Thermal management control method, system and computer device based on adaptive fuzzy logic

Abstract

The application discloses a heat management control method, system and computer equipment based on adaptive fuzzy logic. The application fuses adaptive fuzzy sets, multi-target game decision and extended state observer technology, realizes accurate disturbance tracing through constructing a disturbance causal graph, generates Pareto optimal control parameters with the aid of a four-party game architecture, and dynamically compensates control actions with a three-layer ESO. The implementation process includes signal monitoring and causal graph generation, game decision and parameter optimization, observer execution and compensation, and effect evaluation and self-tuning closed loop. The application significantly improves the dynamic balance capability of the system among stability, response speed and energy efficiency, and has strong adaptability and robustness.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of hydropower generation equipment state monitoring and fault diagnosis, in particular to a thermal management control method and system based on adaptive fuzzy logic and a computer device. BACKGROUND

[0002] In complex systems such as electronic devices, energy systems, and vehicle thermal management, thermal management control strategies have a decisive influence on system stability, energy efficiency, and response speed. The classical PID control method widely used at present shows obvious lack of adaptability when facing nonlinear, time-varying, and multi-disturbance coupled thermal management systems. Experimental data show that when the load power undergoes a step change, the adjustment time of the traditional PID controller generally exceeds 10 seconds, and the overshoot is often more than 15%, which is difficult to meet the rapid stability demand under high dynamic conditions. In addition, the PID controller parameters are usually fixed values, which cannot be adaptively adjusted according to environmental temperature gradient, cooling medium flow rate changes, and other disturbance sources, resulting in a significant decline in control performance under variable conditions and an increase in system operation risk.

[0003] To further improve control performance, some studies have introduced a fuzzy logic control method based on a fixed rule base. This method improves the system's nonlinear processing ability to some extent, but its rule base and membership functions are usually pre-set, lacking an online self-tuning mechanism. For example, when the system is simultaneously subjected to a sudden drop in environmental temperature and an internal load surge, the fixed rule fuzzy controller often produces sustained oscillation due to its inability to accurately quantify the disturbance priority and coupling effect. Test data show that its steady-state error fluctuation range can reach ±2.5°C, seriously affecting control accuracy and equipment life. At the same time, this method generally lacks the ability to semantically attribute and assess the confidence of disturbance sources, resulting in a lack of interpretability and pertinence in control decisions. SUMMARY

[0004] To address the deficiencies in dynamic response delay, poor adaptability to multiple disturbances, and difficulty in balancing stability and energy efficiency of existing technologies, the present application proposes a thermal management control method and system based on adaptive fuzzy logic and a computer device, which realizes accurate identification and semantic attribution of multiple source disturbances by constructing a disturbance causal graph, generates a multi-objective coordinated parameter package that takes into account stability, response speed, and energy efficiency by introducing a four-party game architecture, and realizes dynamic compensation and adaptive execution of control actions by means of a three-layer observer structure, thereby achieving rapid response, high-precision stability, and multi-objective coordinated optimization of the thermal management system under complex operating conditions.

[0005] The technical solutions of the present application are as follows:

[0006] According to an aspect of the present application, a thermal management control method based on adaptive fuzzy logic is provided, comprising:

[0007] The controller monitors multi-source signals in real time and triggers the start of the control cycle when any signal exceeds a preset sensitivity threshold; the adaptive fuzzy preprocessor in the controller delineates the affected area; the fuzzy causal inference engine in the controller judges the type of disturbance source; the fuzzy priority arbitrator in the controller assigns priority weights to various disturbance source types; and the disturbance type, affected area, and priority weights are integrated to generate a disturbance causal map.

[0008] After the controller loads the disturbance causal graph, it initiates a four-party game architecture consisting of a stability guardian, a response accelerator, an energy efficiency arbitrator, and a fuzzy semantic coordinator. The game process follows four steps: parameter proposal, arbitration correction, four-party compromise, and proposal encapsulation, and outputs a semantic control parameter package.

[0009] Configure three-layer observers, ESO-A, ESO-B, and ESO-C, in the controller, so that the three-layer observers focus on parameter proposal execution, thermal inertia response, and environmental disturbance response, respectively; after loading the semantic control parameter package in the controller, start the three-layer observers to execute the semantic control parameter package;

[0010] The controller performs a comprehensive evaluation of the observer's performance; if it is not satisfied with the result of the comprehensive evaluation, it adaptively adjusts the compromise strategy of the four-way compromise through a fuzzy self-tuner to achieve dynamic balance of the control system.

[0011] As a further option of the method of the present invention, the specific method by which the adaptive fuzzy preprocessor delineates the influence region includes:

[0012] The membership function parameters of each input signal are dynamically adjusted based on historical operating data; for the first... Input signal Its corresponding fuzzy subset Membership function center and width Update as follows: ; ;

[0013] in, For fuzzy subset index, For time series indexing, and Let these represent the expected value and variance, respectively. It is a forgetting factor.

[0014] As a further option of the method of the present invention, the fuzzy causal inference engine determines the type of disturbance source and outputs the first... Confidence score of disturbance source The calculation formula is as follows: ; wherein is the output variable of the inference engine, is the th disturbance source of the th inference rule, is the weight of the th inference rule, is the fuzzy subset of the th inference rule.

[0015] As a further option of the method of the present invention, the fuzzy priority arbiter assigns a priority weight to the th disturbance source of the th inference rule, which is calculated as: ; wherein is a specific disturbance source weight adjustment function related to the system health, is the system health.

