Thermal management control method and system for new energy vehicle batteries

Abstract

The present application relates to the technical field of battery thermal management, and discloses a thermal management control method and system for new energy vehicle batteries, the method comprising: multi-level acquisition of battery pack temperature and combining wavelet decomposition algorithm to perform data denoising and temperature field mapping operations to obtain temperature distribution feature vectors; state space modeling of the temperature distribution feature vectors and execution of extended Kalman filter operations to obtain real-time temperature state observations; establishing a global temperature error function based on the real-time temperature state observations and executing Lyapunov stability analysis to obtain a hierarchical control strategy matrix; multi-threshold partitioning and state machine modeling of the temperature field to obtain a temperature control trigger sequence; bringing the temperature control trigger sequence into the global temperature stability index equation for solution, and combining with the exponential decay constraint condition to obtain the optimal control quantity, thereby realizing intelligent switching and smooth transition of the control strategy, and significantly reducing the oscillation and overshoot phenomena in the control process.

Description

Technical Field

[0001] The present application relates to the field of battery thermal management technology, and in particular to a thermal management control method and system for a new energy vehicle battery. Background Art

[0002] With the rapid development of the new energy vehicle industry, the safety and reliability issues of power batteries have become increasingly prominent. Battery packs generate a lot of heat during the charging and discharging process. If the heat cannot be dissipated in time, it is easy to cause problems such as uneven temperature distribution and local overheating, which seriously affect battery performance and service life. Traditional thermal management control methods mainly rely on simple temperature threshold triggers and fixed control strategies, which are difficult to cope with the complex heat dissipation characteristics of batteries under different working conditions.

[0003] Current battery thermal management systems generally have problems such as insufficient control accuracy and delayed response. Since the temperature distribution inside the battery pack has significant spatial non-uniformity and time-varying characteristics, traditional single-point temperature measurement and fixed parameter control methods are difficult to accurately grasp the temperature state of the entire battery pack. At the same time, there is a complex thermal coupling effect between the battery pack and the external environment, and factors such as ambient temperature changes and fluctuations in heat dissipation conditions will affect the control effect. Summary of the invention

[0004] The present application provides a thermal management control method and system for new energy vehicle batteries, thereby realizing intelligent switching and smooth transition of control strategies, and significantly reducing oscillation and overshoot phenomena during the control process.

[0005] In a first aspect, the present application provides a thermal management control method for a new energy vehicle battery, the thermal management control method for a new energy vehicle battery comprising:

[0006] The battery pack temperature is collected at multiple levels and the wavelet decomposition algorithm is used to perform data noise reduction and temperature field mapping operations to obtain the temperature distribution feature vector;

[0007] Performing state space modeling on the temperature distribution feature vector and performing extended Kalman filter operation to obtain a real-time temperature state observation;

[0008] Establishing a global temperature error function based on the real-time temperature state observation and performing Lyapunov stability analysis to obtain a hierarchical control strategy matrix;

[0009] Based on the hierarchical control strategy matrix, the temperature field is partitioned into multiple thresholds and state machine modeling is performed to obtain a temperature control trigger sequence;

[0010] The temperature control trigger sequence is brought into the global temperature stability index equation for solution, and combined with the exponential decay constraint condition, the optimal control quantity is obtained.

[0011] A second aspect of the present application provides a thermal management control device for a new energy vehicle battery, the thermal management control device for a new energy vehicle battery comprising:

[0012] The mapping module is used to collect the battery pack temperature at multiple levels and combine the wavelet decomposition algorithm to perform data noise reduction and temperature field mapping operations to obtain the temperature distribution feature vector;

[0013] A filtering operation module, used for performing state space modeling on the temperature distribution feature vector and performing an extended Kalman filter operation to obtain a real-time temperature state observation;

[0014] A stability analysis module, used to establish a global temperature error function according to the real-time temperature state observation and perform Lyapunov stability analysis to obtain a hierarchical control strategy matrix;

[0015] A modeling module, used to perform multi-threshold partitioning and state machine modeling on the temperature field based on the hierarchical control strategy matrix to obtain a temperature control trigger sequence;

[0016] The solution module is used to bring the temperature control trigger sequence into the global temperature stability index equation for solution, and combine the exponential decay constraint condition to obtain the optimal control amount.

[0017] The third aspect of the present application provides an electronic device, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor calls the instructions in the memory so that the electronic device executes the above-mentioned thermal management control method for new energy vehicle batteries.

[0018] A fourth aspect of the present application provides a computer-readable storage medium, in which instructions are stored. When the computer-readable storage medium is run on a computer, the computer executes the above-mentioned thermal management control method for a new energy vehicle battery.

[0019] Compared with the prior art, the present application has the following beneficial effects: through multi-level temperature acquisition and wavelet decomposition denoising, combined with the multi-scale feature extraction capability of deep neural networks, high-precision reconstruction of the battery pack temperature field is achieved, providing a reliable data basis for subsequent control strategy optimization. The extended Kalman filter algorithm is used to observe and correct the temperature state in real time, effectively overcoming the influence of measurement noise and model uncertainty, and improving the accuracy and reliability of temperature state estimation. Based on the Lyapunov stability theory, a global temperature error function is constructed, and the asymptotic stability of the control system is ensured through rigorous mathematical derivation, avoiding the risk of temperature control divergence. Through multi-threshold partitioning and state machine modeling, intelligent switching and smooth transition of control strategies are achieved, significantly reducing oscillation and overshoot phenomena during the control process. The hierarchical control architecture and pre-trigger mechanism are adopted, combined with the online optimization solution algorithm, which greatly reduces the computational complexity and meets the real-time requirements of the vehicle system. The introduction of adaptive control law and temperature disturbance compensation mechanism enables the system to effectively respond to changes in working conditions and environmental disturbances, and maintain good control performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0021] The structures, proportions, sizes, etc. illustrated in the drawings of this specification are only used to match the contents disclosed in the specification so as to facilitate understanding and reading by persons familiar with this technology. They are not used to limit the conditions under which the present invention can be implemented, and therefore have no substantive technical significance. Any structural modification, change in proportion or adjustment of size, without affecting the effects and purposes that can be achieved by the present invention, should still fall within the scope of the technical contents disclosed by the present invention.

[0022] Figure 1 It is a flow chart of a thermal management control method for a new energy vehicle battery provided by an embodiment of the present invention;

[0023] Figure 2 is a schematic block diagram of the structure of a thermal management control device for a new energy vehicle battery provided by an embodiment of the present invention;

[0024] Figure 3 It is a schematic block diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0025] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0026] The flowcharts shown in the accompanying drawings are only examples and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may also be decomposed, combined or partially merged, so the actual execution order may change according to actual conditions.

[0027] It should also be understood that the terms used in this application specification are only for the purpose of describing specific embodiments and are not intended to limit the application. As used in this application specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0028] It should be further understood that the term "and / or" used in the specification and appended claims of this application refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations. Figure 1 , an embodiment of the thermal management control method of the new energy vehicle battery in the embodiment of the present application includes:

[0029] Step 100, multi-level acquisition of battery pack temperature is performed and a wavelet decomposition algorithm is used to perform data noise reduction and temperature field mapping operations to obtain a temperature distribution feature vector;

[0030] It is understandable that the execution subject of the present application may be a thermal management control device for a new energy vehicle battery, or a terminal or a server, which is not specifically limited here. The present application embodiment is described by taking a server as the execution subject as an example.

