Battery temperature prediction method and device based on EEMD-GRU, and medium
Through the EEMD-GRU method, battery temperature prediction is carried out in combination with specific data of the battery, and the problems of high computational complexity and low prediction accuracy in the prior art are solved, real-time and accurate prediction of battery temperature are achieved.
Patent Information
- Application Number
- CN202410925242.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-11
- Publication Date
- 2024-11-12
AI Technical Summary
The existing battery temperature prediction methods have problems such as high computational complexity, limited predictive capabilities for nonlinear and non-stationary data, and are sensitive to noise, making it difficult to achieve real-time and accurate battery temperature prediction.
By using the EEMD-GRU-based method, the SOC-OCV curve, entropy thermal coefficient, charge and discharge current value, voltage value and heat production data of the battery are obtained, and the battery temperature rise data is decomposed using the EEMD algorithm to obtain the target IMF component and residual signal. And use GRU neural network to train the battery temperature prediction model to reduce the computational complexity and improve the prediction accuracy.
It effectively reduces the calculation complexity of battery temperature prediction, improves the accuracy of prediction results, reduces the impact of noise on prediction results, and realizes real-time and accurate prediction of battery temperature.
Smart Images

Figure CN118940604A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to but is not limited to the field of battery temperature prediction technology, and in particular to a battery temperature prediction method, device, and medium based on EEMD-GRU. Background Art
[0002] The existing battery temperature prediction methods include the following three types: (1) Battery temperature prediction method based on physical model. This method needs to collect accurate model parameters, such as thermal conductivity and heat capacity of battery materials. The accuracy of these model parameters needs to be obtained through a large amount of experimental data and theoretical calculations to identify. In addition, the physical model contains a large number of partial differential equations and parameters, which makes the model solution computationally complex and difficult to achieve real-time prediction of battery temperature; (2) Battery temperature prediction method based on statistics and traditional time series analysis. The accuracy of battery temperature prediction using this method depends on its properties. For example, it works well when processing linear time series data, but has limited prediction ability for nonlinear and non-stationary data, which means that the scope of application of this method for battery temperature prediction is limited; (3) Battery temperature prediction method based on machine learning. This method can capture complex nonlinear relationships in the data to be predicted, but is highly sensitive to noise data and cannot guarantee the accuracy of battery temperature prediction results. Based on this, there is an urgent need for a battery temperature prediction method that can reduce computational complexity, achieve real-time prediction of battery temperature, and ensure temperature prediction accuracy. Summary of the invention
[0003] The embodiments of the present application provide a battery temperature prediction method, device, and medium based on EEMD-GRU, which can effectively reduce the computational complexity of predicting battery temperature while ensuring the accuracy of the prediction results.
[0004] In a first aspect, an embodiment of the present application provides a battery temperature prediction method based on EEMD-GRU, comprising:
[0005] Obtaining a first SOC-OCV curve, a first entropy thermal coefficient, a first charge and discharge current value, a first voltage value, a first state of charge, a first heat generation, and battery temperature rise data corresponding to each first battery to be tested;
[0006] Decomposing each of the battery temperature rise data by a preset EEMD algorithm to obtain a target IMF component and a residual signal, and superimposing all of the target IMF components to obtain a recombined signal;
[0007] A preset GRU neural network is trained based on all of the first SOC-OCV curve, the first entropy thermal coefficient, the first charge and discharge current value, the first voltage value, the first state of charge, the first heat generation, the battery temperature rise data, the residual signal, and the recombined signal to obtain a battery temperature prediction model;
[0008] Obtaining a second SOC-OCV curve, a second entropy thermal coefficient, a second charge and discharge current value, and a second voltage value corresponding to a second battery to be tested, and calculating a second state of charge and a second heat generation corresponding to the second battery to be tested according to the second SOC-OCV curve, the second entropy thermal coefficient, the second charge and discharge current value, and the second voltage value, wherein the second battery to be tested is different from all the first batteries to be tested;
[0009] The second charge and discharge current value, the second voltage value, the second state of charge, and the second heat generation are input into the battery temperature prediction model to obtain a target temperature prediction result.