[0016] As a further option of the method of the present invention, the target functions of the stability guardian, the response accelerator, and the energy efficiency arbiter are respectively , , wherein is the control parameter vector;

[0017] The stability guardian proposes a control parameter proposal aiming at maintaining system stability based on the disturbance causal graph data;

[0018] The response accelerator proposes a control parameter proposal aiming at improving system dynamic response speed based on the disturbance causal graph data;

[0019] The energy efficiency arbiter proposes a control parameter proposal aiming at optimizing system energy efficiency based on the disturbance causal graph data;

[0020] The fuzzy semantic coordinator arbitrates and revises the parameter proposals of the above three parties based on safety constraints;

[0021] Wherein, the arbitration revision process of the fuzzy semantic coordinator is modeled as the following optimization problem with constraints to seek a compromise solution: ; wherein ; wherein and are the ideal value and the worst value of the th objective, is the bargaining weight, is the safety constraint introduced by the fuzzy semantic coordinator, is , , the specific calculation formula of

[0022] As a further option of the method of the present application, the ESO-B translates the parameter proposals in the semantic control parameter package into concrete control actions, including:

[0023] The abstract parameters in the semantic control parameter package are translated into concrete control actions by the ESO-B, including: The standard PID control law is calculated: ; where, is the time instant, the primary control action obtained by the ESO-A, the temperature deviation, , , are the proportional, integral and derivative gains, respectively, is the integral variable; the ESO-A outputs preliminary calculation results .

[0024] As a further option of the method of the present application, the ESO-B is used to estimate and compensate thermal inertia disturbances, which generates a first compensation control quantity is iteratively calculated by the following formula:

[0025] ;

[0026] ;

[0027] ;

[0028] ;

[0029] where, is the defined observation error, i.e. the difference between the temperature tracked by the observer and the actual temperature ; is used to track the real output of the system , whose update depends on the observation error , the primary control action and the estimated total disturbance , , is the observer gain, is the compensation coefficient of the ESO, is a nonlinear function, is the power of the function, is the linear interval width of the function; is the compensation control quantity .

[0030] As a further option of the method of the present application, the ESO-C is used to estimate and compensate environmental disturbances, generating a second compensation control quantity identical to the first compensation control quantity generated by the ESO-B.

[0031] As a further option of the method of the present application, the comprehensive evaluation at least includes quantitative evaluation of stability index, dynamic response speed index and overall energy efficiency index of system output; the fuzzy self-tuner outputs an adjustment factor of the game weight in the next control cycle according to the deviation mode of the evaluation result and the game process summary , and updates it .

[0032] Another aspect of the present application provides a thermal management control system based on adaptive fuzzy logic, which comprises:

[0033] a signal acquisition module for real-time monitoring of multi-source signals, and triggering a control cycle when the value of any signal exceeds a preset sensitive threshold;

[0034] a disturbance causal graph generation module including an adaptive fuzzy preprocessor, a fuzzy causal inference machine and a fuzzy priority arbiter; the adaptive fuzzy preprocessor is used to process the signals exceeding the threshold and dynamically delimit the influence area; the fuzzy causal inference machine is used to judge the type of disturbance source; the fuzzy priority arbiter is used to assign dynamic priority weights to various types of disturbance sources; the disturbance causal graph generation module is used to integrate the disturbance type, the influence area and the priority weight, and generate a disturbance causal graph;

[0035] a four-party game decision module connected to the disturbance causal graph generation module, used to load the disturbance causal graph and including a stability guardian unit, a response accelerator unit, an energy efficiency arbiter unit and a fuzzy semantic coordinator unit; the stability guardian unit, the response accelerator unit and the energy efficiency arbiter unit are used to generate independent control parameter proposals based on the disturbance causal graph from the stability, response speed and energy efficiency targets, respectively; the fuzzy semantic coordinator unit is used to arbitrate and correct the proposals, and guide the four-party units to generate a semantic control parameter package through a compromise process;

[0036] ​​The multi-observer cooperative execution module is connected with the four-party game decision module, and includes an ESO-A observer, an ESO-B observer and an ESO-C observer; the ESO-A observer is used for loading and executing a main control law in a semantic control parameter package; the ESO-B observer is used for estimating system thermal inertia and generating a first compensation control amount; the ESO-C observer is used for estimating environmental disturbance and generating a second compensation control amount; and the multi-observer cooperative execution module is used for synthesizing the main control action and the compensation control amounts, and outputting a final control signal to the controlled object.

[0037] The closed-loop evaluation and self-tuning module is connected with the multi-observer cooperative execution module and the four-party game decision module, respectively, and is used for comprehensively evaluating the execution effect of the final control signal, and when the evaluation result does not meet the preset requirement, adaptively adjusting compromise strategy parameters in the four-party game decision module through a fuzzy self-tuner to analyze performance deviation reasons, so as to realize dynamic balance and continuous optimization of the control system.

[0038] According to an aspect of the present application, a computer device is provided, which comprises a processor and a memory, and the memory stores at least one instruction, at least one program, a code set or an instruction set, which is loaded and executed by the processor to implement the adaptive fuzzy logic-based thermal management control method as described in the above aspect.

[0039] The present application has the following beneficial effects:

[0040] The present method is significantly superior to the traditional PID control in terms of dynamic response and stability. Through the four-party game architecture and the ESO real-time compensation, the system can achieve an optimal balance among multiple targets, and theoretically, the regulation time can be greatly shortened from more than 10 seconds of the traditional PID, and the overshoot can be reduced from more than 15% to within 10°C, effectively suppressing control oscillation and improving device safety and service life.

