Intelligent water cooling control method, system, medium and device

Abstract

The application discloses an intelligent water cooling control method, system, medium and equipment, and the method comprises the following steps: establishing a three-dimensional dynamic model of correlation power consumption, temperature and computing power; predicting a chip junction temperature track according to the three-dimensional dynamic model by using an extended Kalman filter; establishing a prediction model of a water cooling system, and controlling and optimizing the water cooling system according to the chip junction temperature track and the prediction model by using an MPC method; and detecting a temperature sensor, a pressure sensor and a rotating speed sensor in real time, and adjusting the control optimization result according to the detection result. Based on the above data processing procedure, the application provides a 'prediction-optimization-correction' closed-loop water cooling control method for a smart driving domain controller, and high computing power heat dissipation, low energy consumption and functional safety can be considered.

Description

Technical Field

[0001] This invention relates to the field of intelligent driving system technology, and in particular to an intelligent water-cooling control method, system, medium and device. Background Technology

[0002] Existing water cooling solutions generally suffer from drawbacks such as "passive response, fixed MAP control mapping table, and lack of functional safety." 1. The use of an independently cooled electronic control unit (ECU) that interacts with the intelligent driving domain controller via CAN cannot suppress millisecond-level thermal shocks; 2. The coolant pump speed and valve position are set using a calibrated MAP and are not coupled with real-time computing power and power consumption. The energy consumption is 280 W in summer urban congestion conditions and excessive cooling in winter high-altitude cold conditions, with ΔT>20℃. 3. Without ASIL functional safety design, if a single point of failure occurs in the water pump or valve body, the system has no degradation path, triggering the intelligent driving domain controller to power down, causing the vehicle's functions to suddenly stop.

[0003] Therefore, there is an urgent need for a closed-loop water-cooling control method for intelligent driving domain controllers that combines high computing power heat dissipation, low energy consumption, and functional safety. Summary of the Invention

[0004] This invention provides an intelligent water-cooling control method, system, medium, and device that can balance high computing power heat dissipation, low energy consumption, and functional safety.

[0005] Firstly, an intelligent water-cooling control method is provided, including: Establish a three-dimensional dynamic model relating power consumption, temperature, and computing power; The chip junction temperature trajectory is predicted based on the three-dimensional dynamic model using an extended Kalman filter. A predictive model for the water cooling system is established, and the MPC method is used to control and optimize the water cooling system based on the chip junction temperature trajectory and the predictive model. The system monitors temperature, pressure, and speed sensors in real time and adjusts the control optimization based on the monitoring results.

[0006] In some embodiments, the method for establishing the three-dimensional dynamic model is shown in the following formula: ; Among them, the chip's real-time power consumption P diss As shown in the following formula: P diss = α·NPU(t)·V 2 ; The heat Q removed by the water cooling system cool As shown in the following formula: ; In the formula, NPU(t) is the computing power utilization rate; α is a coefficient; V is the real-time voltage; T j T represents the chip junction temperature. c R represents the temperature of the cold plate. jc For junction-cold plate thermal resistance; R ca For cold plate - ambient thermal resistance; c is the coolant mass flow rate; p Specific heat capacity; ΔT(t) is the temperature difference; C j C is the equivalent heat capacity of the chip junction region; c Equivalent heat capacity for cold plate packaging; T a dT represents the coolant inlet temperature. j / dt is the rate of change of the chip junction temperature over time; dT c / dt represents the rate of change of the cold plate temperature over time.

[0007] In some embodiments, predicting the chip junction temperature trajectory based on the three-dimensional dynamic model using extended Kalman filtering includes: The three-dimensional dynamic model is linearized using Taylor series expansion to generate discrete state equations; The extended Kalman filter is used to predict the current state vector based on the current observation data. Based on the discrete state equation, the chip junction temperature trajectory is predicted according to the current state vector.