[0031] Specifically, a high-precision temperature sensor array is arranged on the surface of the battery cell to collect temperature data in real time and generate a raw temperature data stream. The sensor array records the surface temperature of different areas of the battery at a high sampling rate to ensure that the key areas of the three levels of battery cells, modules and battery packs are covered, providing fine-grained thermal distribution information. The raw temperature data stream is input into a three-layer wavelet decomposition model based on the Daubechies 4 (db4) wavelet basis function to complete the multi-level decomposition operation. The temperature signal is multi-scale analyzed by the db4 wavelet basis function to effectively decompose the signal into low-frequency components and high-frequency components, which correspond to the main characteristic information and noise interference characteristics of the signal, respectively. The decomposition results are stored in the form of a temperature signal coefficient matrix. The temperature signal coefficient matrix is ​​thresholded for high-frequency noise and corrected for low-frequency baseline drift. The soft threshold method is used to denoise the high-frequency component. By setting a reasonable noise threshold, the high-frequency signal exceeding the threshold range is eliminated to eliminate the random error caused by environmental interference and sensor noise. At the same time, the baseline drift correction is implemented for the low-frequency component, and the polynomial fitting method is used to eliminate the long-term trend error, making the low-frequency signal more stable and real. After the above processing, the obtained denoised temperature coefficient can accurately reflect the real changes in the thermal field distribution inside the battery. The denoised temperature coefficient is substituted into the wavelet reconstruction equation, and the signal reconstruction is completed by inverse transformation calculation. By recombining and synthesizing the decomposed low-frequency and high-frequency components, the denoised temperature signal sequence is restored to generate a filtered temperature sequence. The filtered temperature sequence is subjected to zero-mean standardization and outlier removal operations to obtain a standardized temperature matrix. Zero-mean standardization adjusts the data to a uniform scale by subtracting the sample mean from each temperature data and dividing it by the standard deviation, which helps to avoid the weight imbalance problem caused by scale differences in subsequent modeling. At the same time, by setting a reasonable upper and lower limit range, outliers are removed to ensure the authenticity and validity of the data. The standardized temperature matrix is ​​organized and reconstructed according to the three levels of battery cells, modules and battery packs. According to the physical structure of the battery and the characteristics of the sensor layout, the data is reorganized to present the thermal distribution relationship at different levels, and a complete pre-processed temperature data set is constructed. This data set can reflect the thermal behavior of the single battery and reflect the overall temperature distribution of the module and the entire battery pack. The preprocessed temperature data set is input into a deep neural network with a multi-scale convolution structure to perform temperature field mapping operations. In the deep neural network, multi-scale convolution kernels are used to extract spatial features of different scales, and the fully connected layer is combined to achieve nonlinear mapping of high-dimensional features, thereby accurately characterizing the temperature distribution of the battery pack. Through end-to-end training of the network, the model can capture the complex distribution of the temperature field and generate a temperature distribution feature vector for describing the thermal state of the battery pack.

[0032] Step 200, performing state space modeling on the temperature distribution feature vector and performing an extended Kalman filter operation to obtain a real-time temperature state observation;

[0033] Specifically, the preprocessed temperature data set is input into a deep neural network with a multi-scale convolution structure, and the data is convolutionally operated through the multi-scale convolution layer to extract feature information at different scales. The multi-scale convolution layer captures the distribution characteristics of the temperature field in different spatial ranges by setting convolution kernels of different sizes, where small-scale convolution kernels extract local features and large-scale convolution kernels obtain global distribution information. After these convolution operations, a feature mapping matrix containing multi-scale feature information is generated to reflect the multi-level structure of the battery temperature field. Deep convolution operations are performed on the multi-scale feature mapping matrix to extract the hierarchical features of the battery temperature distribution. The deep convolution operation extracts high-level features in the temperature field layer by layer by increasing the number and complexity of the convolution layers, and more comprehensively describes the hierarchical characteristics of the temperature distribution. The hierarchical features obtained by the deep convolution are input into the maximum pooling layer for feature dimensionality reduction. The maximum pooling operation extracts the maximum value of the local area, compresses the dimension of the data, reduces redundant information, and retains key features to generate a reduced-dimensional feature tensor. The reduced-dimensional feature tensor is used for temperature uniformity calculation and hot spot area positioning operations to analyze the distribution characteristics of the battery temperature field. The temperature uniformity calculation quantifies the overall temperature equilibrium state of the battery pack by evaluating the consistency of the temperature distribution of each region in the feature tensor, while the hot spot area location operation uses the local extreme points in the feature tensor to locate the area with higher temperature. Through these operations, the high-temperature hot spots inside the battery and their distribution range are identified. The temperature spatial distribution features are input into the fully connected layer to perform temperature field reconstruction calculation. The fully connected layer reorganizes and maps the data by linearly combining and nonlinearly transforming the features to generate reconstructed temperature field data. The reconstructed temperature field data accurately reproduces the temperature distribution state inside the battery pack in space and can reflect its dynamic change trend. The reconstructed temperature field data is subjected to temperature gradient calculation and feature extraction operations. The temperature gradient calculation captures the non-uniformity and change trend of the temperature distribution inside the battery pack by evaluating the direction and rate of temperature change at each point in space. The feature extraction operation aggregates and summarizes the temperature gradient data to generate a temperature distribution feature vector that can fully reflect the thermal field state of the battery pack.

[0034] The key state variables including temperature value, temperature change rate and heat flux density are extracted from the temperature distribution feature vector, and the state equation describing the thermodynamic behavior of the battery is established by combining the heat conduction, convection heat transfer coefficient and radiation coefficient. The state equation is based on the principle of thermodynamics. Through Fourier's heat conduction law, convection heat transfer equation and radiation heat transfer equation, a dynamic model of the internal temperature of the battery changing with time and space is established, and the time-space evolution characteristics of the temperature field are accurately described by mathematical expressions. The measurement equation is constructed based on the temperature dynamic state model to associate the state variables with the measured temperature data of the sensor. The measurement equation provides a basis for the correction of actual observation data and model prediction by defining the mapping relationship between state variables and sensor output. Through this process, the state observation equation group is obtained, which contains a complete mathematical description between the dynamic model and the measurement data, so that theoretical prediction and actual measurement can work together. The state observation equation group is linearized. The nonlinear equation is approximated by Taylor expansion at the current state point, where the Jacobian matrix is ​​used to describe the sensitivity change of the system state to the input variables and noise. The calculation of the Jacobian matrix generates a state transfer matrix, which characterizes the dynamic evolution characteristics of the system from one state to the next state. Through this step, an extended state prediction model is obtained. Based on the extended state prediction model, the system is subjected to a priori state prediction operation. The priori prediction infers the state value at the next moment according to the current system state and input variables, and calculates the filter gain in combination with the measurement noise covariance matrix. The filter gain is a key parameter of the extended Kalman filter, which is used to balance the error influence between the model prediction and the actual measurement data to ensure that the estimated value of the system is more accurate. By optimizing the calculation of the filter gain, the interference of the measurement noise on the state estimation is effectively suppressed, while retaining the true state evolution trend of the system. The temperature state correction amount and the predicted state are weighted and fused to obtain the posterior state update value. The posterior update combines the priori prediction value and the measured correction value by weighted averaging, so that the system's estimate of the state can be closer to the true value. Based on the posterior state update value, the state covariance matrix is ​​dynamically corrected and calculated to adjust the uncertainty assessment of the system's prediction error in real time, so that the filtering algorithm can continuously adapt to complex thermal environment changes, and finally obtain the real-time temperature state observation.