[0041] The present method has strong adaptability and precise disturbance processing capability, overcoming the limitations of fixed rule fuzzy control. The membership function can be dynamically adjusted according to historical data, and combined with the disturbance causal graph to realize semantic attribution and confidence quantification, so that in the case of multiple source disturbances (such as simultaneous occurrence of environmental sudden change and load surge), the priority can be accurately quantified and cooperative decision can be made, and the steady-state error fluctuation range can be narrowed from ±2.5°C, significantly improving control accuracy.

[0042] The present method realizes continuous optimization and strong robustness of the system through closed-loop evaluation and fuzzy self-tuning. The three-level ESO compensation mechanism can effectively estimate and offset internal and external disturbances in the system, and the execution effect evaluation and game weight online adjustment strategy ensures that the system can maintain dynamic optimal performance under variable working conditions, ensuring energy efficiency and stability during long-term operation. BRIEF DESCRIPTION OF DRAWINGS

[0043] Figure 1 The overall flowchart of the heat management control method based on adaptive fuzzy logic is shown in the figure.

[0044] Figure 2 The detailed flowchart of step S100 of the heat management control method based on adaptive fuzzy logic is shown in the figure.

[0045] Figure 3 The detailed flowchart of step S200 of the heat management control method based on adaptive fuzzy logic is shown in the figure.

[0046] Figure 4 The detailed flowchart of step S300 of the heat management control method based on adaptive fuzzy logic is shown in the figure.

[0047] Figure 5 The detailed flowchart of step S400 of the heat management control method based on adaptive fuzzy logic is shown in the figure. DETAILED DESCRIPTION

[0048] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0049] With the development of electronic devices towards high power density and miniaturization, the thermal management system is facing unprecedented challenges. The traditional PID control or fuzzy control method based on fixed rule base is difficult to achieve dynamic optimal balance between system stability, dynamic response speed and energy efficiency when facing nonlinear, time-varying and multi-disturbance coupled thermal management objects. The existing technology often lacks the ability of collaborative analysis and collaborative decision-making of multi-source disturbance, resulting in slow response, large overshoot or high energy consumption of the control system. For example, when the processor load suddenly increases and the ambient temperature suddenly drops at the same time, the traditional method is prone to cause control oscillation. The measured data shows that the time for the system to recover to steady state may be as long as tens of seconds, and the overshoot is more than 10℃, which seriously threatens the safety and life of the equipment.

[0050] The core theoretical basis of the present application is built on three pillars of adaptive fuzzy set theory, multi-objective game decision-making theory and extended state observer (ESO) compensation theory. By constructing a disturbance causal graph, the precise tracing and quantitative evaluation of the disturbance are realized. With the help of a four-party game architecture, the collaborative control parameters in the sense of Pareto optimality are generated, and the dynamic compensation of the control action is realized by using a three-layer ESO observer, so as to realize the intelligent, stable and efficient operation of the thermal management system under multi-disturbance and variable working conditions.

[0051] The derivation of the core theoretical formula is as follows:

[0052] Dynamic adjustment of membership function in adaptive fuzzy preprocessing:

[0053] Let the first The input signals are Its corresponding adaptive membership function Represented as: ;in, It is the first Input signal Belongs to its fuzzy subset The membership degree of , with a value range of [0,1]. For fuzzy subset indexing, The center of the membership function, The width is [specified]. The adaptive mechanism dynamically adjusts [the width] based on the historical data distribution. and :

[0054] ;

[0055] ;

[0056] in, For time series indexing, and Let these represent the expected value and variance, respectively. It is a forgetting factor used to balance the influence of historical and current data. For the first A typical system operating condition.

[0057] Confidence and priority quantification of perturbation causal graphs:

[0058] The output of the fuzzy causal inference engine Confidence score of disturbance source The calculation is as follows: ;in, It is the output variable of the inference engine. It is the first The first type of disturbance source Fuzzy subsets after the inference rules It is the first The weight of each inference rule, It is a fuzzy subset Adaptive membership function. Fuzzy priority arbitrator combined with system health. Priority weight assigned for: ;in, It is a specific disturbance source weight adjustment function related to the system health.

[0059] Objective function of the four-party game and compromise solution:

[0060] Let the objective functions of the stability guardian, response accelerator, and energy efficiency mediator be , , , where is the control parameter vector. The mediation modification process of the fuzzy semantic coordinator is modeled as a constrained optimization problem: ; ; where and are the ideal value and the worst value of the th objective, is the weight of the game, is the safety constraint introduced by the fuzzy semantic coordinator. This model seeks a compromise solution that is relatively satisfactory to all three parties by minimizing the weighted sum.

[0061] Compensation control amount calculation of extended state observer (ESO):

[0062] The first compensation control amount for the control action is estimated and generated by the following ESO: ; ; ; where is the observation error, is the first state variable of the ESO, used to track the system output , is the second state variable of the ESO, and are the derivatives of the state variables and with respect to time, and are the observer gains of the ESO-B, is the compensation coefficient of the ESO, is a nonlinear function used to improve the performance of the observer in different error ranges, is the power of the function, is the linear interval width of the function.

[0063] The final compensation control amount is: , where is the compensation control amount generated by the ESO.

[0064] The above theoretical framework provides a solid theoretical basis for the invention, ensuring the precision and adaptability of the thermal management control process.

[0065] The specific embodiments of the present application will be described in detail below.

[0066] Embodiment one:

[0067] Please refer to Figure 1 , which shows the overall flowchart of a thermal management control method based on adaptive fuzzy logic provided by an embodiment of the present application, the method comprising:

[0068] S100: The controller monitors multiple source signals in real time, and triggers a control cycle when any signal exceeds a preset sensitive threshold, and generates a disturbance causal graph through adaptive fuzzy preprocessing, causal reasoning and priority arbitration.