[0008] In some embodiments, the discrete state equation is as follows: ; In the formula, T j,k T is the chip junction temperature at time k; j,k-1 R is the chip junction temperature at time k-1; jc For junction-cold plate thermal resistance; T c,k T is the cold plate junction temperature at time k; c,k-1 Let k be the cold plate junction temperature at time k-1; Let k be the coolant mass flow rate at time k. C is the coolant mass flow rate at time k-1; j C is the equivalent heat capacity of the chip junction region; c The equivalent heat capacity of the cold plate package; dt is the fixed period; P diss,k-1 Q represents the chip power consumption at time k-1; cool,k-1 This refers to the heat removed by the k-1 water cooling system at any given time.

[0009] In some embodiments, the predictive model for the water-cooling system is shown in the following equation: ; Wherein, the predicted state vector at time k for: ; In the formula, This is the predicted value for the chip junction temperature; This is the predicted temperature of the cold plate; The predicted mass flow rate of the coolant is given; T is the transpose. z is the predicted state vector at time k+1; k+1 Let A be the observation vector at time k+1; k B is the state transition matrix; k To control the input matrix; u k For control input vectors, including control signals for water pumps, proportional valves, and fans; H k This is the observation matrix.

[0010] In some embodiments, the objective function of the MPC method is shown in the following equation: ; Wherein, the predicted state vector at time k for: ; Output vector at time k For: y k = [n pump , u valve , d fan ] T ; The reference target vector r at time k k r k = [T ref , 0,0] T ; In the formula, T j T represents the chip junction temperature. c This refers to the temperature of the cold plate. This is the predicted value of the coolant mass flow rate; This is the predicted value for the chip junction temperature; This is the predicted temperature of the cold plate; ε is the predicted mass flow rate of the coolant; T is the transpose; ε is the slack variable; Q, R, and S are the weights; n pump The pump speed; u valve d represents the proportional valve opening. fan The fan duty cycle is T; N is the prediction time domain length; T is the fan duty cycle. ref This represents the expected temperature value. The predicted state vector at time k+1; u k The control input vector includes control signals for the water pump, proportional valve, and fan; Δu k To control the amount of change in the input vector; u max u minTo control the upper and lower limits of the input vector; x max x min A represents the upper and lower bounds of the predicted state vector. k B is the state transition matrix; k To control the input matrix.

[0011] In some embodiments, the real-time detection of temperature sensors, pressure sensors, and speed sensors, and the adjustment of control optimization results based on the detection results, includes: Fault diagnosis of temperature sensor, pressure sensor and speed sensor; If any two of the temperature sensor, pressure sensor, and speed sensor are detected to be faulty, the water pump speed of the water cooling system will be controlled to the preset speed, and the proportional valve of the water cooling system will be controlled to switch to the small circulation. The fault determination method for any sensor is as follows: when the difference between the current feedback value of any sensor and the corresponding model residual is outside the preset range, it is determined that there is a fault.

[0012] Secondly, an intelligent water-cooling control system is provided, including: The model building module is used to create a three-dimensional dynamic model that correlates power consumption, temperature, and computing power. The filtering module is communicatively connected to the model building module and is used to predict the chip junction temperature trajectory based on the three-dimensional dynamic model using extended Kalman filtering. The target module, which is communicatively connected to the filtering module, is used to control and optimize the water cooling system based on the chip junction temperature trajectory using the MPC objective function; The adjustment module is communicatively connected to the target module and is used to detect the temperature sensor, pressure sensor and speed sensor in real time, and adjust the control optimization results based on the detection results.

[0013] Thirdly, embodiments of the present invention provide a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the intelligent water-cooling control method as described above.

[0014] Fourthly, embodiments of the present invention provide an electronic device, including a storage medium, a processor, and a computer program stored in the storage medium and executable on the processor, wherein the processor executes the computer program to implement the intelligent water-cooling control method as described above.

