triggering and event triggering. (1) Periodic triggering: Automatically start data acquisition according to the preset scheduling period (50ms~100ms) to ensure the real-time performance of parameters and avoid parameter lag caused by excessively long traditional fixed intervals; (2) Event triggering: When the device starts up, the load changes suddenly, or the previously collected parameters are close to the threshold, the instant collection is triggered to supplement the deficiencies of periodic collection and ensure that key node parameters are not missed.

[0041] At the same time, independent sampling frequencies can be set for each sensor based on the changing characteristics of different parameters.

[0042] Furthermore, once the data collection is complete, the environmental perception module 3 sends the collected initial operating environment parameter set to the main control unit 2.

[0043] Step S202: Based on the initial operating environment parameter set, the target environment evaluation value is obtained through data fusion algorithm calculation.

[0044] In one optional embodiment, the data fusion algorithm represents a hierarchical, parameterizable, multi-dimensional data processing and comprehensive analysis algorithm that addresses the characteristics of multi-source environmental parameters and internal state parameters of computing devices. Through an organic combination of "time alignment, robust preprocessing, feature construction, weight adaptation, evidence fusion, and state prediction," it achieves high-precision assessment and advanced prediction of environmental states under complex operating conditions.

[0045] In one optional embodiment, the target environment assessment value represents a quantitative indicator that reflects the overall state of the current environment of the computing device, obtained through hierarchical processing and comprehensive analysis of the initial operating environment parameter set.

[0046] In one optional embodiment, by using a hierarchical, adaptive multi-dimensional data fusion algorithm to comprehensively process and calculate the initial operating environment parameter set, the overall environment assessment value, i.e., the target environment assessment value, can be obtained.

[0047] Step S203: Based on the initial operating environment parameter set and the target environment assessment value, determine the temperature control mode and anti-interference protection mode of the computing power equipment.

[0048] In one optional embodiment, the temperature control mode refers to a dynamic adjustment mode set according to the ambient temperature and internal heat load requirements of the computing device, in order to maintain the core temperature of the chip within a safe range and avoid device failure caused by extreme temperatures.

[0049] In one optional embodiment, the anti-interference protection mode refers to an active protection strategy set up to avoid mechanical damage, signal distortion or system crashes of computing devices in industrial, field and other scenarios, in response to non-temperature risks such as vibration and electromagnetic interference.

[0050] In one optional embodiment, based on the initial operating environment parameter set and the target environment assessment value, the temperature control mode and anti-interference protection mode that are suitable for the current operating conditions can be accurately matched through multi-dimensional condition judgment.

[0051] In one optional embodiment, based on the initial operating environment parameter set and the target environment assessment value, a lightweight deep learning model (such as a CNN-LSTM hybrid model) is used, combined with historical equipment operating data, to predict the compound operating conditions that may overlap within the next 3-5 minutes. Simultaneously, an operating condition risk priority assessment matrix is ​​established, and the various potential operating conditions are prioritized according to the weight of the parameters' impact on the equipment. For example, in high-temperature and strong electromagnetic scenarios, the heat dissipation control mode is prioritized and activated, while the electromagnetic protection mode is activated simultaneously with lower priority.

[0052] Furthermore, by dynamically adjusting the startup order and resource allocation based on real-time load requirements, conflicts in control strategies can be avoided when multiple operating conditions are concurrent, further improving the foresight and adaptability of operating condition judgment in complex and extreme environments. For example, when the real-time load requirement is a high-computing-power task, priority is given to ensuring temperature control stability, while when the real-time load requirement is a low-computing-power task, priority is given to strengthening anti-interference.

[0053] Step S204: Using the temperature control mode, the temperature regulation device is controlled to regulate the temperature of the computing power equipment, and the temperature regulation result of the computing power equipment is obtained.

[0054] In an optional embodiment, according to the determined temperature control mode, the main control unit 2 can drive the temperature regulation device 4 to perform targeted temperature intervention actions. At the same time, through the adjustment logic that dynamically adapts to changes in environment and load, the core temperature of the computing device is stabilized within a safe operating range, and the temperature regulation result that reflects the temperature control effect is finally output.

[0055] Step S205: Using the anti-interference protection mode, the control protection device is used to adjust the external disturbances to the computing power equipment, and the interference adjustment result of the computing power equipment is obtained.

[0056] In an optional embodiment, according to the determined anti-interference protection mode, the main control unit 2 can drive the protection device 5 to perform active protection actions adapted to the characteristics of external disturbances. At the same time, by dynamically adjusting the equipment operating parameters or signal processing methods, it can offset the negative impact of vibration and electromagnetic interference on the equipment, and ultimately output the interference adjustment result that reflects the protection effect.

[0057] Step S206: Using the temperature regulation results and interference regulation results, the energy consumption and redundancy management module is controlled to adjust the energy consumption and redundancy of the computing equipment, so as to obtain the energy consumption optimization adaptive control result of the computing equipment.

[0058] In an optional embodiment, by combining the obtained temperature regulation results and interference regulation results, the main control unit 2 can drive the energy consumption and redundancy management module 6 to perform dynamic energy allocation and fault tolerance actions.

[0059] For example, on the one hand, the energy utilization efficiency in the temperature control and protection process can be quantitatively analyzed to optimize the power allocation of heat dissipation, heating and protection modules and avoid ineffective energy consumption; on the other hand, the operating status of key systems such as temperature control path and power supply link can be monitored to automatically switch to backup resources when performance abnormalities or failures occur, ensuring continuous operation of equipment, and finally outputting energy consumption optimization adaptive control results that take into account both energy saving effect and operational reliability.

[0060] The adaptive control method for optimizing computing power equipment provided in this embodiment helps avoid misjudgment based on a single parameter by acquiring the initial operating environment parameter set of the computing power equipment, thus improving the comprehensiveness of environmental state identification. Furthermore, by processing the initial operating environment parameter set through a data fusion algorithm, it achieves millisecond-level environmental response while outputting a target environmental assessment value reflecting the predicted trend of environmental changes, providing a basis for early activation of temperature control / protection strategies and improving the accuracy of operating condition identification. Furthermore, by determining the temperature control mode and anti-interference protection mode of the computing power equipment through multi-dimensional judgment, it avoids defects such as temperature overshoot and frequent start-stop in a single mode; simultaneously covering dual-dimensional control of temperature and interference, it can balance temperature control accuracy and protection reliability, adapting to complex and extreme environments. Furthermore, by adjusting the temperature of the computing power equipment through the temperature control mode, it can achieve stable control of chip junction temperature in high-temperature environments, while avoiding coolant freezing and start-up delays in low-temperature environments, ensuring stable operation of the equipment across the entire temperature range. Furthermore, by using the anti-interference protection mode to control the protection device to adjust external disturbances, it can actively reduce the impact of vibration resonance on device lifespan, improve electromagnetic interference immunity, reduce calculation errors and system crash risk, and lower the operation interruption rate. Furthermore, by utilizing temperature and interference regulation results to control energy consumption and redundancy management modules, energy redundancy is adjusted, enabling on-demand allocation of heat dissipation power while ensuring continuous system operation during failures, thus balancing energy saving and reliability. Therefore, by implementing this invention, adaptive adjustment of the operating status of computing equipment and energy consumption optimization are achieved.

[0061] In some optional implementations, step S202 above includes: Step S2021: Time-align the initial runtime environment parameter set to obtain the first runtime environment parameter set.

[0062] In one alternative embodiment, a scheduling period dt is determined, and all sensor data are mapped to a discrete-time grid with intervals of dt, thereby determining the time node corresponding to each data point.

[0063] Furthermore, for slow variables with sampling frequencies below 1 / dt (such as humidity and air pressure), zero-order hold or first-order spline interpolation is used to make up to a uniform frequency; for high-frequency quantities (such as vibration and electromagnetic fields), a short-window mean or peak hold strategy covering 1 to 2 dt is used to maintain sensitivity to transient disturbances.