[0069] S200: Based on the disturbance causal graph, a four-party game architecture composed of a stability guardian, a response accelerator, an energy efficiency arbitrator and a fuzzy semantic coordinator is started, and after multiple rounds of proposal, arbitration and compromise, a semantic control parameter package is output.

[0070] S300: The controller loads the semantic control parameter package and starts the three-layer observer ESO-A (parameter proposal observer), ESO-B (thermal inertia observer) and ESO-C (environmental observer), converts the parameter proposal into specific control actions, and injects thermal inertia and environmental disturbance compensation control quantities respectively.

[0071] S400: The controller comprehensively evaluates the control effect of the observer, and if the evaluation result does not meet the requirements, the compromise strategy of the four-party game is adaptively adjusted through the fuzzy self-tuner to realize the dynamic balance and continuous optimization of the control system.

[0072] The specific scheme is as follows:

[0073] In a thermal management control method based on adaptive fuzzy logic, S100 provides disturbance trend perception for the whole control decision through high-frequency, multi-channel signal acquisition and intelligent semantic analysis.

[0074] Please refer to Figure 2 , which shows the flowchart of an exemplary thermal management control method S100 based on adaptive fuzzy logic, the content of which includes:

[0075] S110: Specifically, the controller synchronously collects multiple source signals such as environmental temperature gradient, load power jump rate, cooling medium state, observer health parameters, etc. with fixed sampling frequency through integrated high-precision sensor array.

[0076] In one possible implementation, each signal channel is preset with a sensitive threshold, which is dynamically fine-tuned according to the system operation mode based on historical statistical data. The controller continuously compares the instantaneous value of each signal with its threshold sensitive threshold. When there is at least one signal that meets the instantaneous value greater than its sensitive threshold, a new control cycle is triggered immediately, starting the subsequent analysis process.

[0077] S120: Specifically, when the control cycle is triggered, the adaptive fuzzy preprocessor first preprocesses the hypersensitive threshold signal, and dynamically adjusts the influence boundary of each signal on the system state, thereby defining the influence area.

[0078] In one possible implementation, the step of dynamically defining the influence area based on the adaptive fuzzy preprocessing includes:

[0079] The preprocessor matches the current multi-source signal vector with the pre-stored multiple typical working condition baselines, and finds the most similar baseline working condition.

[0080] According to the historical statistical distribution of the baseline working condition matched, the expected value and the variance of the signal under the baseline working condition are calculated.

[0081] According to the core theoretical formula, the membership function center and width of the corresponding signal are calculated, wherein the forgetting factors and are both valued at 0.85, so that the semantic definitions of the fuzzy system such as large, small, fast and slow can be adapted to the current operating environment.

[0082] Using the adjusted membership function , the membership value of each signal at the current time is calculated to form a fuzzy influence vector. By calculating the confidence ellipse or confidence hyper-ellipsoid of the fuzzy influence vector in the system state space, a multi-dimensional influence area is defined.

[0083] S130: Specifically, the fuzzy causal reasoning machine receives the signal and its influence area information from the preprocessor, performs semantic attribution analysis on the disturbance source, judges the disturbance type and outputs the confidence score.

[0084] In one possible implementation, the fuzzy causal reasoning machine is embedded with a rule base of several IF-THEN rules. For example, IF is NB (Negative Big) AND is ZO (Zero) THEN Perturbation_Type is Env_Cooling_Sudden. In this IF-THEN rule, is the environmental temperature gradient, is the load power ramp rate. The inference process adopts the Mamdani fuzzy inference method, and the output fuzzy set about the perturbation type is obtained through fuzzy composition operation . Finally, the fuzzy set is defuzzified to obtain the clear value , and the confidence score of the current perturbation type is calculated according to the calculation formula derived in the core theory , that is . This process converts abstract and coupled physical signals into perturbation type judgments with clear semantics and credibility.

[0085] S140: Specifically, the fuzzy priority arbiter combines the real-time system health context, and uses the priority weight calculation formula in the theoretical derivation: to give dynamic priority weights to each identified perturbation source, and finally integrate all information to generate a perturbation causal graph.

[0086] In one possible implementation, the system health is calculated by factors such as key component life degradation, historical failure records, and current load rate, and is normalized to the [0, 1] interval. The smaller the value, the worse the health.

[0087] In one possible implementation, the perturbation causal graph at least contains the following elements:

[0088] a list of identified perturbation source types , a confidence score corresponding to each type , a priority weight corresponding to each type , an impact area delineated by the preprocessor , and the causal correlation strength between each perturbation source.

[0089] In one heat management control method based on adaptive fuzzy logic, S200 takes the perturbation causal graph generated by S100 as the core input, starts a simulated four-party game architecture, and generates a set of optimal semantic control parameters through multi-objective coordination and arbitration compromise.

[0090] Please refer to Figure 3 , which shows a flowchart of one example heat management control method S200 based on adaptive fuzzy logic according to the present application, the content of which includes:

[0091] S210: Specifically, the stability daemon, the response accelerator, and the energy efficiency mediator, based on the same perturbation causal graph, independently propose preliminary control parameter proposals from their respective objective functions.

[0092] In one possible implementation,

[0093] The stability daemon aims to maintain the system temperature stability, minimizing overshoot and oscillation. Its objective function is defined as where , is the weight. The stability daemon control parameter proposal is obtained by minimizing .

[0094] The response accelerator aims to improve the dynamic response speed of the system to perturbations. Its objective function is defined as , is the adjustment time. The response accelerator control parameter proposal is obtained by minimizing .