[0015] Compared with existing technologies, the advantages of this invention are as follows: It establishes a three-dimensional dynamic model relating power consumption, temperature, and computing power; it uses extended Kalman filtering to predict the chip junction temperature trajectory based on the three-dimensional dynamic model; it establishes a predictive model for the water cooling system and uses the MPC method to optimize the control of the water cooling system based on the chip junction temperature trajectory and the predictive model; it monitors temperature, pressure, and speed sensors in real time and adjusts the control optimization results based on the monitoring results. Therefore, this invention systematically solves the problems of nonlinearity, time delay, strong coupling, and multiple constraints in the water cooling control of high heat flux chips through a closed-loop architecture of "prediction-optimization-correction," significantly reducing cooling energy consumption and improving system robustness and maintainability while ensuring junction temperature safety and computing power stability. Attached Figure Description

[0016] Figure 1 This is a schematic flowchart of an embodiment of the intelligent water cooling control method of the present invention; Figure 2 This is a schematic diagram of the three-dimensional dynamic model of the correlation between power consumption, temperature and computing power of the present invention; Figure 3 This is a schematic diagram of the EKF-MPC closed-loop control process of the present invention; Figure 4 This is a schematic diagram illustrating how the control optimization results are adjusted based on the detection results according to the present invention. Detailed Implementation

[0017] Referring now to specific embodiments of the invention, examples of which are illustrated in the accompanying drawings. Although the invention will be described in conjunction with specific embodiments, it will be understood that it is not intended to limit the invention to the described embodiments. Rather, it is intended to cover variations, modifications, and equivalents included within the spirit and scope of the invention as defined by the appended claims. It should be noted that the method steps described herein can be implemented by any functional block or functional arrangement, and any functional block or functional arrangement can be implemented as a physical entity or a logical entity, or a combination of both.

[0018] To enable those skilled in the art to better understand the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0019] Note: The examples described below are merely specific examples and are not intended to limit the embodiments of the present invention to the specific steps, values, conditions, data, order, etc. Those skilled in the art can utilize the concept of the present invention to construct more embodiments not mentioned herein by reading this specification.

[0020] Please see Figure 1 The present invention provides an intelligent water cooling control method, the method comprising: Step S100: Establish a three-dimensional dynamic model relating power consumption, temperature, and computing power, and refer to... Figure 2 As shown, the method for establishing the three-dimensional dynamic model is as follows: ; Among them, the chip's real-time power consumption P diss As shown in the following formula: P diss = α·NPU(t)·V 2 ; The heat Q removed by the water cooling system cool As shown in the following formula:

[0021] In the formula, NPU(t) is the computing power utilization rate; α is a coefficient; V is the real-time voltage; T j T represents the chip junction temperature. c R represents the temperature of the cold plate. jc For junction-cold plate thermal resistance; R ca For cold plate - ambient thermal resistance; c is the coolant mass flow rate; p Specific heat capacity; ΔT(t) is the temperature difference; C j C is the equivalent heat capacity of the chip junction region; c Equivalent heat capacity for cold plate packaging; T a dT represents the coolant inlet temperature. j / dt is the rate of change of the chip junction temperature over time; dT c / dt represents the rate of change of the cold plate temperature over time.

[0022] Therefore, the "three-dimensional dynamic model of power consumption-temperature-computing power" of this invention is not a simple list of three independent variables, but rather encapsulates the causal chain of "instantaneous computing power demand" → "power consumption" → "heat generation" → "temperature" into a state space model that can be updated in real time.

[0023] See also Figure 3 As shown, step S200, predicting the chip junction temperature trajectory based on the three-dimensional dynamic model using an extended Kalman filter (EKF), includes: The three-dimensional dynamic model is linearized using Taylor series expansion to generate discrete state equations; The extended Kalman filter is used to predict the current state vector based on the current observation data. Based on the discrete state equation, the chip junction temperature trajectory is predicted according to the current state vector.

[0024] Specifically, in this embodiment of the invention, the discrete state equation is as follows: ; In the formula, T j,k T is the chip junction temperature at time k; j,k-1 R is the chip junction temperature at time k-1; jc For junction-cold plate thermal resistance; T c,k T is the cold plate junction temperature at time k; c,k-1 Let k be the cold plate junction temperature at time k-1; Let k be the coolant mass flow rate at time k. C is the coolant mass flow rate at time k-1; j C is the equivalent heat capacity of the chip junction region; c The equivalent heat capacity of the cold plate package; dt is the fixed period; P diss,k-1 Q represents the chip power consumption at time k-1; cool,k-1 This refers to the heat removed by the k-1 water cooling system at any given time.

