A temperature control system for a liquid-cooled thermal management device

Abstract

The application belongs to the technical field of heat management, and discloses a temperature control system of a liquid-cooled heat management device, which comprises a cooling pipeline, a pump, a first flow valve, a sensor, a first switch and a water-cooled plate connected in series on the cooling pipeline, a liquid storage tank, a second flow valve and a second switch connected in series on the cooling pipeline, and a pipeline with the liquid storage tank, the second flow valve and the second switch connected in parallel with the first switch; and a control console for controlling the opening and closing of the first switch and the second switch and the valve opening degree of the first flow valve and the second flow valve according to the data of the sensor, wherein the first switch and the second switch are not opened and closed at the same time, the opening degree of the first flow valve is controlled by using a fuzzy logic control algorithm when the first switch is closed, and the opening degree of the first flow valve and the second flow valve is controlled by using an improved Q algorithm when the second switch is closed. The application can realize rapid and accurate adjustment of the temperature of a heating element.

Description

Technical Field

[0001] This invention belongs to the technical field of thermal management, and more specifically, relates to a temperature control system for a liquid-cooled thermal management device. Background Technology

[0002] The purpose of thermal management is to ensure that heat-generating components maintain optimal performance within a suitable temperature range and prevent thermal failure. Inappropriate heat dissipation strategies will damage the product, thereby affecting its stability and lifespan, and causing unnecessary economic losses. The heat generated by heat-generating components during operation will cause the product's temperature to rise, affecting its operating status and potentially causing safety issues. Especially for products with high operating temperature requirements, the control strategy of the thermal management system is more complex. Therefore, a good thermal management control strategy can effectively improve product performance and ensure the stable and efficient operation of the entire thermal management system, which has significant application value and prospects.

[0003] The common thermal management control strategies are mainly as follows: (1) Logic threshold control strategy. This method selects the output quantity to achieve the set constraint conditions during the early calibration. The advantages of this method are simplicity, strong robustness, and low hardware requirements. The disadvantage is that the adjustment error is large. (2) PID control strategy. This method is the most widely used strategy in industry. It has a wide range of applications, but the response speed is slow and the adjustment time is long. (3) Model predictive control strategy. This is a composite control method based on model feedforward compensation. It has high accuracy, but requires a long online calculation time and cost. (4) Intelligent control strategy. This uses algorithms such as logical fuzzy logic, neural network, genetic algorithm and deep learning to build system models to cope with complex thermal management systems that are nonlinear and time-varying.

[0004] Although control strategies for thermal management systems have been applied in various fields such as thermal simulation, casting, and energy storage systems, some problems still exist. Current control strategies can generally be divided into on / off control and regulation control. On / off control controls temperature according to pre-set commands; this method is simple but cannot control temperature in real time. Regulation control uses control algorithms to continuously adjust the strategy to achieve the desired goal; this method can achieve good temperature regulation, but it requires collecting large amounts of data and is computationally complex. The quality of the control results largely depends on the accuracy of the dynamic model and the accuracy of real-time detection; large deviations can worsen the results. Therefore, there is an urgent need to design a novel intelligent control strategy for thermal management simulation to achieve real-time temperature regulation and rapid adjustment of system state, thereby improving the operational stability and efficiency of the thermal management system. Summary of the Invention

[0005] In view of the above-mentioned defects or improvement needs of the prior art, the present invention provides a temperature control system for a liquid-cooled thermal management device, which can realize rapid and accurate adjustment of the temperature of the heating element.

[0006] To achieve the above objectives, according to one aspect of the present invention, a temperature control system for a liquid-cooled thermal management device is provided. The system includes: a cooling pipeline, on which a pump, a first flow valve, a sensor, a first switch, and a water-cooling plate are connected in series; the cooling pipeline further includes a liquid storage tank, a second flow valve, and a second switch connected in series, and the pipeline containing the liquid storage tank, the second flow valve, and the second switch is connected in parallel with the first switch; and a control console, which controls the opening and closing of the first and second switches and the valve openings of the first and second flow valves based on sensor data. The first and second switches are not opened and closed simultaneously. When the first switch is closed, a fuzzy logic control algorithm is used to control the opening of the first flow valve; when the second switch is closed, an improved Q algorithm is used to control the opening of the first and second flow valves.

[0007] Preferably, the improved Q algorithm is obtained based on the fuzzy control algorithm and the Q algorithm.