[0095] The energy efficiency mediator aims to optimize the overall energy efficiency of the system. Its objective function is defined as , is the total cooling energy consumption. The energy efficiency mediator control parameter proposal is obtained by minimizing .

[0096] The stability daemon, the response accelerator, and the energy efficiency mediator generate their initial proposals based on the corresponding objective functions and the built-in corresponding optimization algorithms.

[0097] S220: Specifically, the fuzzy semantic coordinator, as an arbitrator, first checks the initial proposals proposed by the stability daemon, the response accelerator, and the energy efficiency mediator, and makes preliminary corrections based on safety constraints.

[0098] In one possible implementation, the fuzzy semantic coordinator has a built-in safety rule library. For example: IF THEN SET WITH High_Confidence. The coordinator uses fuzzy logic to evaluate the safety index of each proposal, and limits or corrects the parameters that do not meet the safety threshold, obtaining the corrected proposal set . All triggered safety rules are formalized as inequality constraints .

[0099] S230: Specifically, based on the revised proposal from the arbitration, the four parties enter into a multi-round compromise process aimed at finding a compromise acceptable to all parties.

[0100] In one possible implementation, the compromise solution is achieved by solving the following constrained min-max optimization problem: ; ;in, It is the first The ideal value for each objective is taken from the historical best value. It is the first The worst-case value for each objective is taken from the worst-case value among the initial proposals of the three parties. Game weights It is not fixed, but rather associated with the perturbation priority weights generated in S100.

[0101] The optimization problem is solved using a constrained nonlinear programming solver, ultimately yielding a compromise parameter vector. The compromised control parameter vector Together with the relevant metadata, it is encapsulated into a complete semantic control parameter package.

[0102] In a thermal management control method based on adaptive fuzzy logic, S300 is responsible for accurately and robustly converting the semantic control parameter package generated by S200 into control actions in the physical world, and responding to the system's internal dynamics and external disturbances through a compensation mechanism.

[0103] Please refer to Figure 4 The diagram illustrates a flowchart of an exemplary thermal management control method S300 based on adaptive fuzzy logic, the contents of which include:

[0104] S310: Specifically, the controller configures three layers of observers, ESO-A, ESO-B, and ESO-C, and loads the semantic control parameter package generated by S200 into the shared memory area of ​​the three layers of observers at the beginning of the control cycle.

[0105] In one possible implementation, ESO-A serves as the core controller, directly executing control laws such as PID and fuzzy PID; ESO-B and ESO-C are compensators, each internally built with an extended state observer model. During initialization, the three-layer observer is fine-tuned based on the fuzzy applicability tags in the parameter package to better adapt to the current operating conditions.

[0106] S320: Specifically, ESO-A, as a parameter proposal observer, has the core responsibility of abstracting parameters from the parameter package. Transform into specific, executable control actions .

[0107] In one possible implementation, if is a set of PID parameters , the ESO-A performs a standard PID control law calculation: ; where is the primary control action obtained by the ESO-A at time , is the temperature error, , , are the proportional gain, integral gain, and derivative gain, respectively, is the integral variable. The ESO-A outputs a preliminary result .

[0108] S330: Specifically, the ESO-B, as a thermal inertia observer, observes the system in real time, estimates and compensates for the thermal inertia effect caused by the system's thermal capacity, thermal resistance, etc., which is regarded as an internal disturbance of the system.

[0109] In one possible implementation, the ESO-B runs through the theoretical formula for iteration:

[0110] ;

[0111] ;

[0112] ;

[0113] ;

[0114] wherein the first line of the formula defines the observation error , i.e., the difference between the temperature tracked by the observer and the actual temperature ; the second and third lines of the formula are the core update law of the observer, used to track the real output of the system, which is updated in dependence of the observation error , the primary control action , and the estimated total disturbance , , is the observer gain, is the compensation coefficient of the ESO, is a nonlinear function, is the power of the function, is the linear interval width of the function. The fourth line of the formula is the compensation control amount generation formula, which reverses and normalizes the estimated total disturbance by the compensation coefficient to obtain the compensation control amount The compensation control quantity is directly added to the main control quantity to offset the disturbance.

[0115] The compensation control quantity aims to offset the negative effects of thermal inertia, making the actual response of the system closer to the ideal control expectation of the ESO-A.

[0116] S340: Specifically, ESO-C, as an environmental observer, focuses on estimating and compensating for external disturbances introduced by environmental temperature fluctuations, cooling medium pressure changes, etc. Its calculation process is similar to ESO-B, but the model parameters and observation objects are different.

[0117] In one possible implementation, ESO-C runs through a theoretical formula for iteration:

[0118] ;

[0119] ;

[0120] ;

[0121] ;

[0122] ESO-C generates a second compensation control quantity , the second compensation control quantity makes the control system have stronger anti-external interference ability.

[0123] S350: Specifically, the control signal acting on the controlled object is the synthesis of the main control action of ESO-A and the two compensation control quantities, that is, .

[0124] This synthesis action not only contains the main control intention based on game decision, but also integrates real-time compensation for system internal dynamics and external environment, significantly improving the accuracy of control.

[0125] In a thermal management control method based on adaptive fuzzy logic, S400 continuously learns and optimizes the control system through closed-loop evaluation of execution effects and policy self-tuning, achieving long-term dynamic balance.

[0126] Please refer to Figure 5 , which shows a flowchart of an example of a thermal management control method S400 based on adaptive fuzzy logic, the content of which includes:

[0127] S410: Specifically, after a period of control action execution, the controller starts the comprehensive evaluation process.

[0128] In a possible implementation, the evaluation index at least includes: stability index, responsiveness index, energy efficiency index.