[0010] In some embodiments, decomposing each of the battery temperature rise data by a preset EEMD algorithm to obtain a target IMF component and a residual signal includes:
[0011] Based on a preset number of times, Gaussian white noise is added to each of the battery temperature rise data for multiple times to obtain multiple intermediate signals, wherein the amount of noise added to each of the intermediate signals of each of the battery temperature rise data is different from each other;
[0012] Performing EMD decomposition on each of the intermediate signals to obtain a corresponding initial IMF component;
[0013] An average value of the initial IMF components corresponding to each of the intermediate signals is taken to obtain the target IMF components and the residual signal corresponding to each of the battery temperature rise data.
[0014] In some embodiments, performing EMD decomposition on each of the intermediate signals to obtain a corresponding initial IMF component includes:
[0015] determining local maxima and local minima in each of the intermediate signals;
[0016] Connecting all the local maxima by an interpolation algorithm to obtain an upper envelope;
[0017] Connecting all the local minimum values by the interpolation algorithm to obtain a lower envelope;
[0018] Calculating the average value of the upper envelope and the lower envelope to obtain the trend component of each of the intermediate signals;
[0019] The target battery temperature rise data is subtracted from the trend component to obtain the initial IMF component, wherein the target battery temperature rise data is the original battery temperature rise data corresponding to the intermediate signal.
[0020] In some embodiments, the GRU neural network includes a first sub-GRU network and a second sub-GRU network, the battery temperature prediction model includes a first target sub-GRU network and a second target sub-GRU network, the GRU neural network includes a first sub-GRU network and a second sub-GRU network, the battery temperature prediction model includes a first target sub-GRU network and a second target sub-GRU network, based on all of the first SOC-OCV curve, the first entropy thermal coefficient, the first charge and discharge current value, the first voltage value, the first state of charge, the first heat generation, the battery temperature rise data, the residual signal and the recombined signal, a preset GRU neural network is trained to obtain a battery temperature prediction model, including:
[0021] Based on a preset data set division ratio, all of the first SOC-OCV curve, the first entropy thermal coefficient, the first charge and discharge current value, the first voltage value, the first state of charge, the first heat generation, the battery temperature rise data, the residual signal, and the recombined signal are divided into a training set, a test set, and a validation set;
[0022] Acquire the first charge-discharge current value, the first voltage value, the first state of charge, and the first heat generation from the training set, and input the first charge-discharge current value, the first voltage value, the first state of charge, and the first heat generation into the first sub-GRU network to obtain a first output result;
[0023] Inputting the first charge-discharge current value, the first voltage value, the first state of charge, and the first heat generation into the second sub-GRU network to obtain a second output result;
[0024] Comparing the residual signal with the first output result to obtain a first comparison result;
[0025] Comparing the recombinant signal with the second output result to obtain a second comparison result;
[0026] Calculate a first loss value based on a first loss function and the first comparison result, and calculate a second loss value based on a second loss function and the second comparison result;
[0027] Training the first sub-GRU network based on the first loss value and the first loss function to obtain a first intermediate network;
[0028] Training the second sub-GRU network based on the second loss value and the second loss function to obtain a second intermediate network;
[0029] Acquire third sample data from the validation set and the test set, and adjust the first intermediate network based on the third sample data to obtain the first target sub-GRU network, wherein the third sample data has the same data type as the first sample data;
[0030] Acquire fourth sample data from the validation set and the test set, adjust the second intermediate network based on the fourth sample data, and obtain the second target sub-GRU network, wherein the fourth sample data has the same data type as the second sample data.
[0031] In some embodiments, calculating the second state of charge and the second heat generation corresponding to the second battery to be tested according to the second SOC-OCV curve, the second entropy thermal coefficient, the second charge and discharge current value, and the second voltage value includes:
[0032] Determining an initial state of charge of the second battery to be tested according to the second SOC-OCV curve;
[0033] Calculating the second state of charge according to the second charge-discharge current value and the initial state of charge based on the ampere-hour integration method;
[0034] Based on the Bernadi heat generation equation, the second heat generation amount is calculated according to the second charge and discharge current value, the second entropy thermal coefficient and the second voltage value.