[0064] Furthermore, through the above alignment strategy, a multi-source input sequence with the same frequency and phase can be obtained, namely the first set of operating environment parameters.

[0065] Step S2022: Perform robust preprocessing on the first operating environment parameter set to obtain the target operating environment parameter set and multiple channel health values.

[0066] In an optional embodiment, the channel health value represents an index obtained by quantitatively evaluating the working status of each sensor channel in the initial operating environment parameter set, with a value range of 0 to 1, and is used to determine the reliability of each sensor data.

[0067] For example, when there are anomalies such as continuous packet loss, data drift, or values ​​exceeding the calibration range, the health value decreases; when the data is stable and there are no anomalies, the health value remains high.

[0068] In one optional embodiment, in order to suppress the impact of sensing anomalies and sudden spikes on the fusion results, robust preprocessing is performed on the data of each channel, i.e., the first set of operating environment parameters.

[0069] For example, based on the upper and lower limits of each parameter's calibration, outliers in the first operating environment parameter set that exceed the range are limited, and the excess is corrected to the calibration boundary value to avoid interference from extreme outliers.

[0070] Furthermore, median filtering combined with Hampel identification is used to remove isolated spikes, i.e., outlier data. Specifically, the data is first smoothed by median filtering, then the local median and absolute deviation are calculated, and values ​​that deviate from the local median ± 3 times the absolute deviation are identified as isolated spikes and removed. Finally, reasonable data from adjacent time points are used for interpolation to complete the data.

[0071] Furthermore, the operating status of each sensing channel is quantified. Specifically, for channels experiencing continuous packet loss or fixed value drift, anomaly counts are recorded and the channel health q (ranging from 0 to 1) is updated. Further, when q falls below a preset threshold (preferably 0.6), the weight of that channel in the fusion process is dynamically reduced or temporarily disabled, thereby ensuring the stability and availability of the fusion calculation under sensing degradation conditions.

[0072] Furthermore, the valid data after amplitude limiting and anomaly removal are integrated into the target operating environment parameter set, and the health value corresponding to each channel is output synchronously, i.e., the channel health value.

[0073] Step S2023: Extract time-domain and frequency-domain features from the target operating environment parameter set to obtain a composite feature set.

[0074] In an optional embodiment, by extracting features from both the time and frequency domains, the original parameters can be transformed into feature vectors that better reflect the essence of the working condition, such as vibration resonance features and electromagnetic harmonic features.

[0075] In an optional embodiment, taking any channel x as an example, the normalized quantity (x minus the historical mean and then divided by the historical standard deviation), the rate of change per unit time (current value minus the previous value and then divided by dt), and the mean, standard deviation, maximum value and linear trend coefficient over a sliding window of 2 to 10 seconds are calculated.

[0076] Simultaneously, short-time Fourier transforms are performed on vibration intensity and electromagnetic field intensity to extract frequency domain features such as main peak frequency, in-band root mean square, and spectral kurtosis, which are used to characterize scenarios such as resonance, harmonics, or electromagnetic interference. Through the above processing, feature vectors containing multi-source statistical and physical parameters can be obtained at each time step, and integrated to form corresponding composite feature sets.

[0077] Among them, the main peak frequency represents the frequency with the highest energy in the frequency distribution; the root mean square within the band represents the average energy within a specific frequency range; and the spectral kurtosis is used to reflect the steepness of the frequency distribution.

[0078] Step S2024: Determine the actual weight values ​​of multiple channels based on the composite feature set and the health values ​​of multiple channels.

[0079] In one optional embodiment, by combining channel health and feature correlation, the actual weight of each channel can be dynamically adjusted to ensure that highly reliable and highly correlated channel data plays a dominant role in the fusion process, thereby improving the accuracy of the fusion results.

[0080] In one optional embodiment, during the factory calibration stage, static weights are set based on the correlation between each channel and the core task (such as temperature control and anti-interference). Then, for each channel, a reliability factor is calculated by combining the composite feature set and the health value, which may include the channel health value q, residual stability, and real-time correlation with the target variable. The residual stability represents the magnitude of the deviation between the feature value and the historical mean.

[0081] Furthermore, the adaptive weight, i.e. the actual weight value of the channel, is calculated using the formula "actual weight = static weight × mapping function (confidence factor)".

[0082] The mapping function can be piecewise linear or logistic in form, ensuring that the weights decay rapidly when the channel degrades and recover smoothly when the channel recovers. Specifically, when the confidence factor is high (i.e., the data is reliable and highly correlated), the mapping function output is close to 1, and the actual weights are close to the static weights; when the confidence factor is low (i.e., the data is abnormal and weakly correlated), the mapping function output decays rapidly, and the actual weights decrease significantly, even approaching 0.

[0083] Step S2025: Based on the composite feature set and the actual weight values ​​of multiple channels, the target environment assessment value is obtained through a two-layer fusion model based on weighted evidence fusion and state space prediction.

[0084] In one optional embodiment, the two-layer fusion model based on weighted evidence fusion and state space prediction is divided into two layers: instantaneous state assessment and short-term trend prediction, which are used to achieve accurate judgment and early warning of environmental state.

[0085] In one optional embodiment, an interpretable mapping is applied to the channel features of the composite feature set, transforming the multi-dimensional features into single-dimensional contribution values ​​that can directly participate in the fusion. The interpretable mapping can be a temperature and humidity coupling term, a threshold-type piecewise function, a small-scale shallow network, etc.

[0086] Furthermore, using the actual weights of each channel as coefficients, the feature contribution values ​​are weighted and summed to obtain preliminary fusion results.

[0087] Furthermore, by combining the Dempster-Shafer evidence synthesis theory, confidence levels are assigned to assumptions about operating conditions such as overheating, strong vibration, and strong electromagnetic interference. After conflict normalization, the weighted summation result is compressed to the 0~1 interval through the Sigmoid function, and a robust score S(t) reflecting the instantaneous state of the current environment is obtained. The joint confidence level of each operating condition assumption, i.e., the prior distribution of the operating conditions, is output simultaneously.

[0088] Furthermore, a state-space model is constructed using Kalman filtering or extended / unscented Kalman filtering. The observation equation takes the output S(t) of the first layer and dominant variables such as temperature and humidity as inputs, and the state evolution equation introduces the parameter change rate and a seasonal term. The seasonal term can be the diurnal temperature and humidity cycle.

[0089] Furthermore, based on current and historical S(t) and key variable data, the predicted values ​​of S(t+h), temperature, and humidity at future time t+h (h=1~5 minutes) can be extrapolated using a state-space model to obtain the environmental change trend.

[0090] Furthermore, through the above processing, we can obtain the target environmental assessment value S(t), the prior distribution of the working condition, the confidence interval, the short-time prediction result S(t+h), and the vibration resonance factor and electromagnetic anomaly factor extracted in parallel from the fusion process.

[0091] In an optional embodiment, to enhance the interpretability of the algorithm and the efficiency of on-site parameter tuning, a physics-heuristic design is adopted for the key mapping. For example, when a rapid rise in external temperature and high humidity are detected, the contribution of the temperature-humidity coupling term to S(t) will be nonlinearly amplified; when a significant increase in electromagnetic anomaly factors occurs simultaneously, the system maintains a conservative interpretation of electromagnetically related temperature rise anomalies and avoids false overheating strategies through weighted suppression. This interpretable structure, combined with evidence theory, enables the system to output stable conclusions that conform to engineering intuition even when multiple parameters conflict.

[0092] In some optional implementations, step S203 above includes: Step S2031: When the ambient temperature value in the initial operating environment parameter set is greater than the target environment assessment value, the temperature control mode is determined to be the heat dissipation control mode.