[0129] The controller compares the measured index with the expected target function value recorded at the end of the four-party game in S200. If the comparison is lower than the preset satisfaction degree, it is determined to be unsatisfactory.

[0130] S420: Specifically, when the evaluation result is unsatisfactory, the fuzzy self-tuner is triggered. The fuzzy self-tuner first analyzes which link in the four-party game leads to substandard performance, and its analysis basis is the relationship between the parameters in the game model and the evaluation index.

[0131] In a possible implementation, the fuzzy self-tuner embeds diagnostic rules, and by analyzing the index deviation mode and the game process summary, the fuzzy self-tuner uses fuzzy reasoning to output a qualitative judgment of the adjustment direction of the weight of each party game.

[0132] S430: Specifically, based on the result of root cause analysis, the fuzzy self-tuner outputs the adjustment amount of the four-party game compromise strategy, which is specifically embodied as a quantitative adjustment of the weight of the game.

[0133] In a possible implementation, the fuzzy self-tuner outputs an adjustment factor according to the qualitative judgment. The new game weight is updated as: The adjusted game weight will be updated to the four-party game architecture of S200 for decision-making in the next control period.

[0134] Example two:

[0135] The application is deployed and verified in a liquid cooling heat dissipation system of a high-performance server. The server is configured with dual CPU and four GPU, and the maximum thermal design power is 1200W. The control target is to maintain the CPU core temperature below 85℃, while reducing the energy consumption of the cooling system as much as possible.

[0136] The system configuration is as follows:

[0137] CPU / GPU power, chip temperature, cooling liquid flow and inlet temperature, micro-pump and fan health status are collected. The controller platform uses an embedded industrial PC running a Linux-based real-time system. The ESO-A control speed of the micro-pump and the opening degree of the regulating valve in the observer; the compensation outputs of ESO-B and ESO-C are superimposed on the control signal of ESO-A.

[0138] The following typical disturbance scenarios are simulated:

[0139] Scenario A: Compute load surge. At t=0s, a GPU full load computation is initiated, and the power jumps from 200W to 800W in 0.5s.

[0140] Scenario B: Ambient temperature step. At t=0s, the coolant inlet temperature is simulated to rise by 5℃ in 2s.

[0141] The comparative test results are shown in the following table:

[0142]

[0143] Specific process analysis:

[0144] In scenario A, when the load power jump rate exceeds the threshold, the controller is triggered. In S120, the adaptive fuzzy preprocessor dynamically adjusts the membership function parameters of the load power jump rate according to the high load working condition baseline, so that the membership is higher. In S130, the fuzzy causal reasoning machine applies the confidence calculation formula, successfully identifies the disturbance source as a compute load surge, and the confidence is 0.94. In S140, since the system health is good, the disturbance is given a high priority weight of 0.45. In the four-party game of S230, the response accelerator's game weight is therefore set to 0.5. The parameter package generated after the game solution tends to the fast response mode. ESO-A accordingly outputs the main control action of increasing the pump speed and valve opening . At the same time, ESO-B accurately estimates the total disturbance due to the large thermal capacity inside the server by running its observer equation, and injects a positive compensation control , so that the micropump generates a pre-acceleration action, effectively hedging the thermal inertia, which is the key to significantly reducing the overshoot. ESO-C does not detect strong environmental disturbance, and the compensation is almost zero.

[0145] In scenario B, the disturbance is identified as environmental warming. In the game of S230, since the disturbance type is different, the energy efficiency arbitrator and the stability guardian obtain relatively higher game weights. The game result outputs a parameter package that is more biased towards energy efficiency priority and stability guarantee. ESO-A takes a relatively moderate adjustment, while ESO-C estimates the additional thermal load brought by the environmental warming in time through its observer and injects a compensation signal to assist the system in smooth transition, avoiding overshoot and maintaining the temperature at a lower pump power in steady state, achieving energy efficiency optimization.

[0146] Embodiment three:

[0147] According to embodiment one of the present application, a thermal management control system based on adaptive fuzzy logic is provided, the system comprising:

[0148] a signal acquisition module, configured to monitor multiple source signals in real time, and trigger a control cycle when a value of any signal exceeds a preset sensitive threshold;

[0149] a disturbance causal graph generation module, including an adaptive fuzzy preprocessor, a fuzzy causal inference machine, and a fuzzy priority arbiter; the adaptive fuzzy preprocessor is configured to process the signal exceeding the threshold and dynamically delimit an influence area; the fuzzy causal inference machine is configured to determine a type of the disturbance source; the fuzzy priority arbiter is configured to assign a dynamic priority weight to each type of the disturbance source; and the disturbance causal graph generation module is configured to integrate the type of the disturbance, the influence area, and the priority weight, and generate a disturbance causal graph;

[0150] a four-party game decision module, connected to the disturbance causal graph generation module, configured to load the disturbance causal graph, and including a stability guardian unit, a response accelerator unit, an energy efficiency arbiter unit, and a fuzzy semantic coordinator unit; the stability guardian unit, the response accelerator unit, and the energy efficiency arbiter unit are configured to generate independent control parameter proposals based on the disturbance causal graph, from the aspects of stability, response speed, and energy efficiency target, respectively; and the fuzzy semantic coordinator unit is configured to arbitrate and correct the proposals, and guide the four-party units to generate a semantic control parameter package through a compromise process;

[0151] a multi-observer cooperative execution module, connected to the four-party game decision module, including an ESO-A observer, an ESO-B observer, and an ESO-C observer; the ESO-A observer is configured to load and execute a main control law in the semantic control parameter package; the ESO-B observer is configured to estimate system thermal inertia and generate a first compensation control amount; the ESO-C observer is configured to estimate environmental disturbance and generate a second compensation control amount; and the multi-observer cooperative execution module is configured to synthesize the main control action and the compensation control amounts, and output a final control signal to a controlled object;

[0152] a closed-loop evaluation and self-tuning module, connected to the multi-observer cooperative execution module and the four-party game decision module, respectively, configured to comprehensively evaluate an execution effect of the final control signal, and when the evaluation result does not meet a preset requirement, analyze a performance deviation reason through a fuzzy self-tuner, and adaptively adjust compromise strategy parameters in the four-party game decision module, so as to realize dynamic balance and continuous optimization of the control system.