[0035] Step 300: Establish a global temperature error function based on the real-time temperature state observation and perform Lyapunov stability analysis to obtain a hierarchical control strategy matrix;

[0036] It should be noted that the real-time temperature state observation is compared with the expected temperature field distribution, and the quadratic global temperature error function is constructed through the difference , which is used to quantify the deviation between the actual temperature distribution and the target temperature distribution. The form is expressed as ,in is the temperature error vector, is a symmetric positive definite matrix, which is used to weigh the impact of temperature errors at different locations on global stability. Through the construction of this error function, a mathematical model describing the dynamic characteristics of temperature field errors is generated. Finding the time derivative , and introduce negative definite constraints To ensure that the temperature error of the system decreases over time and gradually stabilizes. This constraint is based on Lyapunov stability theory. The symbolic nature of ensures that the temperature field dynamic model of the system has global stability. By converting the constraint conditions into stability criterion equations, the dynamic evolution path of the temperature field is defined. The stability criterion equation is substituted into the dynamic control law of the temperature field, and based on the temperature gradient Building a dynamic weight function . The design of the dynamic weight function adjusts the control intensity of each area according to the distribution characteristics of the temperature field. Among them, the area with a larger temperature gradient will be given a higher weight to give priority to eliminating local hot spots or cold spots and achieve balanced control of the global temperature. Combined with this weight function, the temperature balance control law is obtained. The control law aims to optimize the temperature error reduction rate and dynamically adjusts the temperature control strategy at different locations, so that the control process can efficiently adapt to complex heat distribution conditions. The temperature balance control law is hierarchically decomposed and the parameter update equations of the global temperature controller, module coordination controller and single cell precision controller are designed respectively. In the global control layer, the cooling and heating intensity of the battery pack as a whole is adjusted to ensure the uniformity of the macroscopic temperature field; in the module coordination layer, the distribution of the local temperature field is refined by coordinating the heat exchange between different modules; in the single cell precision control layer, the battery cells are fine-tuned to eliminate small temperature fluctuations and ensure the safe operation of the single cell. Through hierarchical decomposition, a hierarchical control parameter group is constructed, and the control objectives and parameter adjustment logic of each layer can work together clearly and efficiently. Based on the hierarchical control parameter group, an adaptive law is designed , the adaptive law combines the temperature error vector and the temperature state vector , dynamic matrix operations are performed to optimize the control parameters. The introduction of the adaptive law enables the control parameters to be automatically adjusted according to the actual changes in the temperature field, improving the adaptability and robustness of the system in complex thermal environments. Through the iterative update of the adaptive law, it gradually converges to the optimal control strategy, providing real-time decision support for the dynamic regulation of the system. The generated control parameter optimization matrix and the positive definite symmetric matrix Perform eigendecomposition operations to extract physically meaningful control characteristics and generate a hierarchical control strategy matrix. The hierarchical control strategy matrix is ​​the core output of the control method, covering all parameters and logic of global, module and single-cell control, and providing a comprehensive control strategy for the battery thermal management system.

[0037] Step 400: Based on the hierarchical control strategy matrix, the temperature field is partitioned into multiple thresholds and state machine modeling is performed to obtain a temperature control trigger sequence;

[0038] Specifically, the battery performance and temperature characteristics are analyzed for the hierarchical control strategy matrix. According to the performance of the battery in different temperature ranges, the operating temperature is divided into the optimal range, normal range and warning range. The optimal range is the working range with the highest battery efficiency and the lowest attenuation; the normal range is the range that the system can accept but the performance is slightly reduced; the warning range indicates that the temperature is close to the safety threshold and strong intervention measures are required for control. The results of the interval division are organized into a temperature threshold partition matrix to clarify the target control strategy in different temperature ranges. Based on the temperature threshold partition matrix, five working states are constructed, namely standby, low load, medium load, high load and emergency cooling. The standby state is used for the situation where the battery has low power consumption or is not working, emphasizing the minimum energy consumption; the low load, medium load and high load states are adapted to different battery operation intensity requirements, and the dynamic balance of the temperature field is achieved by controlling the cooling intensity; the emergency cooling state is specifically for the situation where the temperature is close to the upper limit of the warning range, and the temperature is quickly reduced by activating the maximum heat dissipation capacity to avoid safety hazards. At the same time, the transition conditions between the states are defined, such as dynamically triggering the state switching according to the temperature change rate, ambient temperature and load demand, and building a complete state machine model. The temperature change rate of the state machine model is analyzed to evaluate the dynamic characteristics of temperature change over time. Combined with the temperature prediction control algorithm, a state pre-trigger mechanism is constructed to enable the system to predict in advance that the temperature is about to enter a critical state and take measures. Through the accurate estimation of the future temperature trend by the prediction algorithm, a temperature state warning sequence is generated, which indicates the preventive control measures that the system needs to perform at different time points, thereby effectively reducing the uncontrollable risks caused by high temperature. According to the temperature state warning sequence, the corresponding control parameter configuration table is designed for the cooling fan speed, coolant flow rate and bypass valve opening. The configuration table matches the best control parameters according to different working states. For example, under low load conditions, the cooling fan and coolant flow rate are set to lower values ​​to reduce energy consumption; under high load conditions, the fan speed and coolant flow rate are increased to enhance the heat dissipation capacity; in the emergency cooling state, each cooling component operates at the maximum load to ensure that the system temperature drops rapidly. These parameter configuration tables are optimized and integrated into a hierarchical control parameter set. Based on the hierarchical control parameter set, a set of smooth transition control strategies is designed to achieve disturbance-free switching between different states. When switching states, the operating parameters of the cooling equipment are smoothly adjusted through interpolation algorithms or buffer control mechanisms to avoid system instability caused by sudden changes. Through the disturbance-free switching strategy, a state switching control sequence is generated, which records the time points, target parameters, and transition logic of each state switch. The state switching control sequence is organized according to time and priority to form a temperature control trigger sequence. In the priority setting, emergency cooling has the highest priority to ensure that battery safety is protected first under any circumstances; the priorities of high load, medium load, low load, and standby state decrease in turn.Through hierarchical control and priority trigger design, the temperature control trigger sequence dynamically adapts to different operating conditions, providing an efficient, accurate and safe temperature management solution.

[0039] Step 500: bring the temperature control trigger sequence into the global temperature stability index equation for solution, and combine it with the exponential decay constraint condition to obtain the optimal control amount.

[0040] Specifically, a global temperature stability index is constructed based on the temperature control trigger sequence , which is used to evaluate the stability and uniformity of the system temperature field. By analyzing the deviation between the temperature distribution and the target state, the global temperature stability index equation is defined as ,in represents the mean square error of the temperature field, which is used to measure the degree to which the temperature distribution deviates from the target value. It is a quadratic indicator of the temperature change rate and is used to suppress rapid temperature fluctuations to maintain the dynamic stability of the system. Used to weigh the importance of the two and adjust their values ​​according to specific needs. Analyze the control variables in the stability indicator equation, including coolant flow , Cooling fan speed and bypass valve opening . By performing stability constraint analysis on these control variables, their physical range and operating limits are clarified. For example, the maximum and minimum values ​​of the coolant flow rate depend on the performance of the pump, the cooling fan speed needs to consider the balance between energy consumption and efficiency, and the bypass valve opening is limited by the design of the coolant circulation path. These constraints ensure that the control quantity can meet the hardware limitations in actual operation without affecting the dynamic stability of the temperature field. Based on the control quantity constraints, the optimal convergence trajectory of the temperature field in the sliding time domain is calculated. The control strategy that makes the global temperature field converge to the target distribution along the optimal path is determined by the dynamic optimization method. In order to ensure the convergence speed and stability, an exponential decay constraint is introduced, that is, during the optimization process, the temperature field deviation is limited to decay exponentially over time to a stable range. This constraint is expressed in mathematical form as ,in is the temperature error, is the decay rate, by reasonably choosing Ensure the balance between convergence rate and control cost. Based on the temperature dynamic optimization function, an emergency temperature stabilization control strategy based on Lyapunov function reconstruction is established. By designing the Lyapunov function , analyze the global stability of the system to ensure that the control strategy can quickly converge to a safe temperature range in an emergency. Combined with the robust stability criterion, the control strategy is corrected and calculated to generate a temperature disturbance compensation law. This compensation law focuses on dealing with disturbances under complex operating conditions, such as external ambient temperature fluctuations or load changes, and achieves real-time compensation by dynamically adjusting the control variables to improve the stability and robustness of the system. On this basis, a hierarchical protection mechanism is constructed using the temperature disturbance compensation law to hierarchically manage the control quantities of coolant flow, fan speed, and bypass valve opening with different priorities. The global protection layer is responsible for the overall thermal balance regulation of the system; the local protection layer provides a rapid response to the overheating problem of a specific module; and the single cell protection layer eliminates the hot spots of a single battery by accurately adjusting the cooling parameters. Through this hierarchical design, the efficient operation and safe protection of the system under complex thermal field conditions are ensured. The control quantity is solved online, the optimization results are organized into a control quantity execution sequence, and input into the temperature control actuator. Through real-time closed-loop feedback correction, the deviation between the actual temperature field and the target state is detected, and the control parameters are dynamically adjusted to optimize the execution effect. The closed-loop control mechanism can continuously update the control strategy to ensure that the system always operates in the optimal state and ultimately obtains the optimal control quantity.