[0035] In some embodiments, the second state of charge is calculated based on the ampere-hour integration method according to the second charge-discharge current value and the initial state of charge, and is obtained according to the following formula:
[0036]
[0037] Wherein, SOC0 is the initial state of charge, I is the second charge and discharge current value, t is time, C is the total capacity of the battery, and SOC is the second state of charge;
[0038] The second voltage value includes the battery terminal voltage and the open circuit voltage corresponding to the second battery to be tested. The second heat generation is calculated based on the Bernadi heat generation equation according to the second charge and discharge current value, the second entropy thermal coefficient and the second voltage value, and is obtained according to the following formula:
[0039]
[0040] Wherein, I is the second charge and discharge current value, E is the open circuit voltage, U is the battery terminal voltage, and T is the Kelvin temperature of the second battery to be tested. is the second entropy thermal coefficient.
[0041] In some embodiments, the battery temperature prediction model includes an update gate and a reset gate, and the second charge and discharge current value, the second voltage value, the second state of charge, and the second heat generation are input into the battery temperature prediction model to obtain a target temperature prediction result, which is obtained according to the following formula:
[0042] r t =σ(W r ·[h t-1 ,x t ]);
[0043] z t =σ(W r ·[h t-1 ,x t ]);
[0044]
[0045] h t =(1-z t )*h t-1 +z t *h t ;
[0046] y t =σ(W o ·h t +c);
[0047] Among them, x t is the second charge and discharge current value, the second voltage value, the second state of charge and the second heat generation currently input, h t is the hidden state vector at time t, h t-1 is the hidden state vector at time t-1, σ is the first activation function in the hidden layer of the battery temperature prediction model, r t is the output of the reset gate at time t, z t is the output of the update gate at time t, W r is the weight matrix of the update gate, W o is the output weight matrix, c is the output bias, y t is the target temperature prediction result, is the candidate hidden state vector at time t, and tanh is the second activation function.
[0048] In a second aspect, an embodiment of the present application provides a battery temperature prediction device based on EEMD-GRU, comprising:
[0049] An offline data acquisition module, used to obtain a first SOC-OCV curve, a first entropy thermal coefficient, a first charge and discharge current value, a first voltage value, a first state of charge, a first heat generation, and battery temperature rise data corresponding to each first battery to be tested;
[0050] A temperature rise data decomposition module, used to decompose each of the battery temperature rise data by a preset EEMD algorithm to obtain a target IMF component and a residual signal, and to superimpose all of the target IMF components to obtain a recombined signal;
[0051] A model training module, used for training a preset GRU neural network based on all of the first SOC-OCV curve, the first entropy thermal coefficient, the first charge and discharge current value, the first voltage value, the first state of charge, the first heat generation, the battery temperature rise data, the residual signal and the recombined signal to obtain a battery temperature prediction model;
[0052] an online data acquisition module, for obtaining a second SOC-OCV curve, a second entropy thermal coefficient, a second charge and discharge current value, and a second voltage value corresponding to a second battery to be tested, and calculating a second state of charge and a second heat generation corresponding to the second battery to be tested according to the second SOC-OCV curve, the second entropy thermal coefficient, the second charge and discharge current value, and the second voltage value, wherein the second battery to be tested is different from all the first batteries to be tested;
[0053] The temperature estimation module is used to input the second charge and discharge current value, the second voltage value, the second state of charge and the second heat generation into the battery temperature prediction model to obtain a target temperature prediction result.
[0054] In a third aspect, an embodiment of the present application provides a control device, comprising at least one control processor and a memory for communicating with the at least one control processor; the memory stores instructions executable by the at least one control processor, and the instructions are executed by the at least one control processor so that the at least one control processor can execute the battery temperature prediction method based on EEMD-GRU as described in the first aspect.