[0093] In one optional embodiment, the heat dissipation control mode indicates that when the ambient temperature of the computing device is too high and the internal chip generates a lot of heat due to high load, which may cause the junction temperature to exceed the safe range, a dynamic temperature control mode is activated with the goal of rapid heat dissipation and maintaining temperature stability.

[0094] Furthermore, this heat dissipation control mode coordinates air cooling, liquid cooling, or a combination of both, and adjusts the heat dissipation intensity according to the rate of temperature change and load fluctuations. This ensures that the chip junction temperature remains stable within a safe range under extreme environments such as high temperature and high humidity, avoiding performance degradation or hardware damage caused by overheating.

[0095] In one optional embodiment, the current real-time ambient temperature value is extracted from the initial operating environment parameter set and compared with the target environment assessment value. Further, if the ambient temperature value is greater than the target environment assessment value, it indicates that the current environment needs to reduce the equipment temperature through heat dissipation, thus determining the temperature control mode as heat dissipation control mode.

[0096] Step S2032: When the ambient temperature value is less than the target environmental assessment value, the temperature control mode is determined to be the heating control mode.

[0097] In one optional embodiment, the heating control mode refers to a temperature control mode that is activated when the ambient temperature of the computing device is too low (such as an extremely cold environment below -30°C), which may cause the coolant viscosity to increase, freeze, or delay the device startup or degrade the device performance. The goal is to achieve a stable temperature rise and ensure safe startup and operation.

[0098] Furthermore, this heating control mode, through strategies such as graded heating and preheating of key areas, can avoid thermal stress damage caused by sudden heating, ensuring that the equipment can start up normally and maintain stable operation in extremely cold environments.

[0099] In one optional embodiment, when the ambient temperature value is less than the target environmental assessment value, it indicates that the current environment is under low temperature or extremely cold conditions, and heating is required to ensure the safe start-up and operation of the equipment, thereby determining the temperature control mode as the heating control mode.

[0100] Step S2033: When the vibration intensity value in the initial operating environment parameter set is greater than the preset threshold, the anti-interference protection mode is determined to be the vibration protection mode.

[0101] In one optional embodiment, the vibration protection mode refers to an active protection mode that is activated when the vibration intensity of the environment in which the computing device is located exceeds the safe range, which may lead to stress fatigue of internal solder joints, loosening of components, or structural resonance. The mode aims to reduce vibration damage and ensure the stability of the device structure and operation.

[0102] Furthermore, this vibration protection mode can mitigate the risk of mechanical resonance, extend equipment life, and avoid operational interruptions caused by vibration by adjusting equipment operating parameters (such as reducing operating frequency) or task allocation.

[0103] In one optional embodiment, the current vibration intensity value is extracted from the initial operating environment parameter set and compared with a preset threshold. Further, if the vibration intensity value is greater than the preset threshold, it indicates that the vibration has exceeded the safe range, and thus the anti-interference protection mode is determined to be the vibration protection mode.

[0104] Step S2034: When the electromagnetic field strength in the initial operating environment parameter set does not meet the preset conditions, the anti-interference protection mode is determined to be electromagnetic protection mode.

[0105] In one optional embodiment, the electromagnetic protection mode refers to an active protection mode that is activated when the electromagnetic field strength of the environment in which the computing device is located is abnormal, which may interfere with the transmission of chip logic signals, cause calculation errors or system crashes, in order to block electromagnetic interference and ensure signal stability.

[0106] Furthermore, this electromagnetic protection mode can reduce the impact of electromagnetic interference on equipment and improve the operational reliability of equipment in strong electromagnetic environments by shielding sensitive communication lines and adjusting signal filtering parameters.

[0107] In one optional embodiment, the current electromagnetic field strength is extracted from the initial operating environment parameter set, and when the electromagnetic field strength increases or a sudden change occurs, it indicates that the electromagnetic field strength does not meet the safety conditions, and thus the anti-interference protection mode is determined to be the electromagnetic protection mode.

[0108] In some optional implementations, step S204 above includes: Step S2041: When the temperature control mode is the heat dissipation control mode, obtain multiple working modes of the heat dissipation control mode.

[0109] Among them, there are multiple working modes, including air-cooled control mode, liquid-cooled control mode, and combined air-cooled and liquid-cooled control mode.

[0110] In one optional embodiment, the air-cooled control working mode refers to a low-energy, easy-to-maintain heat dissipation mode that uses forced air convection to remove heat from the inside of the computing device by adjusting the operating parameters of the air-cooling system in the heat dissipation execution module.

[0111] In one optional embodiment, the liquid cooling control working mode refers to a high heat dissipation efficiency mode that utilizes the efficient heat exchange characteristics of the coolant to transfer the core heat of the computing device by adjusting the operating parameters of the liquid cooling system in the heat dissipation execution module.

[0112] In one optional embodiment, the combined air-cooling and liquid-cooling control mode refers to a composite heat dissipation mode that simultaneously activates the air-cooling and liquid-cooling systems and dynamically allocates their operating ratios based on the heat load intensity of the computing equipment and environmental conditions.

[0113] Step S2042: When the computing power device meets the first operating state, the heat dissipation execution module is controlled to adjust the temperature of the computing power device using the air-cooled control working mode.

[0114] In an optional embodiment, when the computing device is in a low-load, low-heat-flux operating state, the low-energy-consumption air-cooling mode is enabled first, that is, the main control unit 2 sends an air-cooling control working mode command to the heat dissipation execution module 401.

[0115] Furthermore, under the control of the air-cooling control working mode command, the heat dissipation execution module 401 adjusts the cooling fan speed to change the airflow speed, thereby achieving rapid heat dissipation on the surface of the equipment.

[0116] Step S2043: When the air-cooled control mode cannot meet the temperature regulation requirements of the computing equipment, the liquid-cooled control mode is used to control the heat dissipation execution module to regulate the temperature of the computing equipment.

[0117] In an optional embodiment, when the air-cooling mode cannot maintain a stable temperature due to insufficient heat dissipation capacity, such as when the ambient temperature is high or the equipment is under high load, causing the heat accumulation rate to exceed the air-cooling heat dissipation rate, the system will switch to a more efficient liquid-cooling mode in a timely manner. That is, the main control unit 2 will send a liquid-cooling control working mode command to the heat dissipation execution module 401.

[0118] Furthermore, under the control of the liquid cooling control working mode command, the heat dissipation execution module 401 achieves efficient heat transfer by adjusting the flow rate and temperature difference of the coolant.

[0119] Step S2044: When the computing power device does not meet the first operating state, the combined air cooling and liquid cooling control mode is used to control the heat dissipation execution module to adjust the temperature of the computing power device and obtain the temperature adjustment result of the computing power device.

[0120] In one optional embodiment, when the computing device is operating under high load and high heat flux, the heat dissipation capacity is maximized through the coordinated operation of air cooling and liquid cooling to ensure that the chip junction temperature remains stable within a safe range.

[0121] For example, when the computing device is operating under high load and high heat flux, the main control unit 2 sends a command for the combined air cooling and liquid cooling control mode to the heat dissipation execution module 401.

[0122] Furthermore, under the control of the combined air cooling and liquid cooling control mode command, the heat dissipation execution module 401 simultaneously activates the air cooling and liquid cooling systems and automatically allocates the working ratio of the two according to the temperature change rate, so as to keep the heat dissipation effect and energy consumption in balance.

[0123] Furthermore, through the aforementioned dynamic coordination mechanism, the temperature control lag problem that occurs in traditional systems under high-temperature environments can be effectively avoided, keeping the chip junction temperature within a safe range.

[0124] In some optional implementations, step S204 above further includes: Step a1: When the temperature control mode is heating control mode, determine the temperature difference based on the ambient temperature and the operating temperature of the computing device.