[0041] In the embodiment of the present application, through multi-level temperature acquisition and wavelet decomposition denoising, combined with the multi-scale feature extraction capability of deep neural network, high-precision reconstruction of battery pack temperature field is achieved, providing a reliable data basis for subsequent control strategy optimization. The extended Kalman filter algorithm is used to observe and correct the temperature state in real time, effectively overcoming the influence of measurement noise and model uncertainty, and improving the accuracy and reliability of temperature state estimation. Based on Lyapunov stability theory, a global temperature error function is constructed, and the asymptotic stability of the control system is ensured through rigorous mathematical derivation, avoiding the risk of temperature control divergence. Through multi-threshold partitioning and state machine modeling, intelligent switching and smooth transition of control strategy are achieved, significantly reducing oscillation and overshoot in the control process. The hierarchical control architecture and pre-trigger mechanism are adopted, combined with the online optimization solution algorithm, which greatly reduces the computational complexity and meets the real-time requirements of the vehicle system. The introduction of adaptive control law and temperature disturbance compensation mechanism enables the system to effectively respond to changes in working conditions and environmental disturbances, and maintain good control performance.

[0042] In a specific embodiment, the process of executing step 100 may specifically include the following steps:

[0043] Arrange a temperature sensor array on the surface of the battery cell to collect data, obtain the original temperature data stream, and input the original temperature data stream into the db4 wavelet basis function to perform a three-layer wavelet decomposition operation to obtain a temperature signal coefficient matrix;

[0044] The temperature signal coefficient matrix is ​​subjected to high-frequency noise thresholding and low-frequency baseline drift correction to obtain the noise reduction temperature coefficient, and the noise reduction temperature coefficient is substituted into the wavelet reconstruction equation for signal reconstruction calculation to obtain the filtered temperature sequence;

[0045] The filtered temperature sequence is subjected to zero-mean normalization and outlier removal operations to obtain a standardized temperature matrix, and the standardized temperature matrix is ​​reconstructed according to the three levels of battery cells, modules, and battery packs to obtain a preprocessed temperature data set;

[0046] The preprocessed temperature data set is input into a deep neural network with a multi-scale convolution structure to perform temperature field mapping operations and obtain the temperature distribution feature vector.

[0047] Specifically, a high-precision temperature sensor array is evenly arranged on the surface of the battery cell to ensure that the key thermal areas of the battery are fully covered. The temperature signals collected by the sensor array are recorded in the form of a time series to generate a raw temperature data stream. ,in Indicates Sensors at time The temperature value recorded at all times, are the spatial position indexes of the sensors respectively. Input the three-layer wavelet decomposition model based on Daubechies 4 (db4) wavelet basis function to perform hierarchical frequency domain analysis on the data. Wavelet decomposition is to transform the signal It is expressed as the sum of approximate components and detail components of different scales, and its expression is:

[0048] ;

[0049] in, is the low-frequency approximation coefficient, which represents the component of the signal in the low-frequency part. is the high-frequency detail coefficient, which indicates the changing characteristics of the signal in the high-frequency part. and They are the scale function and wavelet function of the db4 wavelet basis function. Through wavelet decomposition operation, the temperature signal coefficient matrix is ​​obtained ,in Contains the low-frequency and high-frequency characteristics of the signal. On this basis, the high-frequency coefficient Perform threshold processing on high-frequency noise. The removal of high-frequency noise adopts the soft threshold method, and its expression is:

[0050] ;

[0051] in, is the noise threshold, which is determined by empirical formula or cross-validation method. This process can effectively filter out high-frequency noise while retaining the main details. Perform baseline drift correction to eliminate non-stationary components caused by long-term signal changes and obtain more accurate noise reduction temperature coefficients . The noise reduction temperature coefficient Substitute the wavelet reconstruction equation to reconstruct the signal and restore the filtered temperature series The reconstruction formula is:

[0052] ;

[0053] After filtering is completed, Zero mean standardization is performed to make the data have a uniform scale so that subsequent modeling is more stable. The standardization formula is:

[0054] ;

[0055] in, and Respectively The temperature mean and standard deviation of each location. , replace the temperature values ​​that are out of range with the mean of the neighboring points. Organize and reconstruct the standardized temperature matrix according to the three levels of battery cells, modules and battery packs to form a preprocessed temperature data set The hierarchical structure of this dataset enables it to reflect the thermal distribution characteristics of the battery both internally and as a whole. Enter a deep neural network with a multi-scale convolutional structure. The network consists of multi-scale convolutional layers, pooling layers, and fully connected layers. The convolutional layers use convolution kernels of different sizes. Extract local and global features of temperature data. The convolution operation is expressed as:

[0056] ;

[0057] in, For the The output feature map value of the layer, is the convolution kernel weight, is the bias term, is an activation function (such as ReLU). After being processed by the deep neural network, a temperature distribution feature vector V is generated to describe the spatial distribution characteristics of the temperature field inside the battery.

[0058] In a specific embodiment, the execution step of inputting the preprocessed temperature data set into a deep neural network with a multi-scale convolution structure to perform temperature field mapping operation, and the process of obtaining the temperature distribution feature vector may specifically include the following steps:

[0059] The preprocessed temperature data set is input into a deep neural network with a multi-scale convolution structure, and convolution feature operations are performed on the preprocessed temperature data set through the multi-scale convolution layer of the deep neural network to obtain a multi-scale feature mapping matrix;

[0060] Perform deep convolution operations on the multi-scale feature mapping matrix to obtain the hierarchical features of the battery temperature distribution, and input the hierarchical features of the battery temperature distribution into the maximum pooling layer for feature dimensionality reduction to obtain the reduced-dimensional feature tensor;

[0061] The temperature uniformity calculation and hot spot area positioning operation are performed on the reduced dimension feature tensor to obtain the temperature spatial distribution characteristics, and the temperature spatial distribution characteristics are input into the fully connected layer to perform temperature field reconstruction calculation to obtain the reconstructed temperature field data;

[0062] The temperature gradient calculation and feature extraction operations are performed on the reconstructed temperature field data to obtain the temperature distribution feature vector.