[0153] Embodiment Four:

[0154] According to an embodiment of the present application, a computer device is provided, the computer device comprising: a processor and a memory, the memory storing at least one instruction, at least one program, a code set or an instruction set, the at least one instruction, the at least one program, the code set or the instruction set being loaded and executed by the processor to implement the adaptive fuzzy logic based thermal management control method according to the above aspect.

[0155] Those skilled in the art will understand that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk memory, CD-ROM, optical memory, etc.) containing computer-usable program code. The solutions in the embodiments of the present application can be implemented in various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0156] The present application is described with reference to flowcharts and / or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, as well as combinations of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing apparatus to produce a machine, so that the instructions, which are executed via the processor of the computer or other programmable data processing apparatus, generate a means for implementing the functions specified in the flowcharts and / or block diagrams. Figure 1 The flow or flows and / or blocks in the flowcharts and / or block diagrams Figure 1 The means for carrying out the functions specified in the flow or flows and / or blocks in the flowcharts and / or block diagrams.

[0157] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to work in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including instruction means, which implement the functions specified in the flowcharts and / or block diagrams. Figure 1 The flow or flows and / or blocks in the flowcharts and / or block diagrams Figure 1 The means for carrying out the functions specified in the flow or flows and / or blocks in the flowcharts and / or block diagrams.

[0158] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus, so that a series of operation steps are performed on the computer or other programmable data processing apparatus to produce a computer-implemented process, so that the instructions executed on the computer or other programmable data processing apparatus provide a means for implementing the functions specified in the flowcharts and / or block diagrams. Figure 1 The flow or flows and / or blocks in the flowcharts and / or block diagrams Figure 1steps of the functions specified in the block or blocks.

[0159] While the preferred embodiments of the application have been described, additional variations and modifications can be made to the preferred embodiments by those of skill in the art once they have the benefit of the present disclosure. Therefore, the appended claims are intended to encompass within their scope all possible variations and modifications of the preferred embodiments.

[0160] It is apparent that those skilled in the art can, without departing from the spirit and scope of the application, make various changes and modifications of the application. Thus, it is intended that the present application cover all modifications and variations of this application provided they come within the scope of the appended claims and their equivalents.

Claims

1. A thermal management control method based on adaptive fuzzy logic, characterized in that, include: The controller monitors multiple signals in real time and triggers the start of a control cycle when any signal exceeds a preset sensitivity threshold. The adaptive fuzzy preprocessor in the controller defines the affected area; The fuzzy causal inference engine in the controller determines the type of disturbance source; the fuzzy priority arbitrator in the controller assigns priority weights to various disturbance source types; and the disturbance type, affected area, and priority weights are integrated to generate a disturbance causal graph. After the controller loads the disturbance causal graph, it initiates a four-way game architecture consisting of a stability guardian, a response accelerator, an energy efficiency arbitrator, and a fuzzy semantic coordinator. The game process follows four steps: parameter proposal, arbitration correction, four-party compromise, and proposal encapsulation, and outputs a semantic control parameter package. Configure three-layer observers, ESO-A, ESO-B, and ESO-C, in the controller, so that the three-layer observers focus on parameter proposal execution, thermal inertia response, and environmental disturbance response, respectively; after loading the semantic control parameter package in the controller, start the three-layer observers to execute the semantic control parameter package; The controller performs a comprehensive evaluation of the observer's performance; if it is not satisfied with the result of the comprehensive evaluation, it adaptively adjusts the compromise strategy of the four-way compromise through a fuzzy self-tuner to achieve dynamic balance of the control system. Among them, the objective functions of the stability guardian, the response accelerator, and the energy efficiency arbitrator are respectively , , ,in, For control parameter vectors; The Stability Guardian proposes control parameters aimed at maintaining system stability based on disturbance causal graph data. The Response Accelerator proposes control parameters based on disturbance causality graph data, aiming to improve the dynamic response speed of the system. Energy efficiency arbitrators propose control parameters to optimize system energy efficiency targets based on disturbance causality graph data. The fuzzy semantic coordinator arbitrates and modifies the parameter proposals of the three parties based on security constraints; The arbitration and correction process of the fuzzy semantic coordinator is modeled as a constrained optimization problem to find a compromise solution: ; ;in, and The first Ideal and worst values ​​for each target It is the game weight. It is a security constraint introduced by the fuzzy semantic coordinator. yes , , The specific calculation formula.

2. The thermal management control method based on adaptive fuzzy logic according to claim 1, characterized in that, The specific methods by which the adaptive fuzz preprocessor delineates the affected region include: The membership function parameters of each input signal are dynamically adjusted based on historical operating data; for the first... Input signal Its corresponding fuzzy subset Membership function center and width Update as follows: ; ; in, For fuzzy subset indexing, For time series indexing, and Let these represent the expected value and variance, respectively. It is a forgetting factor.