[0055] In a fourth aspect, an embodiment of the present application further provides a computer-readable storage medium storing computer-executable instructions, wherein the computer-executable instructions are used to execute the battery temperature prediction method based on EEMD-GRU as described in the first aspect.
[0056] The embodiment of the present application provides a battery temperature prediction method, device, and medium based on EEMD-GRU, and the method includes: obtaining a first SOC-OCV curve, a first entropy thermal coefficient, a first charge and discharge current value, a first voltage value, a first state of charge, a first heat generation, and battery temperature rise data corresponding to each first battery to be tested; decomposing each of the battery temperature rise data by a preset EEMD algorithm to obtain a target IMF component and a residual signal, and superimposing all of the target IMF components to obtain a recombined signal; based on all of the first SOC-OCV curves, the first entropy thermal coefficient, the first charge and discharge current value, the first voltage value, the first state of charge, the first heat generation, the battery temperature rise data According to the present invention, the preset GRU neural network is trained with the residual signal and the recombined signal to obtain a battery temperature prediction model; the second SOC-OCV curve, the second entropy thermal coefficient, the second charge and discharge current value and the second voltage value corresponding to the second battery to be tested are obtained, and the second state of charge and the second heat generation corresponding to the second battery to be tested are calculated according to the second SOC-OCV curve, the second entropy thermal coefficient, the second charge and discharge current value and the second voltage value, wherein the second battery to be tested is different from all the first batteries to be tested; the second charge and discharge current value, the second voltage value, the second state of charge and the second heat generation are input into the battery temperature prediction model to obtain a target temperature prediction result. According to the solution provided in the embodiment of the present application, the EEMD algorithm is combined with the GRU model to realize battery temperature prediction, which effectively reduces the computational complexity of predicting the battery temperature while reducing the influence of noise on the temperature prediction result, thereby ensuring the accuracy of the prediction result. BRIEF DESCRIPTION OF THE DRAWINGS
[0057] Figure 1 is a flowchart of the steps of a battery temperature prediction method based on EEMD-GRU provided in one embodiment of the present application;
[0058] Figure 2 is a module schematic diagram of a battery temperature device based on EEMD-GRU provided in another embodiment of the present application;
[0059] Figure 3 It is a structural diagram of a control device provided in another embodiment of the present application. DETAILED DESCRIPTION
[0060] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.
[0061] It is understood that although the functional modules are divided in the device schematic diagram and the logical order is shown in the flow chart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flow chart. The terms "first", "second", etc. in the specification, claims or the above drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.
[0062] The existing battery temperature prediction methods include the following three types: (1) Battery temperature prediction method based on physical model. This method needs to collect accurate model parameters, such as thermal conductivity and heat capacity of battery materials. The accuracy of these model parameters needs to be obtained through a large amount of experimental data and theoretical calculations to identify. In addition, the physical model contains a large number of partial differential equations and parameters, which makes the model solution computationally complex and difficult to achieve real-time prediction of battery temperature; (2) Battery temperature prediction method based on statistics and traditional time series analysis. The accuracy of battery temperature prediction using this method depends on its properties. For example, it works well when processing linear time series data, but has limited prediction ability for nonlinear and non-stationary data, which means that the scope of application of this method for battery temperature prediction is limited; (3) Battery temperature prediction method based on machine learning. This method can capture complex nonlinear relationships in the data to be predicted, but is highly sensitive to noise data and cannot guarantee the accuracy of battery temperature prediction results. Based on this, there is an urgent need for a battery temperature prediction method that can reduce computational complexity, achieve real-time prediction of battery temperature, and ensure temperature prediction accuracy.