[0125] In an optional embodiment, the temperature difference can be calculated using the formula: "Temperature difference = Operating temperature of computing device - Ambient temperature".

[0126] In an optional embodiment, the calculated temperature difference can be verified to be within a reasonable range by combining the preset temperature measurement range of the computing device. Furthermore, if the difference is invalid, a sensor anomaly warning can be triggered and a historical average value can be temporarily used as a substitute to ensure that subsequent heating decisions are not interrupted.

[0127] By calculating the temperature difference, the current low-temperature gap of the equipment can be accurately determined. This can avoid thermal stress damage caused by traditional blind heating and provide a quantitative basis for adjusting the heating intensity as needed, which helps to ensure that the heating strategy fits the actual temperature requirements of the equipment.

[0128] Step a2: When the temperature difference meets the preset temperature difference condition, based on the preset heating temperature curve, the extreme cold heating module is controlled to adjust the temperature of the computing device using the heating control mode to obtain the temperature adjustment result of the computing device.

[0129] In an optional embodiment, when the temperature difference is within a safe range that does not require high-intensity heating and thus meets the preset temperature difference condition, a preset heating temperature curve can be used for refined heating. This can avoid thermal stress damage to the device caused by sudden heating (such as solder joint cracking and material deformation) and ensure that the equipment temperature rises steadily to the target range. The heating temperature curve is used to reflect the relationship between time and heating power.

[0130] For example, the main control unit 2 sends a heating command to the extreme cold heating module 402. Further, under the control of the heating command, the extreme cold heating module 402 outputs heating power according to a preset curve while collecting the device's operating temperature in real time. If the temperature rise rate deviates from the curve (such as being too fast or too slow), the heating power is finely adjusted until the temperature of the computing device rises steadily, and the corresponding temperature adjustment result is obtained.

[0131] In one optional embodiment, the heating intensity is automatically selected based on the ambient temperature difference. When the temperature difference is large, a graded heating strategy is adopted, that is, low-power preheating is performed first, and then the heating output is gradually increased to make the temperature of the computing equipment rise steadily and avoid thermal stress damage caused by sudden heating.

[0132] Step a3: When the temperature difference does not meet the preset temperature difference condition and the ambient temperature meets the preset temperature condition, the heating control mode is used to control the extreme cold heating module to adjust the temperature of the computing device until the temperature difference meets the preset temperature difference condition and the temperature adjustment result is obtained.

[0133] In an optional embodiment, when the temperature difference is within the safe range requiring high-intensity heating and the ambient temperature is below a preset threshold ( 30℃ indicates that the current environment is in extremely cold conditions. At this time, it is necessary to make up for the low temperature gap by continuous heating to ensure that the computing equipment reaches the temperature conditions for safe start-up and operation.

[0134] For example, when the temperature difference does not meet the preset temperature difference condition but the ambient temperature meets the preset temperature condition, the main control unit 2 sends a heating command to the extreme cold heating module 402.

[0135] Furthermore, under the control of the heating command, the extreme cold heating module 402 preheats the key areas before starting, so that they reach the allowable start-up temperature range before entering normal operation.

[0136] Furthermore, once the temperature difference is within a safe range where high-intensity heating is not required, step a2 is executed until the temperature of the computing device rises steadily and the corresponding temperature adjustment result is obtained.

[0137] By controlling the heating rate and time, a gradual temperature transition was achieved, thereby ensuring that the computing equipment could start up safely and maintain stable operation in a low-temperature environment.

[0138] In some optional implementations, step S205 above includes: Step S2051: When the anti-interference protection mode is vibration protection mode, control the vibration protection module to reduce the operating frequency of the computing power device or adjust the tasks allocated to the computing power device, and obtain the interference adjustment result of the computing power device.

[0139] In one optional embodiment, when the vibration intensity value is greater than a preset threshold, the vibration protection mode is used to actively protect against external disturbances to the computing device.

[0140] For example, when the vibration intensity value exceeds a preset threshold, the main control unit 2 sends a vibration protection command to the vibration protection module 501. Furthermore, under the control of the vibration protection command, the vibration protection module 501 automatically reduces the operating frequency of the computing device to reduce the impact of mechanical resonance.

[0141] In an optional embodiment, the vibration protection module 501 can also automatically adjust the tasks assigned to the computing device under the control of the vibration protection command to reduce the impact of mechanical resonance.

[0142] By dynamically adjusting the operating status (frequency / task) of computing equipment, risks can be mitigated from two dimensions: reducing the impact of vibration sources and reducing the equipment's vibration sensitivity. This not only solves the problem that traditional passive vibration reduction cannot eliminate resonance damage, but also avoids internal solder joint stress fatigue and component loosening caused by vibration, ultimately ensuring the structural safety and operational continuity of computing equipment.

[0143] Step S2052: When the anti-interference protection mode is electromagnetic protection mode, the electromagnetic protection module is controlled to process the target communication line or signal filtering parameters of the computing power device and obtain the interference adjustment result of the computing power device.

[0144] In one optional embodiment, when the electromagnetic field strength increases or a sudden change occurs, an electromagnetic protection mode is used to actively protect against external disturbances to the computing device.

[0145] For example, when the electromagnetic field strength increases or a sudden change occurs, the main control unit 2 sends an electromagnetic protection command to the electromagnetic protection module 502. Furthermore, under the control of this electromagnetic protection command, the electromagnetic protection module 502 temporarily shields highly sensitive communication lines or adjusts signal filtering parameters to reduce the impact of electromagnetic interference on control logic and data transmission.

[0146] By specifically blocking the transmission path of electromagnetic interference (shielding lines) and optimizing signal processing logic (adjusting filtering), the problem that traditional fixed shielding cannot cope with sudden strong electromagnetic interference and is prone to signal distortion is solved, ensuring stable transmission of chip logic signals and avoiding calculation errors or system crashes.

[0147] In some optional implementations, step S206 above includes: Step S2061: Determine the cooling efficiency of the computing power equipment based on the temperature adjustment results and the interference adjustment results.

[0148] In one optional embodiment, the cooling capacity, i.e. the cooling efficiency, under unit energy consumption is calculated by the temperature regulation effect (whether heat dissipation / heating meets the standard) and the indirect impact of interference regulation on temperature control.

[0149] For example, the temperature difference of the chip core before and after adjustment, the working time of the heat dissipation module, and the actual power consumption of air cooling / liquid cooling can be extracted from the temperature adjustment effect. At the same time, the interference adjustment results can be used to determine whether the anti-interference action affects the temperature control and to correct the energy consumption statistics.

[0150] Furthermore, the current cooling efficiency can be calculated using the formula: "Cooling efficiency = (Temperature difference × Chip thermal capacity) / (Total heat dissipation power × Operating time)". A higher cooling efficiency indicates better heat dissipation per unit of energy consumption.

[0151] In step S2062, based on cooling efficiency, the control energy consumption and redundancy management module adjusts the air cooling and liquid cooling ratio of the computing equipment and obtains the power utilization rate and heat dissipation path performance of the computing equipment.

[0152] In one optional embodiment, the power utilization rate represents the ratio of the actual output power of heat dissipation-related modules (such as air-cooled fans and liquid-cooled circulating pumps) in the computing device to the rated power, and is used to determine whether there is power waste or overload risk in the heat dissipation module.

[0153] In one optional embodiment, the heat dissipation path performance represents the working status and efficiency index of the physical path (such as air-cooled duct, liquid-cooled pipeline) responsible for heat transfer in the computing device, and is used to determine whether the heat dissipation hardware is normal and whether the heat dissipation capacity meets the standard.

[0154] In one optional embodiment, the power allocation ratio between air cooling and liquid cooling is dynamically adjusted according to the cooling efficiency. This can avoid the energy waste caused by the traditional fixed ratio operation, and also ensure that the heat dissipation effect does not decrease during the optimization process by monitoring the power utilization rate (whether the energy consumption matches the demand) and the performance of the heat dissipation path (whether the hardware is normal).