[0063] Specifically, the temperature dataset is preprocessed Input deep neural network, its tensor form is ,in is the sample size, and Respectively represent the spatial height and width of the data, Indicates the number of channels. The multi-scale convolution layer of the deep neural network uses convolution kernels of different sizes to perform convolution feature operations on the input data. The convolution operation is expressed as:

[0064] ;

[0065] in It is Position in the feature map output by the convolution layer and Channel The value of is the convolution kernel weight, which represents the interaction between the input data and the convolution kernel. is the input value of the previous layer, is the bias term, is the activation function (e.g. ReLU: After the multi-scale convolution layer, the multi-scale feature mapping matrix is ​​obtained. ,in is the number of feature channels. Perform deep convolution operations to extract the hierarchical features of the battery temperature field. The deep convolution layer increases the number of convolution kernels and the number of convolution operations, allowing the model to capture the global trend and local details of the temperature distribution and obtain a higher-dimensional feature map. ,in and is the feature size after downsampling, is the number of feature channels of the output. Input the maximum pooling layer for dimensionality reduction. The formula for the pooling operation is:

[0066] ;

[0067] in is the position after pooling and Channel The value of Represents the input area corresponding to the pooling window. After the pooling operation, a reduced-dimensional feature tensor is generated. . For the reduced feature tensor Perform temperature uniformity calculation and hot spot location operation. Temperature uniformity is quantified by calculating the global variance of the feature tensor, and the formula is:

[0068] ;

[0069] in express The mean of is the total number of elements in the feature tensor. Hotspot location is achieved by extracting local extreme values ​​in the feature tensor to identify abnormal temperature rise areas. The temperature spatial distribution features are input into the fully connected layer for reconstruction calculation. The formula is:

[0070] ;

[0071] in represents the fully connected layer output values, is the input feature, and are weight and bias terms respectively, is the activation function. The reconstructed temperature field data generated by the fully connected layer It can accurately reproduce the temperature distribution of the battery pack. The temperature gradient is calculated for the reconstructed temperature field data to capture the changing trend of the temperature field. The gradient formula is:

[0072] ;

[0073] in and The temperature field is and Combine the temperature gradient characteristics and global uniformity characteristics to extract the final temperature distribution feature vector , used to fully characterize the thermal state of the battery.

[0074] In a specific embodiment, the process of executing step 200 may specifically include the following steps:

[0075] The state variables including temperature value, temperature change rate and heat flux density are constructed for the temperature distribution characteristic vector, and the state equation is established according to the heat conduction, convection heat transfer coefficient and radiation coefficient to obtain the temperature dynamic state model;

[0076] The measurement equation is established according to the temperature dynamic state model, and the measurement equation is mapped and calculated with the actual measured data of the sensor to obtain the state observation equation group;

[0077] The state observation equations are linearized, and the state transfer matrix is ​​calculated based on the Jacobian matrix to obtain the extended state prediction model;

[0078] Based on the extended state prediction model, the prior state prediction operation is performed, and the filter gain is calculated in combination with the measurement noise covariance matrix to obtain the temperature state correction value;

[0079] The temperature state correction value and the predicted state are weightedly fused to obtain the posterior state update value, and the state covariance matrix is ​​dynamically corrected and calculated based on the posterior state update value to obtain the real-time temperature state observation value.

[0080] Specifically, according to the temperature distribution feature vector , decompose it into three core state variables: temperature value , temperature change rate and heat flux .in, Indicates the instantaneous temperature of the battery surface; is the rate of change of temperature with time; Indicates heat flux density. Combined with heat conduction and convection heat transfer coefficients and emissivity The temperature dynamic state equation is established based on the physical parameters. According to Fourier's heat conduction law and energy conservation, the state equation is expressed as:

[0081] ;

[0082] in is the material density; is the specific heat capacity; k is the thermal conductivity; is the ambient temperature; is the Stefan-Boltzmann constant. The continuous partial differential equation is converted into a discrete-time state equation by discretization method, and the discrete form is obtained as follows:

[0083] ;

[0084] in is the state vector, is the state transition matrix, is the control input matrix, is the input vector, is the process noise. Based on the state equation, the measurement equation is established to associate the state variables with the actual sensor data. The mathematical form of the measurement equation is:

[0085] ;

[0086] in is the measurement vector, containing the temperature value measured by the sensor; is the measurement matrix, mapping the state variables to the measurement space; is the measurement noise. By combining the above measurement equation and state equation, a state observation equation group is formed. Since the state equation and measurement equation have nonlinear forms, they are linearized to facilitate the subsequent filtering operation. Linearization is performed by and Perform Taylor expansion and retain the first-order terms. The linearized expression of the state equation is:

[0087] ;

[0088] in is the Jacobian matrix of the state equation, defined as:

[0089] ;

[0090] Likewise, the linearized form of the measurement equation is:

[0091] ;

[0092] in is the Jacobian matrix of the measurement equation:

[0093] ;

[0094] Based on the linearized equation, an extended state prediction model is established. According to the a priori update formula of the extended Kalman filter, the system state is predicted and the a priori state is calculated. and the predicted covariance :

[0095] ;

[0096] ;

[0097] in is the process noise covariance matrix. Combined with the measurement noise covariance matrix , calculate the filter gain :

[0098] ;

[0099] Use the filter gain to update the state and get the posterior state and update covariance :

[0100] ;

[0101] ;

[0102] Through the above steps, the real-time temperature state observation is finally obtained , used for real-time monitoring and regulation of battery thermal management.

[0103] In a specific embodiment, the process of executing step 300 may specifically include the following steps:

[0104] The difference between the real-time temperature state observation and the expected temperature field distribution is calculated, and the quadratic global temperature error function V(t) is constructed to obtain the temperature field error dynamic model.

[0105] Based on the temperature field error dynamic model, the time derivative of the error function V(t) is obtained, and the negative definite constraint dV(t) / dt<0 is established to obtain the stability criterion equation.

[0106] Substitute the stability criterion equation into the dynamic control law of the temperature field, and construct the dynamic weight function ω(t) based on the temperature gradient ∇T to obtain the temperature equilibrium control law.

[0107] The temperature balance control law is decomposed hierarchically, and the parameter update equations of the global temperature controller, module coordination controller and single precision controller are established respectively to obtain a hierarchical control parameter group;

[0108] Based on the hierarchical control parameter group, an adaptive law Θ(t) is constructed, and the temperature error vector e(t) and the temperature state vector φ(t) are combined for matrix operation to obtain the control parameter optimization matrix;

[0109] Perform eigendecomposition on the control parameter optimization matrix and the positive definite symmetric matrix P to obtain the hierarchical control strategy matrix.

[0110] Specifically, the real-time temperature state observation and expected temperature field distribution Perform difference calculation to obtain temperature error distribution Based on the error distribution, a quadratic global temperature error function is constructed. , which is of the form:

[0111] ;

[0112] in is the temperature field domain, is a symmetric positive definite weight matrix used to adjust the error weights of different regions. is the error vector, which indicates the degree to which the temperature field deviates from the target distribution. The quadratic form ensures that ,and The system reaches the target temperature distribution when , find its time derivative , and combined with the temperature field error dynamic model The stability criterion equation is derived. Using the chain rule, the time derivative is expressed as:

[0113] ;

[0114] In stability analysis, in order to satisfy the Lyapunov stability condition, a negative definite constraint is established By constraining and properly designing the nonlinear terms of the system, it is ensured that the temperature error of the system can decrease over time and achieve global stability. Substitute the stability criterion equation into the dynamic control law of the temperature field and combine it with the temperature gradient Building a dynamic weight function The dynamic weight function is in the form of:

[0115] ;

[0116] in is the proportionality coefficient, Represents the square norm of the temperature field gradient, which is used to measure the intensity of temperature change. Combined with the weight function , the dynamic control law is expressed as:

[0117] ;

[0118] in is the control input vector, It is a control gain matrix that dynamically responds to temperature field changes by adjusting the control law. In order to achieve more detailed control, the temperature balance control law is hierarchically decomposed and the global temperature controller, module coordination controller and single precision controller are designed respectively. The global controller achieves the balance of macroscopic temperature distribution by adjusting the overall parameters of the cooling system (such as fan speed and coolant flow); the module coordination controller achieves the balance of macroscopic temperature distribution by adjusting the heat exchange coefficient between modules. Optimize the temperature distribution in the local area; the cell controller fine-tunes the temperature of a single battery cell. The parameter update equations are expressed as:

[0119] ;

[0120] ;

[0121] ;

[0122] in are the learning rate parameters of each layer. After completing the hierarchical control decomposition, the adaptive law is constructed based on the hierarchical control parameter group. , by dynamically adjusting the control gain to adapt to changes in different working conditions. The mathematical expression of the adaptive law is:

[0123] ;

[0124] in is the initial control gain, is the adaptive gain matrix, and They are the error vector and the temperature state vector respectively. , dynamically optimize the control effect. The control parameter optimization matrix With positive definite symmetric matrix Perform eigendecomposition operation to extract the hierarchical control strategy matrix. The eigendecomposition formula is:

[0125] ;

[0126] in is the eigenvalue, indicating the control strength; is an eigenvector, indicating the control direction. The hierarchical control strategy matrix achieves multi-level optimization from global to local by adjusting the eigenvalues ​​and eigenvectors of each level.