[0008] Preferably, the Q-algorithm is as follows:

[0009] θ Q =θ[Q=v π (v)]·λ1+θ[u=e·Rf]·2

[0010] Where, θ[Q=v π [v] represents the output of the Q-algorithm, and θ[u=e·Rf] represents the output of the fuzzy control algorithm. π(v) Let v be the value function. π (v)=E[R(v(0))+γR(v(1))+γ 2 R(v(2))+...+γ i R(v(i))], E is the expected value of the cumulative reward, and the goal of the Q-algorithm is to find the policy with the largest value function among all possible policies. v(i) is the i-th state of the system, v(i) = [v1(i), v2(i)], representing the coolant flow rate through the two control valves, where the initial state v(0) = [v1(0), v2(0)]. R(v(i)) is the reward, γ is the decay function, γ∈[0,1]. λ1 is the weight coefficient of the Q-algorithm, and λ2 is the weight coefficient of the fuzzy control algorithm. s1 is the surface temperature sensitivity factor of the heating element, s2 is the surface temperature difference sensitivity factor of the heating element, and λ1∈[λ1 min ,λ1 max ],λ2∈[λ2 min ,λ2max ].

[0011] Preferably, the console controls the opening and closing of the first switch and the second switch in the following manner: if the real-time temperature collected by the sensor is greater than the safe temperature threshold, the first switch is turned on; if the real-time temperature collected by the sensor is less than or equal to the safe temperature threshold, the second switch is turned on.

[0012] Preferably, the control variable in the fuzzy logic control algorithm is the coolant flow rate, which is adjusted by regulating the opening of the first flow valve.

[0013] Preferably, the specific method for controlling the opening degree of the first flow valve using a fuzzy logic control algorithm is as follows: S1: Set the difference between the target temperature and the current temperature as the observation quantity e, and use the change in the observation quantity e as the input quantity of the fuzzy control algorithm; simultaneously, set the opening degree u of the first flow valve as the control quantity; S2: Divide the observation quantity into multiple fuzzy sets; preferably, establish an evaluation index system based on the safety critical temperature, operating temperature, and ambient temperature of the thermal management system, and use a grey relational evaluation model to divide the observation quantity into five fuzzy sets. S3: Fuzzy inference: the larger the temperature deviation e, the faster the system needs to cool down, and the larger the corresponding u. By looking up the fuzzy rule table and combining it with the actual operation of the system, define the fuzzy relation matrix R. f The fuzzy reasoning process is as follows:

[0014] u=e·R f

[0015] Preferably, the observations are divided into five fuzzy sets using a grey relational evaluation model.

[0016] In summary, compared with the prior art, the temperature control system of the liquid-cooled thermal management device provided by the present invention has the following advantages:

[0017] 1. This application constructs a single-valve temperature control loop based on fuzzy logic fast response control to achieve rapid response of the thermal management system, enabling rapid cooling of heating elements, avoiding overheating, and reducing the time cost of temperature control. Simultaneously, it constructs a dual-valve temperature control loop based on improved Q-algorithm adaptive control, further improving the model's control accuracy. This can be used to reduce the maximum temperature difference on the surface of the heating element, improving the working performance of the heating element. Combining the advantages of both methods, appropriate temperature control measurements are selected based on real-time temperature and temperature difference, improving the system's cooling efficiency and enabling the system to respond promptly and efficiently in a replicated environment.

[0018] 2. An evaluation index system is established based on the safety critical temperature, operating temperature and ambient temperature of the thermal management system. The observed quantities are divided into multiple fuzzy sets using a grey relational evaluation model.

[0019] 3. The improved Q algorithm is based on fuzzy control algorithm and Q algorithm, which improves the response time of traditional Q algorithm, and can realize the measurement of heating elements with rapid temperature rise, thus improving response speed and accuracy.

[0020] 4. The control console selects the appropriate temperature adjustment method according to the control strategy, which improves both the temperature control response speed and the temperature control accuracy of the heating element. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the temperature control system of a liquid-cooled thermal management device.

[0022] Figure 2 This is a schematic diagram of the control flow of the console.

[0023] Figure 3 It is a hierarchical structure of fuzzy sets of observations.

[0024] In all the accompanying drawings, the same reference numerals are used to denote the same elements or structures, wherein:

[0025] 1-Pump; 2-First flow valve; 3-Heat exchanger; 4-First switch; 5-Second switch; 6-Storage tank; 7-Second flow valve; 8-Flow meter; 9-Water cooling plate; 10-Thermometer; 11-Control console; 12-Power supply; 13-Power control switch; 14-Cooling channel; 15-Control channel; 16-Measurement channel. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0027] The present invention provides a temperature control system for a liquid-cooled thermal management device, the system comprising cooling pipes and a control console, as detailed below.