[0155] For example, if the efficiency is higher than the benchmark, the proportion of high-energy-consuming liquid cooling can be appropriately reduced, while the proportion of low-energy-consuming air cooling can be increased to further reduce the total power consumption. Furthermore, if the efficiency is lower than the benchmark, the proportion of liquid cooling can be increased to enhance heat dissipation capacity, while simultaneously investigating whether dust blockage in the air cooling system is causing a decrease in efficiency.

[0156] In an optional embodiment, the main control unit 2 determines an adjustment strategy based on the cooling efficiency and sends an adjustment command to the energy consumption and redundancy management module 6. Furthermore, under the control of this adjustment command, the energy consumption and redundancy management module 6 automatically adjusts the ratio of air cooling to liquid cooling. Simultaneously, it collects real-time data on the power utilization rate and heat dissipation path performance of the computing equipment.

[0157] Step S2063: When the power utilization rate or heat dissipation path performance does not meet the preset conditions, the control energy consumption and redundancy management module switches to the backup circuit.

[0158] In one optional embodiment, by monitoring energy consumption efficiency (power utilization rate) and hardware health status (heat dissipation path), backup resources are activated when an anomaly occurs. This not only solves the problem of traditional systems having no redundancy and shutting down upon failure, but also avoids chip overheating or interference caused by heat dissipation failure, ensuring continuous and stable operation of the device in extreme environments.

[0159] In an optional embodiment, when power utilization decreases or a certain heat dissipation path malfunctions, the main control unit sends a switching command to the energy consumption and redundancy management module 6. Furthermore, under the control of this switching command, the energy consumption and redundancy management module 6 automatically switches to the backup circuit to ensure continuous and stable system operation.

[0160] For example, if the power utilization rate is low, the system will switch to the low-power backup heat dissipation circuit first; if the heat dissipation circuit fails, the backup heat dissipation circuit (such as the backup liquid cooling circuit) or the hybrid backup scheme (backup air cooling + main liquid cooling bypass) will be activated.

[0161] Step S2064: Use the backup circuit to control the operation of the computing equipment and obtain the energy consumption optimization adaptive control results.

[0162] In one alternative embodiment, the stable operation of the backup circuit can ensure the safety of the computing equipment while continuously maintaining an optimized energy consumption state.

[0163] In an optional embodiment, the main control unit 2 can set the operating parameters of the backup circuit according to the current computing power equipment load and environmental conditions to ensure that the heat dissipation effect meets the standard, and at the same time control the power consumption of the backup circuit to avoid energy waste.

[0164] Furthermore, the main control unit 2 can continuously monitor the cooling efficiency, power utilization, and equipment operation stability of the backup circuit, and confirm that the backup circuit meets both heat dissipation requirements and energy consumption optimization goals.

[0165] Furthermore, the entire process data, including temperature regulation effect, interference protection effect, heat dissipation ratio optimization scheme, backup circuit operating parameters, and total energy consumption data, are integrated to form corresponding energy consumption optimization adaptive control results.

[0166] In some optional implementations, the above method further includes: Step b1: Based on the temperature regulation results, disturbance regulation results, and energy consumption optimization adaptive control results, the energy consumption optimization adaptive control process of the computing equipment is modified.

[0167] In one optional embodiment, by statistically analyzing and predicting the trends of temperature regulation results, disturbance regulation results, and energy consumption optimization adaptive control results, the judgment threshold and control parameters can be automatically corrected, enabling the system to have adaptive optimization capabilities during long-term operation.

[0168] In one example, an environmental adaptation and heat dissipation control method for artificial intelligence computing devices suitable for extreme environments is provided. This method addresses the problems that AI chips are prone to, such as insufficient heat dissipation, large temperature fluctuations, and unstable operation, under complex operating conditions including high temperature, low temperature, high humidity, strong vibration, and electromagnetic interference. Through multi-layered collaborative steps including environmental perception, state recognition, control execution, and feedback optimization, it achieves adaptive adjustment of the device's operating state and optimization of energy consumption.

[0169] The core idea of ​​this example is to continuously sense the external environment and internal temperature during the operation of computing devices, determine the current operating conditions in real time, and automatically select the most suitable heat dissipation or heating strategy to ensure that the system maintains stable temperature, energy efficiency balance, and long-term reliable operation in complex environments. This method can be applied to scenarios such as artificial intelligence servers, edge computing terminals, and high-performance industrial control equipment.

[0170] In this implementation, the environmental perception and data fusion step is performed first. This step involves setting up multiple sensing units to monitor key parameters of the equipment's operating environment in real time, primarily including ambient temperature, humidity, air pressure, vibration intensity, and electromagnetic field strength. Each sensor data point has an independent sampling frequency, and after acquisition, a data fusion algorithm is used to comprehensively calculate and obtain an overall environmental assessment value. This assessment value not only reflects the current environmental state but can also be used to predict environmental change trends. For example, when the external temperature rises rapidly and humidity is high, the system can enter an early warning state to prepare for subsequent heat dissipation control. The key technology of this step lies in using multi-parameter comprehensive judgment, rather than relying on a single temperature signal, thereby significantly improving the accuracy and response speed of operating condition identification.

[0171] The "Environmental Perception and Data Fusion Algorithm" proposed in this example adopts a hierarchical and parameterizable implementation. Within a unified scheduling cycle, the algorithm completes time alignment, robust preprocessing, feature construction, confidence assessment, weighted evidence fusion, and short-term trend prediction of multi-source sensor data, and finally outputs environmental assessment values ​​and related confidence information, providing directly callable quantitative basis for operating condition determination and control strategies.

[0172] In one embodiment, the system first performs time alignment on data from sensor channels such as temperature, humidity, air pressure, vibration intensity, and electromagnetic field intensity. Specifically, the system sets a scheduling period dt (50ms~100ms recommended), and all sensor data are mapped to a discrete-time grid. For slow variables with sampling frequencies below 1 / dt (such as humidity and air pressure), zero-order hold or first-order spline interpolation is used to round them to a uniform frequency; for high-frequency quantities (such as vibration and electromagnetic fields), a short-window mean or peak hold strategy covering 1~2 dt is used to maintain sensitivity to transient disturbances. Through this alignment strategy, the system obtains multi-source input sequences with the same frequency and phase, laying the foundation for subsequent fusion.

[0173] To mitigate the impact of sensor anomalies and sudden spikes on the fusion results, this example performs robust preprocessing on the data from each channel. Preferred methods include: amplitude limiting based on calibration upper and lower limits; using median filtering combined with Hampel identification to remove isolated spikes (the preferred threshold is the local median plus or minus three times the absolute deviation); for channels with continuous packet loss or fixed-value drift, recording the anomaly count and updating the channel health q (value 0~1). When q falls below a preset threshold (preferably 0.6), the weight of that channel in the fusion process will be dynamically reduced or temporarily disabled, thereby ensuring the stability and availability of the fusion calculation under sensor degradation conditions.

[0174] Subsequently, the system constructs time-domain and frequency-domain composite features for the cleaned multi-source sequences to enhance the ability to identify changes in operating conditions and external interference. Taking any channel x as an example, the system calculates the normalized quantity (x minus the historical mean and then divided by the historical standard deviation), the rate of change per unit time (current value minus the previous value and then divided by dt), and the mean, standard deviation, maximum value, and linear trend coefficient over a 2-10 second sliding window. For vibration intensity and electromagnetic field intensity, a short-time Fourier transform is further performed to extract frequency-domain features such as the main peak frequency, in-band root mean square, and spectral kurtosis to characterize scenarios such as resonance, harmonics, or electromagnetic interference. Through the above processing, a feature vector containing multi-source statistical and physical parameters can be obtained at each time step.