[0127] In a specific embodiment, the process of executing step 400 may specifically include the following steps:

[0128] The battery performance-temperature characteristics are analyzed for the hierarchical control strategy matrix, and the operating temperature range is divided into the optimal range, the normal range and the warning range to obtain the temperature threshold partition matrix;

[0129] Based on the temperature threshold partition matrix, five working states, namely standby, low load, medium load, high load and emergency cooling, are constructed, and the state transition conditions are defined to obtain the state machine model.

[0130] The temperature change rate of the state machine model is analyzed, and a state pre-trigger mechanism is established in combination with the temperature prediction control algorithm to obtain the temperature state warning sequence;

[0131] According to the temperature status warning sequence, a control parameter configuration table of cooling fan speed, coolant flow and bypass valve opening is set to obtain a hierarchical control parameter set;

[0132] A disturbance-free switching strategy is established based on the hierarchical control parameter set, and the control quantity is smoothly transitioned to obtain a state switching control sequence. The state switching control sequence is organized in time sequence and a trigger priority is established to obtain a temperature control trigger sequence.

[0133] Specifically, for the hierarchical control strategy matrix Conduct battery performance-temperature characteristics analysis, including Indicates the control level (such as global control, module control, and single unit control). Indicates control parameters (such as cooling fan speed, coolant flow, bypass valve opening, etc.). Combined with the battery's performance-temperature characteristic curve, the operating temperature range is divided into the optimal range by analyzing the relationship between battery efficiency, life and safety and temperature. , Normal range and warning zone The division of these intervals is determined by the following conditions:

[0134] ;

[0135] in represents the battery efficiency, is the temperature. The optimal interval corresponds to the temperature range with the highest efficiency and the smallest life attenuation. The normal interval is the acceptable range, and the warning interval requires protective regulation. Based on this analysis, the temperature threshold partition matrix is ​​generated. , each column of the matrix corresponds to the temperature threshold of a different control level. According to the temperature threshold partition matrix, five working states of standby, low load, medium load, high load and emergency cooling are constructed, and the transition conditions of each state are defined. The standby state is suitable for low-power scenarios with the lowest cooling demand; the low-load and medium-load states correspond to medium working intensities, and the cooling system operates in a lower energy consumption mode; the high-load state adapts to high-power output scenarios, and the cooling intensity is significantly increased; the emergency cooling state is used when the temperature approaches or exceeds the warning interval, and the system runs at full speed to reduce the temperature. Based on the state machine model, combined with the temperature change rate analysis and the temperature prediction control algorithm, a state pre-trigger mechanism is established to generate a temperature state warning sequence. Through the predictive control algorithm, the temperature change trend at future moments is calculated, and the prediction formula is:

[0136] ;

[0137] in is the predicted temperature at the next moment, is the sampling time interval, is the current temperature change rate. If the prediction results show that the temperature will enter the warning range in the future, emergency cooling or high load mode will be triggered in advance to ensure safe operation of the system. The pre-trigger status at each time point is recorded. Based on the temperature status warning sequence, the cooling fan speed is set , Coolant flow and bypass valve opening Each working state corresponds to a set of fixed control parameter values. For example, in the emergency cooling state, the fan speed is set to the maximum value. , the coolant flow rate is , bypass valve fully open The configuration table is as follows:

[0138]

[0139] On this basis, in order to achieve disturbance-free operation during state switching, a smooth transition strategy is designed, and the control parameters are adjusted using a linear interpolation algorithm. The formula is:

[0140] ;

[0141] in and are the control parameters of the two states respectively, and It is the start and end time of the state switching. Organize the state switching control sequence according to the time sequence, and build the temperature control trigger sequence in combination with the trigger priority.

[0142] In this embodiment, before setting the control parameter configuration table of the cooling fan speed, coolant flow rate and bypass valve opening according to the temperature control trigger sequence, it also includes: establishing a proxy Lagrangian relaxation model for the battery thermal management control problem, and decomposing the temperature control constraints to obtain multiple independent temperature subsystems; constructing the temperature subsystem as a Markov decision process consisting of a state space S, an action space A and a reward function R, and defining a temperature state transition probability matrix to obtain a reinforcement learning environment model; constructing a dual network of a strategy network and a value network for the reinforcement learning environment model, and setting a data storage structure of an experience replay pool to obtain a deep reinforcement learning framework; and performing a deep reinforcement learning on each temperature based on the deep reinforcement learning framework. The temperature subsystems are trained in parallel, and the time difference algorithm is used to iteratively calculate the value function to obtain the optimal action strategy; the mapping relationship of the temperature control parameters is constructed according to the optimal action strategy, and the action space is discretized to obtain the control parameter optimization table; the control parameter optimization table is updated by the proxy Lagrange multiplier, and the parameters are coordinated in combination with the temperature constraints to obtain the global optimal control configuration; based on the global optimal control configuration, the control parameters of each temperature subsystem are fine-tuned online, and a parameter smooth transition mechanism is established to obtain a dynamic control parameter set; the dynamic control parameter set is organized according to the priority of the temperature subsystem, and mapped and matched with the temperature threshold partitions to obtain a hierarchical control parameter set.

[0143] In a specific embodiment, the process of executing step 500 may specifically include the following steps:

[0144] The global temperature stability index J(t) is constructed based on the temperature control trigger sequence, and the global temperature stability index equation including the mean square error of the temperature field E(t) and the quadratic form of the temperature change rate Q(t) is established.

[0145] The stability constraint analysis is performed on the control variables of coolant flow u1(t), cooling fan speed u2(t) and bypass valve opening u3(t) in the global temperature stability index equation to obtain the control variable constraint conditions.

[0146] Based on the control quantity constraint condition, the optimal convergence trajectory of the temperature field in the sliding time domain is calculated, and the exponential decay constraint condition is introduced to obtain the temperature dynamic optimization function;

[0147] According to the temperature dynamic optimization function, an emergency temperature stabilization control strategy based on Lyapunov reconstruction is established, and correction operation is performed in combination with the robust stability criterion to obtain the temperature disturbance compensation law.

[0148] A hierarchical protection mechanism is constructed based on the temperature disturbance compensation law, and the control quantity is solved online to obtain the control quantity execution sequence. The control quantity execution sequence is input into the temperature control actuator, and closed-loop feedback correction is performed to obtain the optimal control quantity.