[0063] In order to solve the above-mentioned problems, the embodiments of the present application provide a battery temperature prediction method, device, and medium based on EEMD-GRU, the method comprising: obtaining a first SOC-OCV curve, a first entropy thermal coefficient, a first charge and discharge current value, a first voltage value, a first state of charge, a first heat generation, and battery temperature rise data corresponding to each first battery to be tested; decomposing each of the battery temperature rise data by a preset EEMD algorithm to obtain a target IMF component and a residual signal, and superimposing all of the target IMF components to obtain a recombined signal; based on all of the first SOC-OCV curves, the first entropy thermal coefficient, the first charge and discharge current value, the first voltage value, the first state of charge, the first heat generation, the The battery temperature rise data, the residual signal and the recombined signal are used to train a preset GRU neural network to obtain a battery temperature prediction model; the second SOC-OCV curve, the second entropy thermal coefficient, the second charge and discharge current value and the second voltage value corresponding to the second battery to be tested are obtained, and the second state of charge and the second heat generation corresponding to the second battery to be tested are calculated according to the second SOC-OCV curve, the second entropy thermal coefficient, the second charge and discharge current value and the second voltage value, wherein the second battery to be tested is different from all the first batteries to be tested; the second charge and discharge current value, the second voltage value, the second state of charge and the second heat generation are input into the battery temperature prediction model to obtain a target temperature prediction result. According to the solution provided in the embodiment of the present application, the EEMD algorithm is combined with the GRU model to realize battery temperature prediction, which effectively reduces the computational complexity of predicting the battery temperature while reducing the influence of noise on the temperature prediction result, thereby ensuring the accuracy of the prediction result.
[0064] The embodiments of the present application are further described below in conjunction with the accompanying drawings.
[0065] refer to Figure 1 , a flowchart of a battery temperature prediction method based on EEMD-GRU provided by an embodiment of the present application, an embodiment of the present application provides a battery temperature prediction method based on EEMD-GRU, the method includes but is not limited to the following steps:
[0066] Step S10, obtaining a first SOC-OCV curve, a first entropy thermal coefficient, a first charge and discharge current value, a first voltage value, a first state of charge, a first heat generation, and battery temperature rise data corresponding to each first battery to be tested;
[0067] Step S20, decomposing the temperature rise data of each battery by a preset EEMD algorithm to obtain a target IMF component and a residual signal, and superimposing all the target IMF components to obtain a recombined signal;
[0068] Step S30, training a preset GRU neural network based on all the first SOC-OCV curves, the first entropy thermal coefficient, the first charge and discharge current value, the first voltage value, the first state of charge, the first heat generation, the battery temperature rise data, the residual signal and the recombined signal to obtain a battery temperature prediction model;
[0069] Step S40, obtaining a second SOC-OCV curve, a second entropy thermal coefficient, a second charge and discharge current value, and a second voltage value corresponding to the second battery to be tested, and calculating a second state of charge and a second heat generation corresponding to the second battery to be tested according to the second SOC-OCV curve, the second entropy thermal coefficient, the second charge and discharge current value, and the second voltage value, wherein the second battery to be tested is different from all the first batteries to be tested;
[0070] Step S50, inputting the second charge and discharge current value, the second voltage value, the second state of charge and the second heat generation into a battery temperature prediction model to obtain a target temperature prediction result.
[0071] It should be noted that, in the embodiment of the present application, the first battery to be tested is a sample battery, and the specific method of obtaining the first SOC-OCV curve, the first entropy thermal coefficient, the first charge and discharge current value, the first voltage value, the first state of charge, the first heat generation and the battery temperature rise data corresponding to each first battery to be tested can be obtained by performing an offline experiment on the first battery to be tested. For example, the first battery to be tested is subjected to an HPPC test, and the HPPC test includes multiple groups of current pulse tests, and the state of charge SOC of each group of current pulse tests differs by 10%. When a group of current pulse tests is performed on the first battery to be tested, the next group of current pulse tests can be performed only after the first battery to be tested enters electrochemical and thermal equilibrium. Specifically, the process of the HPPC test is as follows: (1) Fully charge the first battery to be tested (2) Let it stand for 1 hour (3) Perform a constant current pulse test on the first battery to be tested (4) Discharge the first battery to be tested by 10% SOC (5) and then stand for 1 hour (6) repeat steps (1) to (5) until the power of the first battery to be tested is consumed by 90%, and the SOC-OCV curve of the first battery to be tested is obtained after the HPPC test is completed; the first entropy thermal coefficient of the present application can be obtained according to the following steps: a temperature characteristic analysis experiment is performed on the first battery to be tested, and the specific experimental process is as follows: (1) the temperature of the thermostat is set so that the thermostat changes slowly within a certain temperature range, for example, from a low temperature (such as 0°C) to a high temperature (such as 50°C), and data is recorded every preset temperature value (such as 5°C); (2) after each change of the thermostat temperature, the first battery to be tested is allowed to stand for a preset time at the new temperature to ensure that the temperature and voltage reach a balanced state, and at each temperature point, the open circuit voltage of the first battery to be tested is measured and recorded; (3) a curve is drawn between the measured open circuit voltage and the corresponding temperature, and the entropy thermal coefficient of the first battery to be tested is calculated according to the slope of the curve.