3. The thermal management control method based on adaptive fuzzy logic according to claim 1 or 2, characterized in that, The fuzzy causal inference engine determines the type of disturbance source and outputs the first... Confidence score of disturbance source The calculation formula is as follows: ;in, It is the output variable of the inference engine. It is the first The first type of disturbance source Fuzzy subsets after the inference rules It is the first The weight of each inference rule, It is a fuzzy subset Adaptive membership function.

4. The thermal management control method based on adaptive fuzzy logic according to claim 3, characterized in that, The fuzzy priority arbitrator is the first... Priority weights assigned to perturbation sources The calculation formula is as follows: ;in, It is a specific disturbance source weight adjustment function related to system health. It refers to the system's health status.

5. The thermal management control method based on adaptive fuzzy logic according to claim 1, characterized in that, The stability guardian aims to maintain system temperature. To achieve stability, minimize overshoot and oscillation, the objective function is defined as follows: ,in , Proposal for weighted, stability-guardian control parameters By minimizing get; The objective of the response accelerator is to improve the dynamic response speed of the system to disturbances, and the objective function is defined as follows: , To adjust the time, in response to the accelerator control parameter proposal By minimizing get; The goal of the energy efficiency arbitrator is to optimize the overall energy efficiency of the system, and the objective function is defined as follows: , Proposed control parameters for total cooling energy consumption by the energy efficiency arbitrator. By minimizing get.

6. The thermal management control method based on adaptive fuzzy logic according to claim 1, characterized in that, The ESO-B translates parameter proposals in the semantic control parameter package into specific control actions, including: Abstract parameters in the semantic control parameter package Perform standard PID control law calculation: ;in, At any moment Main control actions obtained from ESO-A For temperature deviation, , , These are proportional gain, integral gain, and derivative gain, respectively. For integration variables; ESO-A outputs preliminary calculation results. .

7. The thermal management control method based on adaptive fuzzy logic according to claim 1, characterized in that, The ESO-B is used to estimate and compensate for thermal inertial disturbances, and the first compensation control quantity it generates is... Iterative calculations are performed using the following formula: ； ； ； ； in, It defines the observation error, which is the temperature tracked by the observer. Compared with actual temperature difference; Used to track the system's actual output. Its update depends on observation error Main control action and the estimated total disturbance , , For observer gain, This is the compensation coefficient for ESO. It is a nonlinear function. for The power of a function is the width of the linear interval of the function; To compensate for control quantity .

8. The thermal management control method based on adaptive fuzzy logic according to claim 7, characterized in that, The ESO-C is used to estimate and compensate for environmental disturbances, and the second compensation control quantity it generates is... The first compensation control quantity generated with ESO-B The iterative calculation principle is the same.

9. The thermal management control method based on adaptive fuzzy logic according to claim 1, characterized in that, The comprehensive evaluation includes at least a quantitative assessment of the system's output stability index, dynamic response speed index, and overall energy efficiency index; the fuzzy self-tuner outputs the game weights for the next control cycle based on the deviation pattern between the evaluation results and the game process summary. Adjustment factor , and according to Update.

10. A thermal management control system based on adaptive fuzzy logic, characterized in that, The thermal management control method based on adaptive fuzzy logic according to any one of claims 1-9, the system includes: The signal acquisition module is used to monitor multi-source signals in real time and trigger a control cycle when the value of any signal exceeds a preset sensitivity threshold. The perturbation causal graph generation module includes an adaptive fuzzy preprocessor, a fuzzy causal inference engine, and a fuzzy priority arbitrator. The adaptive fuzzy preprocessor is used to process signals exceeding the threshold and dynamically delineate the affected area. The fuzzy causal inference engine is used to determine the type of perturbation source. The fuzzy priority arbitrator is used to assign dynamic priority weights to various types of perturbation sources. The perturbation causal graph generation module is used to integrate the perturbation type, affected area, and priority weights to generate a perturbation causal graph. The four-party game decision-making module, connected to the perturbation causal graph generation module, is used to load the perturbation causal graph and includes a stability guardian unit, a response accelerator unit, an energy efficiency arbitrator unit, and a fuzzy semantic coordinator unit. The stability guardian unit, response accelerator unit, and energy efficiency arbitrator unit are used to generate independent control parameter proposals based on the perturbation causal graph, starting from stability, response speed, and energy efficiency objectives, respectively. The fuzzy semantic coordinator unit is used to arbitrate and modify the proposals and guide the four-party units to generate semantic control parameter packages through a compromise process. The multi-observer collaborative execution module, connected to the four-party game decision module, includes ESO-A observer, ESO-B observer, and ESO-C observer. The ESO-A observer is used to load and execute the main control law in the semantic control parameter package. The ESO-B observer is used to estimate the system thermal inertia and generate the first compensation control quantity. The ESO-C observer is used to estimate the environmental disturbance and generate the second compensation control quantity. The multi-observer collaborative execution module is used to synthesize the main control action and the compensation control quantity, and output the final control signal to the controlled object. The closed-loop evaluation and self-tuning module is connected to the multi-observer collaborative execution module and the four-party game decision module, respectively. It is used to comprehensively evaluate the execution effect of the final control signal. When the evaluation result does not meet the preset requirements, the fuzzy self-tuner analyzes the cause of the performance deviation and adaptively adjusts the compromise strategy parameters in the four-party game decision module to achieve dynamic balance and continuous optimization of the control system.

11. A computer device, characterized in that, The computer device includes a processor and a memory, wherein the memory stores at least one instruction, at least one program, a code set, or an instruction set, and the at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by the processor to implement the thermal management control method based on adaptive fuzzy logic as described in any one of claims 1 to 9.

Images