[0149] Specifically, the sequence is triggered by temperature control , define the global temperature stability index , which includes the mean square error of the temperature field and the quadratic form of the temperature change rate The mathematical expression of the global temperature stability index is:

[0150] ;

[0151] in is the temperature field domain, is the weight coefficient, which is used to balance the influence of mean square error and rate of change on stability. Defined as:

[0152] ;

[0153] in is the actual temperature field distribution, is the target temperature field distribution, is the area of ​​the domain. The quadratic form of the temperature change rate is It is expressed as:

[0154] ;

[0155] in is the rate of change of temperature over time. In order to realize the stability constraint analysis of the control system, we focus on the coolant flow , Cooling fan speed and bypass valve opening The physical constraints of these control quantities are described in the following form:

[0156] ;

[0157] in The upper and lower limits of the coolant flow rate and the fan speed, the bypass valve opening The range of is limited to [0,1], indicating fully closed and fully open. Based on the control quantity constraint, the optimal convergence trajectory of the temperature field in the sliding domain is calculated. Assuming that the length of the sliding domain is , the goal is to The temperature field deviation is converged to the allowable range. By introducing the exponential decay constraint, the exponential decay behavior of the temperature error over time is limited, and its mathematical expression is:

[0158] ;

[0159] in is the attenuation coefficient, which indicates the speed at which the temperature field converges. is the initial time. Combined with the above conditions, the goal of the temperature dynamic optimization function is to minimize :

[0160] ;

[0161] In order to ensure that the temperature quickly stabilizes in an emergency, a reconstruction control strategy based on the Lyapunov function is established. Defined as:

[0162] ;

[0163] And find its time derivative:

[0164] ;

[0165] By introducing control input Correct the temperature change rate and perform correction calculations in combination with the robust stability criterion to obtain the temperature disturbance compensation law:

[0166] ;

[0167] in is the gain matrix, is the temperature error vector, Represents the temperature gradient and is used to quickly respond to local temperature changes. A hierarchical protection mechanism is constructed based on the compensation law, and the control strategy is divided into a global control layer, a module coordination layer, and a single-unit precision layer. The global layer optimizes the fan speed, the module layer adjusts the coolant flow, and the single-unit layer controls the bypass valve opening. Real-time online solution of the control quantity execution sequence:

[0168] ;

[0169] And input temperature control actuators, such as cooling fans, liquid pumps and valves. Through closed-loop feedback correction, monitor the deviation between actual temperature and target temperature, dynamically adjust the control input, and ensure that the system achieves optimal temperature control under all working conditions.

[0170] The above describes the thermal management control method of the new energy vehicle battery in the embodiment of the present application. The following describes the thermal management control device 10 of the new energy vehicle battery in the embodiment of the present application. Figure 2 In the embodiment of the present application, an embodiment of the thermal management control device 10 for a new energy vehicle battery includes:

[0171] A mapping module 11 is used to collect the battery pack temperature at multiple levels and perform data noise reduction and temperature field mapping operations in combination with a wavelet decomposition algorithm to obtain a temperature distribution feature vector;

[0172] The filtering operation module 12 is used to perform state space modeling on the temperature distribution feature vector and perform extended Kalman filtering operation to obtain a real-time temperature state observation;

[0173] A stability analysis module 13 is used to establish a global temperature error function according to the real-time temperature state observation and perform Lyapunov stability analysis to obtain a hierarchical control strategy matrix;

[0174] A modeling module 14 is used to perform multi-threshold partitioning and state machine modeling on the temperature field based on a hierarchical control strategy matrix to obtain a temperature control trigger sequence;

[0175] The solution module 15 is used to bring the temperature control trigger sequence into the global temperature stability index equation for solution, and combine the exponential decay constraint condition to obtain the optimal control amount.

[0176] Through the collaboration of the above components, multi-level temperature acquisition and wavelet decomposition noise reduction, combined with the multi-scale feature extraction capability of deep neural networks, high-precision reconstruction of the battery pack temperature field is achieved, providing a reliable data basis for subsequent control strategy optimization. The extended Kalman filter algorithm is used to observe and correct the temperature state in real time, effectively overcoming the influence of measurement noise and model uncertainty, and improving the accuracy and reliability of temperature state estimation. Based on the Lyapunov stability theory, a global temperature error function is constructed, and the asymptotic stability of the control system is ensured through rigorous mathematical derivation, avoiding the risk of temperature control divergence. Through multi-threshold partitioning and state machine modeling, intelligent switching and smooth transition of control strategies are achieved, significantly reducing oscillation and overshoot during the control process. The hierarchical control architecture and pre-trigger mechanism are adopted, combined with the online optimization solution algorithm, which greatly reduces the computational complexity and meets the real-time requirements of the vehicle system. The introduction of adaptive control law and temperature disturbance compensation mechanism enables the system to effectively respond to changes in working conditions and environmental disturbances, and maintain good control performance.

[0177] See also Figure 3 , Figure 3 This is a schematic block diagram of the structure of an electronic device 300 provided in an embodiment of the present application. The electronic device 300 includes a processor 301 and a memory 302. The processor 301 and the memory 302 are connected via a device bus 303, wherein the memory 302 may include a non-volatile storage medium and an internal memory.

[0178] The non-volatile storage medium can store a computer program. The computer program includes program instructions, and when the program instructions are executed by the processor 301, the processor 301 can execute any of the above-mentioned thermal management control methods for new energy vehicle batteries.

[0179] The processor 301 is used to provide computing and control capabilities to support the operation of the entire electronic device 300 .

[0180] The internal memory provides an environment for the operation of the computer program in the non-volatile storage medium. When the computer program is executed by the processor 301, the processor 301 can execute any of the above-mentioned thermal management control methods for new energy vehicle batteries.

[0181] Those skilled in the art will understand that Figure 3 The structure shown in the figure is only a block diagram of a partial structure related to the present application scheme, and does not constitute a limitation on the electronic device 300 involved in the present application scheme. The specific electronic device 300 may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0182] It should be understood that the processor 301 may be a central processing unit (CPU), and the processor 301 may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0183] It should be noted that technicians in the relevant field can clearly understand that for the convenience and simplicity of description, the specific working process of the electronic device 300 described above can refer to the corresponding process of the thermal management control method of the aforementioned new energy vehicle battery, and will not be repeated here.

[0184] An embodiment of the present application also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by one or more processors, the one or more processors implement the thermal management control method for a new energy vehicle battery as provided in an embodiment of the present application.

[0185] The computer-readable storage medium may be an internal storage unit of the electronic device 300 in the aforementioned embodiment, such as a hard disk or memory of the electronic device 300. The computer-readable storage medium may also be an external storage device of the electronic device 300, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped with the electronic device 300.

[0186] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0187] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable an electronic device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store program codes.

[0188] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A thermal management control method for a new energy vehicle battery, characterized in that: The method comprises: The battery pack temperature is collected at multiple levels and the wavelet decomposition algorithm is used to perform data noise reduction and temperature field mapping operations to obtain the temperature distribution feature vector; Performing state space modeling on the temperature distribution feature vector and performing extended Kalman filter operation to obtain a real-time temperature state observation; Establishing a global temperature error function based on the real-time temperature state observation and performing Lyapunov stability analysis to obtain a hierarchical control strategy matrix; Based on the hierarchical control strategy matrix, the temperature field is partitioned into multiple thresholds and state machine modeling is performed to obtain a temperature control trigger sequence; The temperature control trigger sequence is brought into the global temperature stability index equation for solution, and combined with the exponential decay constraint condition, the optimal control quantity is obtained.

2. The thermal management control method of new energy vehicle battery according to claim 1, characterized in that: The multi-level acquisition of the battery pack temperature and the combination of wavelet decomposition algorithm to perform data noise reduction and temperature field mapping operations to obtain the temperature distribution feature vector include: Arrange a temperature sensor array on the surface of the battery cell to collect data, obtain an original temperature data stream, and input the original temperature data stream into the db4 wavelet basis function to perform a three-layer wavelet decomposition operation to obtain a temperature signal coefficient matrix; Performing high-frequency noise thresholding and low-frequency baseline drift correction on the temperature signal coefficient matrix to obtain a noise reduction temperature coefficient, and substituting the noise reduction temperature coefficient into a wavelet reconstruction equation to perform signal reconstruction calculation to obtain a filtered temperature sequence; Performing zero-mean normalization and outlier elimination operations on the filtered temperature sequence to obtain a standardized temperature matrix, and reconstructing the standardized temperature matrix according to three levels of battery cells, modules, and battery packs to obtain a preprocessed temperature data set; The preprocessed temperature data set is input into a deep neural network with a multi-scale convolution structure to perform temperature field mapping operations to obtain a temperature distribution feature vector.