[0028] like Figure 1As shown, the cooling pipeline includes a pump 1, a first flow valve 2, a sensor, a first switch 4, and a water-cooled plate 9 connected in series. The cooling pipeline also includes a liquid storage tank 6, a second flow valve 7, and a second switch 5 connected in series. The pipeline containing the liquid storage tank 6, the second flow valve 7, and the second switch 5 is connected in parallel with the first switch 4.

[0029] The sensors mainly include a flow meter 8 and a thermometer 10, which are the main sensors for collecting real-time data from the device.

[0030] Overall, the entire device can be divided into a cooling channel 14, a measuring channel 16, and a control channel 15. The cooling channel 14 is the flow path for the coolant, transferring heat within the device from the pump. The measuring channel 16 transmits the coolant flow rate and temperature information monitored by sensors to the control console. The control console sends control signals through the control channel 15 to regulate the entire system. The cooling channel 14 also includes a heat exchanger 3, a power supply 12 to provide power to the pump 1, and a power control switch 13.

[0031] The control console 11 controls the opening and closing of the first switch 4 and the second switch 5, as well as the valve opening of the first flow valve 2 and the second flow valve 7, based on sensor data. The first switch 4 and the second switch 5 are not opened and closed at the same time. When the first switch 4 is closed, a fuzzy logic control algorithm is used to control the opening of the first flow valve 2. When the second switch 5 is closed, an improved Q algorithm is used to control the opening of the first flow valve 2 and the second flow valve 7.

[0032] When the first switch 4 is closed and the second switch 5 is open, it is a single-valve temperature control circuit. At this time, the fuzzy logic control algorithm is used to adjust the opening of the first flow valve 2.

[0033] The specific method for controlling the opening degree of the first flow valve using a fuzzy logic control algorithm is as follows:

[0034] S1: Set the difference between the target temperature and the current temperature as the observation quantity e, and use the change in the observation quantity e as the input quantity of the fuzzy control algorithm; at the same time, set the opening degree u of the first flow valve as the control quantity.

[0035] The flow rate of the coolant is v, V = (π / 4) × d 2 / A, where d is the diameter of the water-cooled pipe and A is the cross-sectional area of ​​the water-cooled pipe.

[0036] The observed quantity e is set to the target temperature T. p and the current temperature T i The difference between the two values, and the change in the observed values ​​before and after the time interval, are used as the inputs for fuzzy control.

[0037] De = e(t+1) - e(t)

[0038] The coolant flow rate varies with the opening of the flow valve, therefore the valve opening u is set as the control variable.

[0039] S2: Divide the observations into multiple fuzzy sets;

[0040] The grey relational analysis method is preferably used to divide the observed quantity into multiple fuzzy sets. Based on three indicators—critical safety temperature, operating temperature, and ambient temperature—this invention establishes a hierarchical structure of the fuzzy set system for the observed quantity e, as follows: Figure 3 As shown. In this embodiment, there are 5 fuzzy sets, as follows: Figure 3 As shown, the values ​​include negative large (fuzzy set I), negative small (fuzzy set II), zero (fuzzy set III), positive small (fuzzy set IV), and significant (fuzzy set V). If e is negative, it indicates that the current average temperature of the heating element is lower than the target temperature. If e is positive, it indicates that the current average temperature of the heating element is higher than the target temperature. The range of values ​​for input e and output u is set according to the actual situation of the thermal management system.

[0041] A factor ranking model was established using grey relational analysis, {x i (k)} is the reference sequence, {x} j (k)} is the comparison sequence.

[0042] {x i (k)}={x i (1),x i (2),...,x i (m)},

[0043] {x j (k)}={x j (1), x j (2),...,x j (n)},

[0044] In the formula, i = 1, 2, ..., m; j = 1, 2, ..., n, and the sequence {x} j (k)} for the reference sequence {x i The correlation coefficient of the k terms at point i, i.e., the degree of correlation between the comparison sequence and the reference sequence at a certain point, δ. ij (k) is:

[0045]

[0046] In the formula It is the minimum difference between the two poles; β is the maximum difference between the two poles; β is the resolution coefficient, which is between 0 and 1, and is usually taken as β = 0.5.