[0175] In terms of weight setting, this example introduces a two-level weight model of "static weight plus adaptive correction". Static weights are assigned during factory calibration based on the correlation between each channel and the target task (e.g., heat dissipation safety), preferably with temperature weighting higher than humidity and air pressure. In interference identification tasks, the basic weights of vibration and electromagnetic field channels are relatively increased. Based on this, the system periodically calculates the confidence factor for each channel (including health q, residual stability, real-time correlation with the target variable, or mutual information, etc.), and obtains the adaptive weight using "actual weight = static weight × mapping function (confidence)". The mapping function can be piecewise linear or logistic in form, ensuring that the weight decays rapidly when a channel degrades and smoothly recovers when a channel recovers.

[0176] In terms of the fusion framework selection, this example adopts a two-level structure of "robust weighted evidence fusion plus state-space prediction". The first layer provides a robust score S(t) for the instantaneous environmental state. The preferred implementation is a weighted generalized linear model with confidence constraints: by applying interpretable mappings (such as temperature and humidity coupling terms, threshold piecewise functions, or small-scale shallow networks) to the features of each channel, and then compressing the score to the 0~1 range through adaptive weight summation and Sigmoid-type mapping. To improve the ability to handle conflicting evidence, Dempster-Shafer evidence synthesis can be superimposed on the above model. Confidence levels are assigned to hypotheses such as "overheating", "humidity", and "strong interference", and a joint confidence distribution is obtained through conflict normalization. Finally, the weighted data is backfilled into S(t) and the prior distribution of the operating conditions, thus maintaining stable judgment under strong interference or single-channel anomaly conditions.

[0177] In terms of trend prediction, the second layer employs a state-space model with exogenous input to perform short-term extrapolation of key variables and scores to achieve prediction results for time t+h (preferably h is 1~5 minutes), including S(t+h), temperature prediction, and humidity prediction. Preferably, this is implemented using a Kalman filter or an equivalent linear Gaussian state-space model, whose observation equation consists of S(t) obtained from the first layer fusion and the dominant variable (such as temperature). The state evolution equation introduces the rate of change and seasonal terms (such as the diurnal temperature and humidity cycle). In nonlinear or strongly coupled scenarios, extended Kalman filtering or unscented Kalman filtering can be used to improve the predictive accuracy in situations of rapid temperature rise and high humidity coexistence.

[0178] To achieve real-time indication of anomalies and disturbances, this example integrates the parallel calculation of vibration resonance factor and electromagnetic anomaly factor on the backbone. The vibration resonance factor is based on the ratio of the amplitude of the main peak of the vibration spectrum to the root mean square within the band, combined with the rate of change of the main peak frequency to determine whether mechanical resonance has occurred. The electromagnetic anomaly factor is based on the energy proportion of the electromagnetic spectrum near the power frequency and harmonics, and the amplitude of broadband noise surges to determine the strength of external electromagnetic disturbances. In addition to issuing early warnings upstream, these two types of factors also serve as suppression terms in adaptive weight correction, avoiding misjudging disturbances as actual environmental deterioration.

[0179] In terms of parameter learning and online self-calibration, this example provides a combined mechanism of offline calibration and online fine-tuning. In the offline phase, historical laboratory and field data are used to determine statistical parameters, static weights, and model structure hyperparameters, and cross-validation is used to determine the S(t) threshold and hysteresis band. In the online phase, recent data is continuously accumulated, and stable statistics are updated using an exponentially weighted shift (attenuation factor recommended 0.95~0.995). The inflection point of the weight mapping function is adjusted in small steps to adapt to seasonal variations, sensor aging, and differences in installation location. To prevent accuracy degradation caused by online drift, online updates are constrained by safety boundaries and confidence intervals, and are automatically frozen during periods of abnormal operating conditions.

[0180] Regarding the output interface and connection method, this example generates four types of results in each scheduling cycle: First, the environmental assessment value S(t) (range 0~1, with a confidence interval provided); second, the prior distribution of operating conditions (vector form, representing the joint confidence level of each operating condition, used for upstream mode switching and rule engine); third, short-term prediction S(t+h) and key variable prediction sequences (used for proactive control and preheating of actuators such as fans and pumps); fourth, anomaly factors (vibration resonance factor and electromagnetic anomaly factor, used to directly trigger interference suppression strategies and fault diagnosis processes). All outputs are timestamped and versioned, and can be encapsulated in a lightweight message format at the edge or gateway.

[0181] Preferably, to enhance the interpretability of the algorithm and the efficiency of on-site parameter tuning, this example employs a physics-heuristic design for the key mapping. For example, when a rapid rise in external temperature and high humidity are detected, the contribution of the temperature-humidity coupling term to S(t) will be nonlinearly amplified; when a significant increase in electromagnetic anomaly factors occurs simultaneously, the system maintains a conservative interpretation of electromagnetically related temperature rise anomalies and avoids false overheating strategies through weighted suppression. This interpretable structure, combined with evidence theory, enables the system to output stable conclusions that conform to engineering intuition even when multiple parameters conflict.

[0182] In terms of engineering implementation, the algorithm pipeline in this example supports operation on resource-constrained embedded platforms. Time alignment, robust filtering, and feature computation are implemented using fixed-point friendly methods; the size of the first-layer fusion model is controlled to a parameter count of around 100, and the second-layer state space prediction dimension does not exceed 10. The overall computational complexity and memory usage can meet the requirements for operation on a cycle time of hundreds of milliseconds. To further improve reliability, the system provides a degradation mechanism: when the number of effective sensing channels is lower than a preset threshold, it automatically degrades to a robust univariate or bivariate estimation and increases the output confidence interval to alert the upstream strategy to enter a conservative mode.

[0183] The fusion algorithm in this example achieves high-precision assessment and advanced prediction of environmental conditions under complex operating conditions through an organic combination of "time alignment, robust preprocessing, feature construction, weight adaptation, evidence fusion, and state prediction." This algorithm not only explains the technical effect of "multi-parameter comprehensive judgment" but also provides structured and traceable quantitative evidence for subsequent operating condition identification, early warning triggering, and heat dissipation control strategies, significantly improving the accuracy and response speed of operating condition identification.

[0184] In the operating condition determination and control strategy matching step, the system compares comprehensive environmental information with the equipment's operating status (including chip core temperature, power consumption, voltage, etc.) and determines the operating condition type of the equipment based on the results. This method categorizes operating conditions into four types: normal temperature, overheating, extreme cold, and interference. By setting multi-level threshold ranges and a hysteresis judgment mechanism, frequent switching caused by short-term fluctuations can be avoided. When the ambient temperature exceeds the upper threshold, the system automatically enters "heat dissipation control mode"; when the temperature is below the lower threshold, it enters "heating control mode"; if vibration or electromagnetic signal intensity exceeds the set value, it enters "anti-interference protection mode." During this stage, the system also dynamically adjusts the thresholds based on historical equipment data, giving the judgment logic self-learning characteristics, thus making the control strategy more flexible and adaptable.

[0185] In the heat dissipation control and temperature balancing step, the system executes specific temperature adjustment actions based on the operating condition judgment. This step includes three operating modes: air cooling control, liquid cooling control, and a combination of both. Air cooling control adjusts the cooling fan speed to change the airflow speed, achieving rapid heat dissipation from the equipment surface; liquid cooling control adjusts the coolant flow rate and temperature difference to achieve efficient heat transfer. When the equipment is detected to be operating under high load and high heat flux, the system automatically activates the combined control mode, simultaneously turning on both the air cooling and liquid cooling systems and automatically allocating their operating ratios based on the temperature change rate, maintaining a balance between heat dissipation and energy consumption. Through this dynamic coordination mechanism, this example effectively avoids the "temperature control lag" problem that occurs in traditional systems under high-temperature environments, keeping the chip junction temperature within a safe range. The core technology of this step lies in the real-time prediction and response adjustment of temperature changes. Through continuous sampling and feedback correction, the control process possesses rapid response capability and stability.