[0025] Definition: Predicted state vector x = [T] j ^, T c ^, ^] The parameters in the predicted state vector x correspond to the junction temperature, cold plate temperature, and mass flow rate; the input u = [P diss Q cool ] The parameters in the input u correspond to the heating power and the power carried away by water cooling; the observation vector y=[T c , ΔP] The parameters in the observation vector y correspond to the cold plate temperature sensor and the pump pressure difference; the process noise covariance Q=diag(0.1, 0.05, 0.01) -- model error; the observation noise covariance R=diag(0.2, 0.1) -- sensor noise.

[0026] The transition state matrix F can be derived from the offline state equations as follows: ; In the formula, T in This refers to the coolant inlet temperature. The observation equation is as follows: T csens = T Ck / / Directly measure the temperature of the cold plate ΔP = K p · k ² / / Pump pressure difference ∝ Flow rate squared In the formula, T csens The sensor reading for the cold plate temperature; T Ck K represents the temperature of the cold plate at time k; ΔP represents the pressure difference of the water pump; K p This is the differential pressure coefficient of the water pump; From the above, the observation matrix H can be derived as follows: [010; 002·K p · ].

[0027] The prediction time domain is 2 s⁻²⁰ steps, meaning the extended Kalman filter is only responsible for providing the optimal estimate of k at the current time. k|k To predict the next 20 steps, directly use the control sequence u k … u k+19 By inputting the same discrete state equation, we can obtain: T j,k+1|k ,T j,k+2|k … T j,k+20|k This is the chip junction temperature profile required by MPC.

[0028] Step S300: Establish a predictive model for the water cooling system. Utilize the MPC (Model Predictive Control) method to optimize the water cooling system based on the chip junction temperature trajectory and the predictive model. Specifically, employ model predictive control (MPC) to continuously optimize the water pump speed n. pump Proportional valve opening u valve and fan duty cycle d fan Ultimately, this achieves optimized water-cooling control of the domain controller.

[0029] Specifically, the prediction model for the water cooling system is shown in the following equation: ; Wherein, the predicted state vector at time k for: ; In the formula, This is the predicted value for the chip junction temperature; This is the predicted temperature of the cold plate; The predicted mass flow rate of the coolant is given; T is the transpose. z is the predicted state vector at time k+1; k+1 Let k+1 be the observation vector. , representing the output value that can be directly obtained through sensors or other means at the next moment, n pump The pump speed; u valve d represents the proportional valve opening. fan A represents the fan duty cycle. k The state transition matrix describes how the system transitions from the current state. Evolving into the next state B k The control input matrix describes the control input vector u. k State The direct impact; u k For control input vectors, including control signals for water pumps, proportional valves, and fans; H k The observation matrix describes the state vector. Mapped to observable output z k+1 .

[0030] The objective function of the MPC method is shown in the following equation: ; Wherein, the predicted state vector at time k for: ; Output vector at time k For: y k = [n pump , u valve , d fan ] T ; The reference target vector r at time k k r k = [T ref , 0,0] T ; In the formula, T j T represents the chip junction temperature. c This refers to the temperature of the cold plate. This is the predicted value of the coolant mass flow rate; This is the predicted value for the chip junction temperature; This is the predicted temperature of the cold plate; ε is the predicted mass flow rate of the coolant; T is the transpose; ε is the slack variable; Q, R, and S are the weights; n pump The pump speed; u valve d represents the proportional valve opening. fan The fan duty cycle is T; N is the prediction time domain length; T is the fan duty cycle. ref This represents the expected temperature value. The predicted state vector at time k+1; u k The control input vector includes control signals for the water pump, proportional valve, and fan; Δu k To control the amount of change in the input vector; u max u min To control the upper and lower limits of the input vector; x max x min A represents the upper and lower bounds of the predicted state vector. k B is the state transition matrix; k To control the input matrix.