[0047] By accumulating the correlation coefficients of each term (k = 1, 2, ..., m), we obtain the comparison sequence and the reference sequence {x}.i The degree of correlation of (k)}, i.e.

[0048]

[0049] Based on system requirements, four correlation indices (p, q, r, s) are determined, and the correlation values ​​are divided into five regions. The reference series is selected from the environmental temperature data set, and the comparison series is selected from the system input data e. j Calculate the correlation r between the comparison sequence and the reference sequence. ij Divide the data into five fuzzy sets as follows:

[0050]

[0051] S3: Fuzzy Reasoning

[0052] In this invention, if e increases, then u increases; if e = 0, then u = 0; if e decreases, then u decreases. A fuzzy relation matrix R is defined by consulting the fuzzy rule table and considering the actual operation of the system. f The fuzzy reasoning process is as follows:

[0053] u=e·R f

[0054] Using a single-valve temperature control loop can significantly improve the temperature control response speed of the thermal management system, and it is easy to control, enabling the thermal management system to cool down quickly.

[0055] When the first switch is closed and the second switch is closed, it is a dual-valve temperature control circuit. At this time, the improved Q algorithm is used to control the opening degree of the first flow valve and the second flow valve.

[0056] When the first switch is closed and the second switch is closed, a liquid storage tank and a new flow valve are connected in series after the heat exchanger to form a dual-valve control loop. The coolant flowing from the pump is temporarily stored in the liquid storage tank after passing through the first flow valve and the heat exchanger. The second flow valve further regulates the coolant flow rate. This method ensures reduced temperature loss of the coolant flowing through the heating element and minimizes flow loss during transmission, resulting in a more stable cooling rate and temperature for the heating element. The coolant flow rates through the flow valves are denoted as v1 and v2, respectively. Reinforcement learning is used for system temperature control. The dual-valve control enables more precise flow regulation, reduces the maximum temperature difference on the heating element surface, and makes its temperature changes smoother.

[0057] In reinforcement learning, the agent in state s t Take action a t To obtain new states t+1 Then take a new action a t+1 Iterate. The goal of reinforcement learning is to maximize the reward r. t .exist Figure 1 In this diagram, the coolant state through the first flow valve is denoted as s(j), the coolant state through the second flow valve is denoted as s(k), and the action adjusting the coolant flow rate is denoted as a. When the cooling circuit performs action a in state s, the reward based on the action and state is represented as R(s,a). The state-value function is expressed as:

[0058]

[0059] The initial state of the system is v(0) = [v1(0), v2(0)]. T When the controller takes an action, h(0) = [h1(0), h2(0)] T The system transitions to a new state v(1) = [v1(1), v2(1)] T The controller takes a new action h(1) and iterates until the desired state is reached:

[0060]

[0061] The goal of reinforcement learning methods is to maximize the reward value. The value function v π (v) defines the expected total of the discount rewards:

[0062] v π (v)=E[R(v(0))+γR(v(1))+γ 2 R(v(2))+...+γ i R(v(i))]

[0063] Where E is the expected value of the cumulative reward, v(i) is the i-th state of the coolant flow rate, v(i) = [v1(i), v2(i)], R(v(i)) is the reward, γ is the decay function, γ∈[0,1], and h is the coolant flow rate.

[0064] The Q algorithm is as follows:

[0065] θ Q =θ[Q=v π (v)]·λ1+θ[u=e·Rf]·λ2

[0066] Where, θ[Q=v π [v] represents the output of the Q-algorithm, and θ[u=e·Rf] represents the output of the fuzzy control algorithm. π(v) Let λ1 be the value function; λ2 be the weight coefficients of the Q-algorithm and λ3 be the weight coefficients of the fuzzy control algorithm. s1 is the surface temperature sensitivity factor of the heating element, s2 is the surface temperature difference sensitivity factor of the heating element, and λ1∈[λ1 min ,λ1 max ],λ2∈[λ2min ,λ2 max ].

[0067] The Q-algorithm can effectively reduce the surface temperature difference of the heating element. If the surface temperature difference sensitivity factor s2 is large, the weighting coefficient λ2 should be increased to enhance the influence of the Q-algorithm. The fuzzy control algorithm can quickly reduce the surface temperature of the heating element. If we place greater emphasis on the surface temperature sensitivity factor s1, the weighting coefficient λ1 should be increased to enhance the influence of the fuzzy control algorithm. The mapping relationship between the weighting coefficients and the sensitivity factor is set based on expert experience and testing of the actual operation of the thermal management system.