[0186] When the system detects that the external environment is at low or extremely cold conditions, it executes heating control and start-up protection steps. This step achieves stable temperature rise of the equipment by setting a heating temperature curve. The system automatically selects the heating intensity based on the ambient temperature difference. When the temperature difference is large, it adopts a "staged heating" strategy, that is, it first performs low-power preheating and then gradually increases the heating output to ensure a smooth temperature rise of the equipment and avoid thermal stress damage caused by sudden heating. Under extremely cold conditions (such as below...), (30℃) Before startup, the system preheats critical areas to the allowable startup temperature range before entering normal operation. This method achieves a smooth temperature transition by controlling the heating rate and time, thereby ensuring safe startup and stable operation of the equipment in low-temperature environments.

[0187] In the vibration and electromagnetic interference protection steps, this example continuously monitors changes in environmental vibration and electromagnetic signals to determine the presence of external disturbances. When the vibration intensity exceeds a set threshold, the system automatically reduces the operating frequency or adjusts task allocation to minimize the impact of mechanical resonance. When an increase or sudden change in electromagnetic field intensity is detected, the system immediately triggers a signal protection program, temporarily shielding highly sensitive communication lines or adjusting signal filtering parameters to reduce the impact of electromagnetic interference on control logic and data transmission. The key technology in this part lies in the adoption of an adaptive response approach, enabling the system to automatically select protective measures based on disturbance characteristics, thereby improving overall anti-interference capability and operational reliability.

[0188] In the energy management and redundancy control steps, this example sets up an energy efficiency analysis and fault switching mechanism to ensure continuous system operation and optimal energy utilization. The energy efficiency analysis section dynamically calculates the cooling efficiency per unit energy consumption by monitoring the operating power and temperature control effect of the heat dissipation device in real time, and automatically adjusts the air cooling and liquid cooling ratio based on the results. When a decrease in power utilization or an abnormality in the performance of a heat dissipation path is detected, the redundancy control module automatically switches to the backup loop to ensure continuous and stable system operation. Simultaneously, the system records each switching event and energy consumption data, providing a reference for subsequent algorithm optimization and achieving self-learning adjustment of energy consumption over operating time, thereby achieving the dual optimization goals of energy saving and reliability.

[0189] Finally, in the feedback and self-learning optimization steps, the system records and analyzes various data collected during operation (including temperature curves, energy consumption data, environmental parameters, and control signals). Through statistical analysis and trend prediction of historical operating data, the control algorithm can automatically correct judgment thresholds and control parameters, enabling the system to have adaptive optimization capabilities over long-term operation. This mechanism allows control accuracy to gradually improve with increasing operating time, thereby realizing an intelligent and continuously optimized temperature control process.

[0190] In summary, the technical solution presented in this example utilizes multiple methods, including environmental monitoring, intelligent judgment, dynamic control, and feedback optimization, to achieve automatic temperature control and stable operation of AI computing devices in complex environments. This method not only improves heat dissipation efficiency and reduces energy consumption but also significantly enhances the reliability of the equipment in high-temperature, low-temperature, and high-interference environments. Its overall structure is clear, the steps are highly implementable, it requires no specific hardware limitations, and it is applicable to a variety of industrial and intelligent computing devices.

[0191] The environmental adaptation and heat dissipation control method for artificial intelligence computing devices in extreme environments provided in this example has achieved significant improvements in many aspects of technical performance compared to existing technologies. Specifically, it has obvious technical and economic advantages in terms of intelligence, adaptability, energy saving and system stability.

[0192] Regarding environmental adaptability, this example establishes a dynamic environmental perception mechanism based on multi-source data fusion, enabling real-time response to changes in the external environment in the system's operating status. Traditional heat dissipation or environmental control methods often employ fixed thresholds or single-point sensor control, resulting in sluggish responses and a tendency to misjudge. This invention combines a weighted fusion algorithm with a self-learning model, accurately assessing comprehensive operating conditions such as temperature, humidity, and electromagnetic interference, achieving millisecond-level response to environmental changes, thereby significantly improving the operational stability and reliability of the equipment. Experimental results show that, with external temperature fluctuations of ±15℃, the system's core temperature fluctuation is controlled within ±2℃, and the environmental response time is shortened by approximately 40% compared to traditional PID control.

[0193] In terms of energy consumption control, this example introduces dynamic power prediction and redundancy management algorithms to automatically adjust energy allocation based on load change trends, achieving on-demand control of heat dissipation and heating power. Continuous optimization of the operating state through an energy efficiency learning model effectively reduces ineffective heat dissipation and overheating. Real-world test data shows that in typical AI computing task environments, this method achieves an average energy saving of approximately 18% to 25% compared to the traditional constant power mode, and improves the performance per watt by approximately 22%, providing significant energy-saving potential for data centers and high-power devices.

[0194] Regarding system safety and anti-interference capabilities, this example establishes an active protection mechanism against vibration and electromagnetic interference through a signal analysis and state recognition module. This mechanism can identify abnormal spectral characteristics and automatically switch to a safe mode. This mechanism effectively reduces the risk of equipment malfunction caused by structural resonance and electromagnetic shock. Test results show that after adopting the method of this invention, the system's average immunity to electromagnetic pulse interference increases by approximately 1.6 times, and the operation interruption rate decreases by approximately 70%.

[0195] Furthermore, the self-learning feedback mechanism in this example enables long-term adaptive optimization. By modeling and analyzing historical operating data, the control strategy can be dynamically adjusted according to environmental characteristics and equipment aging status, thereby maintaining high control accuracy and thermal balance efficiency across different service cycles. This long-term self-adjustment characteristic allows the system to achieve an intelligent evolution effect of "the longer it runs, the more precise the control," significantly extending equipment lifespan and reducing maintenance costs.

[0196] From a social and economic perspective, this example can be deployed in existing computing or control systems without complex hardware modifications, exhibiting excellent compatibility and scalability. Its application can not only reduce energy consumption and carbon emissions but also improve the reliable operation of critical equipment in extreme environments, making it of significant promotional value in the fields of energy conservation, environmental protection, and intelligent manufacturing.

[0197] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention.

[0198] The following is a detailed reference. Figure 3 The diagram illustrates a structural schematic suitable for implementing an electronic device according to embodiments of the present invention. The electronic device may include a processor (e.g., a central processing unit, graphics processor, etc.) 301, which can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) 302 or a program loaded from memory 308 into random access memory (RAM) 303. The RAM 303 also stores various programs and data required for the operation of the electronic device. The processor 301, ROM 302, and RAM 303 are interconnected via a bus 304. An input / output (I / O) interface 305 is also connected to the bus 304.

[0199] Typically, the following devices can be connected to I / O interface 305: input devices 306 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 307 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; memory devices 308 including, for example, magnetic tapes, hard disks, etc.; and communication devices 309. Communication device 309 allows electronic devices to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 3 Electronic devices with various devices are shown, but it should be understood that it is not required to implement or have all of the devices shown, and more or fewer devices may be implemented or have instead.

[0200] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 309, or installed from a memory 308, or installed from a ROM 302. When the computer program is executed by the processor 301, it performs the functions defined in the adaptive control method for optimizing computing power device energy consumption according to embodiments of the present invention.

[0201] Figure 3 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments of the present invention.

[0202] This invention also provides a computer-readable storage medium. The methods described above according to embodiments of the invention can be implemented in hardware or firmware, or implemented as computer code that can be recorded on a storage medium, or implemented as computer code downloaded via a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and then stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code. When the software or computer code is accessed and executed by the computer, processor, or hardware, the adaptive control method for optimizing computing power consumption shown in the above embodiments is implemented.