[0031] Therefore, the final goal is to solve 20 steps every 100 ms: to ensure that the chip does not overheat in the next 2 seconds, to minimize the power consumption and noise of the pump / fan, and to always meet the physical limits.

[0032] See also Figure 4 As shown, step S400 involves real-time monitoring of the temperature sensor, pressure sensor, and speed sensor, and adjusting the control optimization results based on the monitoring results, including: Fault diagnosis of temperature sensor, pressure sensor and speed sensor; If any two of the temperature sensor, pressure sensor, and speed sensor are detected to be faulty, the water pump speed of the water cooling system will be controlled to the preset speed, and the proportional valve of the water cooling system will be controlled to switch to the small circulation. The fault determination method for any sensor is as follows: when the difference between the current feedback value of any sensor and the corresponding model residual is outside the preset range, it is determined that there is a fault.

[0033] Specifically, in this embodiment of the invention, the dual-core lockstep comparison begins by voting on the temperature sensor, pressure sensor, and speed sensor in a 2-out-of-3 vote; if two sensors report a fault, a fault is considered to have occurred. Actuator fault diagnosis is based on the current feedback value (referring to the operating current in the sensor's own circuit) and the model residual (the difference between the actual sensor measurement value and the model prediction value; the model prediction can be referenced in the prediction model above), with a threshold of ±10%. When a single point of failure occurs, a degradation mode is triggered: the water pump defaults to 40% speed, the valve body switches to small circulation, and the computing power is limited to 70%.

[0034] The following specific embodiment is provided; in this embodiment, the vehicle drives continuously for 30 minutes under urban congestion conditions, with an ambient temperature of 42 ℃, SoC computing power utilization of 80%, and NPU utilization of 70%. The following results are obtained using the method of this invention: EKF predicts that the junction temperature will reach 93 ℃ in 2 seconds; MPC optimized output: pump speed 4800 rpm, valve opening 65%, fan duty cycle 50%; The actual junction temperature remained stable at 86 ℃, with energy consumption of 185 W, a 34% reduction compared to traditional MAP solutions; The interior noise level is 52 dB(A), which meets the NVH requirements for luxury vehicles. If the water pump in the functional safety monitoring layer experiences a circuit breaker failure, the system will switch to degraded mode within 12 ms, and the vehicle functions will not be lost.

[0035] This invention also provides an intelligent water-cooling control system, comprising: The model building module is used to create a three-dimensional dynamic model that correlates power consumption, temperature, and computing power. The filtering module is communicatively connected to the model building module and is used to predict the chip junction temperature trajectory based on the three-dimensional dynamic model using extended Kalman filtering. The target module, which is communicatively connected to the filtering module, is used to control and optimize the water cooling system based on the chip junction temperature trajectory using the MPC objective function; The adjustment module is communicatively connected to the target module and is used to detect the temperature sensor, pressure sensor and speed sensor in real time, and adjust the control optimization results based on the detection results.

[0036] In summary, the integrated method proposed in this invention, which combines "establishing a three-dimensional dynamic model of power consumption, temperature, and computing power + EKF junction temperature prediction + water-cooling system prediction model + MPC optimization + real-time sensor calibration," is expected to bring the following beneficial effects: 1. By using EKF to reconstruct the junction temperature trajectory in real time under noise and hysteresis conditions, estimation bias caused by inlet and outlet temperatures can be avoided.

[0037] 2. MPC uses predictive models to anticipate load and environmental disturbances, and adjusts pump speed / valve position / fan in advance to achieve small overshoot, fast convergence and stable junction temperature control, reducing peak junction temperature and temperature fluctuation amplitude.

[0038] 3. The 3D model incorporates temperature and computing power into the optimization objectives and constraints, proactively preventing thermal throttling and computing power fluctuations, and improving the sustainable full load time and service quality.