[0068] The control console controls the opening degrees of the first and second switches, as well as the valve opening degrees of the first and second flow valves, thereby altering the temperature control loop. When the surface temperature of the heating element is too high and rapid cooling is required, a single-valve loop with fuzzy control is selected through the decision module to avoid the risk of thermal runaway in the thermal management system. When the surface temperature difference of the heating element is large and the system temperature inconsistency is significant, a dual-valve loop with reinforcement learning control is adopted through the decision module to achieve precise temperature regulation and improve system efficiency.

[0069] This application cleans and transforms the real-time data collected by the sensor, inputs it into the control console, and then filters and analyzes the surface temperature and temperature difference data using a threshold filtering method. For example... Figure 2 As shown, during the operation of the heating element, the sensor monitors the real-time temperature Ti and maximum temperature difference Td of the element surface. When Ti is higher than the safe temperature threshold Tm, the system directly adopts a single-valve loop with fuzzy control to achieve rapid cooling and avoid the risk of thermal runaway. When Ti is lower than Tm, the system adopts a dual-valve loop with an improved Q algorithm to reduce the temperature difference on the heating element surface. When both the surface temperature and temperature difference are within the normal range, the system adjusts the system according to the pre-set sensitivity factor requirements using the improved Q algorithm to maintain stable and efficient operation.

[0070] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A temperature control system for a liquid-cooled thermal management device, characterized in that, The system includes: The cooling pipeline includes a pump, a first flow valve, a sensor, a first switch, and a water-cooled plate connected in series. The cooling pipeline also includes a liquid storage tank, a second flow valve, and a second switch connected in series. The pipeline containing the liquid storage tank, the second flow valve, and the second switch is connected in parallel with the first switch. The control console controls the opening and closing of the first and second switches, as well as the valve opening of the first and second flow valves, based on sensor data. The first and second switches do not open and close simultaneously. When the first switch is closed, a fuzzy logic control algorithm is used to control the opening of the first flow valve. When the second switch is closed, an improved Q algorithm is used to control the opening of the first and second flow valves. The improved Q algorithm is obtained based on the fuzzy control algorithm and the Q algorithm; The Q algorithm is as follows: in, The output of the Q algorithm, The output of the fuzzy control algorithm, For value function, E is the expected value of the cumulative reward. The goal of the Q-algorithm is to find the policy that maximizes the value function among all possible policies. Let i be the i-th state of the system. This represents the coolant flow rate through the two control valves, where the initial state... , R( In return, The decay function, , These are the weighting coefficients for the Q-algorithm. These are the weighting coefficients for the fuzzy control algorithm. , s1 is the surface temperature sensitivity factor of the heating element, and s2 is the surface temperature difference sensitivity factor of the heating element. , .

2. The system according to claim 1, characterized in that, The control console controls the opening and closing of the first and second switches in the following manner: If the real-time temperature collected by the sensor is greater than the safe temperature threshold, the first switch is turned on; if the real-time temperature collected by the sensor is less than or equal to the safe temperature threshold, the second switch is turned on.

3. The system according to claim 1, characterized in that, The control variable in the fuzzy logic control algorithm is the flow rate of the coolant, which is adjusted by regulating the opening of the first flow valve.

4. The system according to claim 1 or 3, characterized in that, The specific method for controlling the opening degree of the first flow valve using a fuzzy logic control algorithm is as follows: S1: Set the difference between the target temperature and the current temperature as the observation value. e The observed quantity e The change in the quantity is used as the input to the fuzzy control algorithm; at the same time, the opening degree of the first flow valve is used as the input. u Set as a control quantity; S2: Divide the observed quantity e into five fuzzy sets according to the range of temperature deviation, and divide the control quantity u into five fuzzy sets according to the degree of opening of the flow valve. Establish an evaluation index system based on the safety critical temperature, operating temperature and ambient temperature of the thermal management system, and divide the fuzzy sets based on the grey relational evaluation model. S3: Fuzzy inference. The larger the temperature deviation 'e', ​​the faster the system needs to cool down, and the larger the corresponding 'u'. By looking up the fuzzy rule table and combining it with the actual operating conditions of the system, a fuzzy relation matrix is ​​defined. Perform fuzzy reasoning calculations 。 5. The system according to claim 4, characterized in that, The observations were divided into five fuzzy sets using a grey relational evaluation model.

Images