3. The thermal management control method of the new energy vehicle battery according to claim 2, characterized in that: The preprocessed temperature data set is input into a deep neural network with a multi-scale convolution structure to perform temperature field mapping operation to obtain a temperature distribution feature vector, including: Inputting the preprocessed temperature data set into a deep neural network with a multi-scale convolution structure, performing convolution feature operation on the preprocessed temperature data set through the multi-scale convolution layer of the deep neural network, and obtaining a multi-scale feature mapping matrix; Performing deep convolution operations on the multi-scale feature mapping matrix to obtain battery temperature distribution level features, and inputting the battery temperature distribution level features into a maximum pooling layer to perform feature dimensionality reduction processing to obtain a reduced dimensionality feature tensor; Performing temperature uniformity calculation and hot spot area positioning operation on the dimension-reduced feature tensor to obtain temperature spatial distribution features, and inputting the temperature spatial distribution features into a fully connected layer to perform temperature field reconstruction calculation to obtain reconstructed temperature field data; The reconstructed temperature field data is subjected to temperature gradient calculation and feature extraction operations to obtain a temperature distribution feature vector.

4. The thermal management control method of the new energy vehicle battery according to claim 3, characterized in that: The state space modeling of the temperature distribution feature vector and the execution of the extended Kalman filter operation to obtain the real-time temperature state observation quantity include: Constructing state variables including temperature value, temperature change rate and heat flux density for the temperature distribution characteristic vector, and establishing state equations according to heat conduction, convection heat transfer coefficient and radiation coefficient to obtain a temperature dynamic state model; Establishing a measurement equation according to the temperature dynamic state model, and mapping and calculating the measurement equation with the actual measured data of the sensor to obtain a state observation equation group; Linearizing the state observation equation group, and calculating the state transfer matrix based on the Jacobian matrix to obtain an extended state prediction model; Performing a priori state prediction operation based on the extended state prediction model, and calculating filter gain in combination with the measurement noise covariance matrix to obtain a temperature state correction amount; A weighted fusion operation is performed on the temperature state correction amount and the predicted state to obtain a posterior state update value, and a dynamic correction calculation is performed on the state covariance matrix based on the posterior state update value to obtain a real-time temperature state observation amount.

5. The thermal management control method of the new energy vehicle battery according to claim 4, characterized in that: The method of establishing a global temperature error function according to the real-time temperature state observation and performing Lyapunov stability analysis to obtain a hierarchical control strategy matrix includes: Calculate the difference between the real-time temperature state observation and the expected temperature field distribution, and construct a quadratic global temperature error function V(t) to obtain a temperature field error dynamic model; Based on the temperature field error dynamic model, the time derivative of the error function V(t) is obtained, and a negative definite constraint condition dV(t) / dt<0 is established to obtain a stability criterion equation; Substituting the stability criterion equation into the temperature field dynamic control law, and constructing a dynamic weight function ω(t) based on the temperature gradient ∇T, to obtain the temperature equilibrium control law; Performing hierarchical decomposition operation on the temperature balance control law, and establishing parameter update equations of the global temperature controller, the module coordination controller and the single-body precise controller respectively, to obtain a hierarchical control parameter group; Based on the hierarchical control parameter group, an adaptive law Θ(t) is constructed, and a matrix operation is performed on the temperature error vector e(t) and the temperature state vector φ(t) to obtain a control parameter optimization matrix; Performing eigendecomposition operation on the control parameter optimization matrix and the positive definite symmetric matrix P, a hierarchical control strategy matrix is ​​obtained.

6. The thermal management control method of the new energy vehicle battery according to claim 5, characterized in that: The temperature field is partitioned into multiple threshold values ​​and state machine modeled based on the hierarchical control strategy matrix to obtain a temperature control trigger sequence, including: Performing battery performance-temperature characteristic analysis on the hierarchical control strategy matrix, and dividing the operating temperature range into an optimal range, a normal range, and a warning range, to obtain a temperature threshold partition matrix; Based on the temperature threshold partition matrix, five working states of standby, low load, medium load, high load and emergency cooling are constructed, and state transition conditions are defined to obtain a state machine model; Performing temperature change rate analysis on the state machine model, and establishing a state pre-trigger mechanism in combination with a temperature prediction control algorithm to obtain a temperature state warning sequence; According to the temperature state warning sequence, a control parameter configuration table of cooling fan speed, coolant flow rate and bypass valve opening is set to obtain a hierarchical control parameter set; Based on the hierarchical control parameter set, a disturbance-free switching strategy is established, and the control amount is smoothly transitioned to obtain a state switching control sequence. The state switching control sequence is organized in time sequence and a trigger priority is established to obtain a temperature control trigger sequence.

7. The thermal management control method of a new energy vehicle battery according to claim 6, characterized in that: The temperature control trigger sequence is brought into the global temperature stability index equation for solution, and combined with the exponential decay constraint condition to obtain the optimal control quantity, including: Based on the temperature control trigger sequence, a global temperature stability index J(t) is constructed, and a global temperature stability index equation including a temperature field mean square error E(t) and a temperature change rate quadratic form Q(t) is established; Perform stability constraint analysis on the control quantities of coolant flow u1(t), cooling fan speed u2(t) and bypass valve opening u3(t) in the global temperature stability index equation to obtain control quantity constraint conditions; Based on the control quantity constraint condition, the optimal convergence trajectory of the temperature field in the sliding time domain is calculated, and an exponential decay constraint condition is introduced to obtain a temperature dynamic optimization function; According to the temperature dynamic optimization function, an emergency temperature stabilization control strategy based on Lyapunov reconstruction is established, and correction operation is performed in combination with a robust stability criterion to obtain a temperature disturbance compensation law; A hierarchical protection mechanism is constructed based on the temperature disturbance compensation law, and the control quantity is solved online to obtain a control quantity execution sequence, which is input into a temperature control actuator, and a closed-loop feedback correction is performed to obtain an optimal control quantity.

8. A thermal management control device for a new energy vehicle battery, characterized in that: Used to execute the thermal management control method of a new energy vehicle battery according to any one of claims 1 to 7, the thermal management control device of the new energy vehicle battery comprises: The mapping module is used to collect the battery pack temperature at multiple levels and combine the wavelet decomposition algorithm to perform data noise reduction and temperature field mapping operations to obtain the temperature distribution feature vector; A filtering operation module, used for performing state space modeling on the temperature distribution feature vector and performing an extended Kalman filter operation to obtain a real-time temperature state observation; A stability analysis module, used to establish a global temperature error function according to the real-time temperature state observation and perform Lyapunov stability analysis to obtain a hierarchical control strategy matrix; A modeling module, used to perform multi-threshold partitioning and state machine modeling on the temperature field based on the hierarchical control strategy matrix to obtain a temperature control trigger sequence; The solution module is used to bring the temperature control trigger sequence into the global temperature stability index equation for solution, and combine the exponential decay constraint condition to obtain the optimal control amount.

9. An electronic device, characterized in that: The electronic device comprises: a memory and at least one processor, wherein instructions are stored in the memory; The at least one processor calls the instructions in the memory so that the electronic device executes the thermal management control method for a new energy vehicle battery as described in any one of claims 1-7.

10. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instructions are executed by the processor, the thermal management control method for the new energy vehicle battery as described in any one of claims 1 to 7 is implemented.

Images