[0039] 4. Significantly reduce system energy consumption and operating costs. Under the premise of meeting temperature and pressure safety constraints, MPC minimizes the energy consumption of pumps / fans / secondary refrigeration and avoids the conservative strategy of "overcooling / overflow".

[0040] Therefore, this invention systematically solves the problems of nonlinearity, time delay, strong coupling and multiple constraints in the water cooling control of high heat flux chips through a closed-loop architecture of "prediction-optimization-correction". While ensuring junction temperature safety and computing power stability, it significantly reduces cooling energy consumption and improves system robustness and maintainability, and has clear engineering application value and promotion significance.

[0041] Specifically, this embodiment corresponds one-to-one with the above method embodiments. The functions of each module have been described in detail in the corresponding method embodiments, so they will not be repeated here.

[0042] Based on the same inventive concept, embodiments of this application also provide a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements all or part of the method steps of the above method.

[0043] The present invention can implement all or part of the processes in the above methods, or it can be accomplished by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when the computer program is executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content contained in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium does not include electrical carrier signals and telecommunication signals.

[0044] Based on the same inventive concept, embodiments of this application also provide an electronic device, including a memory and a processor. The memory stores a computer program that runs on the processor. When the processor executes the computer program, it implements all or part of the method steps described above.

[0045] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the computer device, connecting all parts of the computer device through various interfaces and lines.

[0046] Memory can be used to store computer programs and / or modules. The processor performs various functions of the computer device by running or executing the computer programs and / or modules stored in the memory, and by accessing data stored in the memory. Memory can primarily include a program storage area and a data storage area. The program storage area can store the operating system and at least one application program required for a function (e.g., sound playback, image playback, etc.); the data storage area can store data created based on the use of the mobile phone (e.g., audio data, video data, etc.). Furthermore, memory can include high-speed random access memory, and can also include non-volatile memory, such as hard disks, RAM, plug-in hard disks, SmartMedia Cards (SMC), Secure Digital (SD) cards, Flash Cards, at least one disk storage device, flash memory device, or other volatile solid-state storage devices.

[0047] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, servers, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage and optical storage) containing computer-usable program code.

[0048] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), servers, and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations 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, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0049] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0050] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0051] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. An intelligent water-cooling control method, characterized in that, include: Establish a three-dimensional dynamic model relating power consumption, temperature, and computing power; The chip junction temperature trajectory is predicted based on the three-dimensional dynamic model using an extended Kalman filter. A predictive model for the water cooling system is established, and the MPC method is used to control and optimize the water cooling system based on the chip junction temperature trajectory and the predictive model. The system monitors temperature, pressure, and speed sensors in real time and adjusts the control optimization based on the monitoring results.

2. The intelligent water-cooling control method as described in claim 1, characterized in that, The method for establishing the three-dimensional dynamic model is shown in the following formula: ； Wherein, the real-time power consumption P of the chip diss As shown in the following formula: P diss = α·NPU(t)·V 2 ; The heat Q removed by the water cooling system cool As shown in the following formula: ； In the formula, NPU(t) is the computing power utilization rate; α is a coefficient; V is the real-time voltage; T j T represents the chip junction temperature. c R represents the temperature of the cold plate. jc For junction-cold plate thermal resistance; R ca For cold plate - ambient thermal resistance; c is the coolant mass flow rate; p Specific heat capacity; ΔT(t) is the temperature difference; C j C is the equivalent heat capacity of the chip junction region; c Equivalent heat capacity for cold plate packaging; T a dT represents the coolant inlet temperature. j / dt is the rate of change of the chip junction temperature over time; dT c / dt represents the rate of change of the cold plate temperature over time.

3. The intelligent water-cooling control method as described in claim 1, characterized in that, The method of predicting the chip junction temperature trajectory based on the three-dimensional dynamic model using extended Kalman filtering includes: The three-dimensional dynamic model is linearized using Taylor series expansion to generate discrete state equations; The extended Kalman filter is used to predict the current state vector based on the current observation data. Based on the discrete state equation, the chip junction temperature trajectory is predicted according to the current state vector.