[0203] A portion of this invention can be applied as a computer program product, such as computer program instructions, which, when executed by a computer, can invoke or provide the methods and / or technical solutions according to the invention through the operation of the computer. Those skilled in the art will understand that the forms in which computer program instructions exist in a computer-readable medium include, but are not limited to, source files, executable files, installation package files, etc. Correspondingly, the ways in which computer program instructions are executed by a computer include, but are not limited to: the computer directly executing the instructions, or the computer compiling the instructions and then executing the corresponding compiled program, or the computer reading and executing the instructions, or the computer reading and installing the instructions and then executing the corresponding installed program. Here, the computer-readable medium can be any available computer-readable storage medium or communication medium accessible to a computer.

[0204] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A method for adaptive control of energy consumption optimization in computing equipment, characterized in that, The method is used for a main control unit, which is connected to a temperature regulation device, a protection device, and an energy consumption and redundancy management module; the main control unit is used for a main control unit, which is connected to a temperature regulation device, a protection device, and an energy consumption and redundancy management module respectively; the method includes: Obtain the initial operating environment parameter set of the computing power device; Based on the initial set of operating environment parameters, the target environment evaluation value is obtained through data fusion algorithm calculation; Based on the initial operating environment parameter set and the target environment evaluation value, the temperature control mode and anti-interference protection mode of the computing power device are determined; Using the temperature control mode, the temperature adjustment device is controlled to adjust the temperature of the computing power device, and the temperature adjustment result of the computing power device is obtained; Using the aforementioned anti-interference protection mode, the protection device is controlled to adjust external disturbances to the computing power equipment, thereby obtaining the interference adjustment result of the computing power equipment; Using the temperature regulation results and the interference regulation results, the energy consumption and redundancy management module is controlled to adjust the energy consumption and redundancy of the computing device, thereby obtaining the energy consumption optimization adaptive control result of the computing device.

2. The method according to claim 1, characterized in that, Based on the initial operating environment parameter set, the target environment evaluation value is obtained through data fusion algorithm calculation, including: The initial runtime environment parameter set is time-aligned to obtain the first runtime environment parameter set; The first set of operating environment parameters is subjected to robust preprocessing to obtain the target set of operating environment parameters and multiple channel health values. The target operating environment parameter set is subjected to time-domain and frequency-domain feature extraction to obtain a composite feature set; Based on the composite feature set and the multiple channel health values, determine the actual weight values ​​of multiple channels; Based on the composite feature set and the actual weight values ​​of the multiple channels, the target environment assessment value is obtained through a two-layer fusion model based on weighted evidence fusion and state space prediction.

3. The method according to claim 1, characterized in that, Based on the initial operating environment parameter set and the target environment assessment value, the temperature control mode and anti-interference protection mode are determined, including: When the ambient temperature value in the initial operating environment parameter set is greater than the target environment assessment value, the temperature control mode is determined to be the heat dissipation control mode. When the ambient temperature value is lower than the target environmental assessment value, the temperature control mode is determined to be a heating control mode. When the vibration intensity value in the initial operating environment parameter set is greater than a preset threshold, the anti-interference protection mode is determined to be a vibration protection mode. When the electromagnetic field strength in the initial operating environment parameter set does not meet the preset conditions, the anti-interference protection mode is determined to be the electromagnetic protection mode.

4. The method according to claim 3, characterized in that, The temperature regulation device includes a heat dissipation execution module; using the temperature control mode, the temperature regulation device is controlled to regulate the temperature of the computing power equipment, and the temperature regulation result of the computing power equipment is obtained, including: When the temperature control mode is the heat dissipation control mode, multiple working modes of the heat dissipation control mode are obtained, including air cooling control mode, liquid cooling control mode, and air cooling and liquid cooling combined control mode. When the computing power device meets the first operating state, the heat dissipation execution module is controlled to adjust the temperature of the computing power device using the air-cooling control working mode. When the air-cooling control mode cannot meet the temperature regulation requirements of the computing device, the liquid-cooling control mode is used to control the heat dissipation execution module to regulate the temperature of the computing device. When the computing power device does not meet the first operating state, the cooling and liquid cooling combined control working mode is used to control the heat dissipation execution module to adjust the temperature of the computing power device, thereby obtaining the temperature adjustment result of the computing power device.

5. The method according to claim 3, characterized in that, The temperature regulation device further includes an extreme cold heating module; using the temperature control mode to control the temperature regulation device to regulate the temperature of the computing power equipment, and obtaining the temperature regulation result of the computing power equipment, it also includes: When the temperature control mode is the heating control mode, the temperature difference is determined based on the ambient temperature value and the operating temperature of the computing device; When the temperature difference meets the preset temperature difference condition, based on the preset heating temperature curve, the heating control mode is used to control the extreme cold heating module to adjust the temperature of the computing device, and the temperature adjustment result of the computing device is obtained. When the temperature difference does not meet the preset temperature difference condition but the ambient temperature meets the preset temperature condition, the heating control mode is used to control the extreme cold heating module to adjust the temperature of the computing device until the temperature difference meets the preset temperature difference condition, and the temperature adjustment result is obtained.

6. The method according to claim 3, characterized in that, The protective device includes a vibration protection module and an electromagnetic protection module; using the anti-interference protection mode, the protective device is controlled to adjust external disturbances to the computing power equipment, resulting in interference adjustment of the computing power equipment, including: When the anti-interference protection mode is the vibration protection mode, the vibration protection module is controlled to reduce the operating frequency of the computing device or adjust the tasks allocated to the computing device, and the interference adjustment result of the computing device is obtained. When the anti-interference protection mode is the electromagnetic protection mode, the electromagnetic protection module is controlled to process the target communication line or signal filtering parameters of the computing power device and obtain the interference adjustment result of the computing power device.

7. The method according to claim 1, characterized in that, Using the temperature regulation results and the interference regulation results, the energy consumption and redundancy management module is controlled to adjust the energy consumption and redundancy of the computing power device, thereby obtaining the energy consumption optimization adaptive control result of the computing power device, including: The cooling efficiency of the computing device is determined based on the temperature adjustment results and the interference adjustment results. Based on the cooling efficiency, the energy consumption and redundancy management module adjusts the air cooling and liquid cooling ratio of the computing device and obtains the power utilization rate and heat dissipation path performance of the computing device. When the power utilization rate or the heat dissipation path performance does not meet the preset conditions, the energy consumption and redundancy management module is controlled to switch to the backup circuit. The backup circuit is used to control the operation of the computing device, and the energy consumption optimization adaptive control result is obtained.

8. The method according to claim 1, characterized in that, The method further includes: Based on the temperature adjustment results, the interference adjustment results, and the energy consumption optimization adaptive control results, the energy consumption optimization adaptive control process of the computing device is corrected.

9. An adaptive control system for optimizing energy consumption of computing equipment, characterized in that, The system includes: a main control unit, an environmental sensing module, a temperature regulating device, a protective device, an energy consumption and redundancy management module, and a power management unit. The main control unit is connected to the environmental sensing module, the temperature regulating device, the protective device, the energy consumption and redundancy management module, and the power management unit. The power management unit is also connected to the energy consumption and redundancy management module. The temperature regulation device includes a heat dissipation execution module and an extreme cold heating module; the protection device includes a vibration protection module and an electromagnetic protection module. The environment perception module is used to acquire the operating environment parameter set of the computing power device and send the operating environment parameter set to the main control unit; The main control unit is used to execute the energy consumption optimization adaptive control method according to any one of claims 1 to 8.

10. A computer program product, characterized in that, It includes computer instructions for causing a computer to execute the energy consumption optimization adaptive control method according to any one of claims 1 to 8.

Images