4. The intelligent water-cooling control method as described in claim 3, characterized in that, The discrete state equation is shown below: ； In the formula, T j,k T is the chip junction temperature at time k; j,k-1 R is the chip junction temperature at time k-1; jc For junction-cold plate thermal resistance; T c,k T is the cold plate junction temperature at time k; c,k-1 Let k be the cold plate junction temperature at time k-1; Let k be the coolant mass flow rate at time k. C is the coolant mass flow rate at time k-1; j C is the equivalent heat capacity of the chip junction region; c The equivalent heat capacity of the cold plate package; dt is the fixed period; P diss,k-1 Q represents the chip power consumption at time k-1; cool,k-1 This refers to the heat removed by the k-1 water cooling system at any given time.

5. The intelligent water-cooling control method as described in claim 1, characterized in that, The prediction model for the water cooling system is shown in the following equation: ； Wherein, the predicted state vector at time k for: ; In the formula, This is the predicted value for the chip junction temperature; This is the predicted temperature of the cold plate; The predicted mass flow rate of the coolant is given; T is the transpose. z is the predicted state vector at time k+1; k+1 Let A be the observation vector at time k+1; k B is the state transition matrix; k To control the input matrix; u k For control input vectors, including control signals for water pumps, proportional valves, and fans; H k This is the observation matrix.

6. The intelligent water-cooling control method as described in claim 1, characterized in that, The objective function of the MPC method is shown in the following equation: ； Wherein, the predicted state vector at time k for: ; Output vector at time k For: y k = [n pump , u valve , d fan ] T ; The reference target vector r at time k k r k = [T ref , 0,0] T ; In the formula, T j T represents the chip junction temperature. c This refers to the temperature of the cold plate. This is the predicted value of the coolant mass flow rate; This is the predicted value for the chip junction temperature; This is the predicted temperature of the cold plate; ε is the predicted mass flow rate of the coolant; T is the transpose; ε is the slack variable; Q, R, and S are the weights; n pump The pump speed; u valve d represents the proportional valve opening. fan The fan duty cycle is T; N is the prediction time domain length; T is the fan duty cycle. ref This represents the expected temperature value. The predicted state vector at time k+1; u k The control input vector includes control signals for the water pump, proportional valve, and fan; Δu k To control the amount of change in the input vector; u max u min To control the upper and lower limits of the input vector; x max x min A represents the upper and lower bounds of the predicted state vector. k B is the state transition matrix; k To control the input matrix.

7. The intelligent water-cooling control method as described in claim 1, characterized in that, The real-time detection of temperature, pressure, and speed sensors, and the adjustment of control optimization results based on the detection results, include: Fault diagnosis of temperature sensor, pressure sensor and speed sensor; If any two of the temperature sensor, pressure sensor, and speed sensor are detected to be faulty, the water pump speed of the water cooling system will be controlled to the preset speed, and the proportional valve of the water cooling system will be controlled to switch to the small circulation. The fault determination method for any sensor is as follows: when the difference between the current feedback value of any sensor and the corresponding model residual is outside the preset range, it is determined that there is a fault.

8. An intelligent water-cooling control system, characterized in that, include: The model building module is used to create a three-dimensional dynamic model that correlates power consumption, temperature, and computing power. The filtering module is communicatively connected to the model building module and is used to predict the chip junction temperature trajectory based on the three-dimensional dynamic model using extended Kalman filtering. The target module, which is communicatively connected to the filtering module, is used to control and optimize the water cooling system based on the chip junction temperature trajectory using the MPC objective function; The adjustment module is communicatively connected to the target module and is used to detect the temperature sensor, pressure sensor and speed sensor in real time, and adjust the control optimization results based on the detection results.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the intelligent water-cooling control method as described in any one of claims 1 to 7.

10. An electronic device comprising a storage medium, a processor, and a computer program stored in the storage medium and executable on the processor, characterized in that, When the processor runs the computer program, it implements the intelligent water-cooling control method as described in any one of claims 1 to 7.

Images