[0072] It should be noted that the second SOC-OCV curve, the second entropy thermal coefficient, the second charge and discharge current value and the second voltage value of the present embodiment are obtained by online real-time acquisition.
[0073] It should be noted that, in this embodiment, the first SOC-OCV curve, the second SOC-OCV curve, the first entropy thermal coefficient and the second entropy thermal coefficient are battery characteristic attribute parameters, and the first charge and discharge current value, the second charge and discharge current value and the first voltage value and the second voltage value are temperature characteristic value parameters. Acquiring the above data in this embodiment can provide an effective data basis for training the preset GRU neural network.
[0074] Additionally, in some embodiments, Figure 1 The step of decomposing each battery temperature rise data by a preset EEMD algorithm to obtain a target IMF component and a residual signal in step S20 includes but is not limited to the following steps:
[0075] Step S21, adding Gaussian white noise to each battery temperature rise data for multiple times based on a preset number of times to obtain multiple intermediate signals, wherein the amount of noise added to each intermediate signal of each battery temperature rise data is different from each other;
[0076] Step S22, performing EMD decomposition on each intermediate signal to obtain a corresponding initial IMF component;
[0077] Step S23, taking the average value of the initial IMF components corresponding to each intermediate signal to obtain the target IMF components and residual signals corresponding to each battery temperature rise data.
[0078] It should be noted that the embodiment of the present application does not limit the multiple addition of Gaussian white noise to each battery temperature rise data. The specific number of times of obtaining multiple intermediate signals and the amount of Gaussian white noise added can be determined by technical personnel in this field according to actual needs. It can be understood that multiple addition of Gaussian white noise to each battery temperature rise data can ensure that different characteristics of the battery temperature rise data can be fully manifested under different noises.
[0079] It can be understood that the embodiment of the present application decomposes the battery temperature rise data through the EEMD algorithm to obtain multiple target IMF components and residual signals, and then superimposes all the target IMF components to obtain a recombined signal, that is, the battery temperature rise data is decomposed into two parts (residual signal and recombined signal) through the EEMD algorithm. Since the GRU neural network of the present application includes two sub-neural networks, the residual signal and the recombined signal can be used as learning targets of the two sub-neural networks respectively. In this way, in the process of training the GRU neural network, the two sub-neural networks of the GRU neural network can learn the temperature rise characteristics of the battery system more accurately, providing an effective data basis for improving the accuracy of battery temperature prediction.
[0080] In addition, in some embodiments, step S22 includes but is not limited to the following steps:
[0081] Step S221, determining the local maximum and local minimum in each intermediate signal;
[0082] Step S222, connecting all local maxima by an interpolation algorithm to obtain an upper envelope;
[0083] Step S223, connecting all local minimum values by an interpolation algorithm to obtain a lower envelope;
[0084] Step S224, calculating the average value of the upper envelope and the lower envelope to obtain the trend component of each intermediate signal;
[0085] Step S225 , subtracting the trend component from the target battery temperature rise data to obtain an initial IMF component, wherein the target battery temperature rise data is the original battery temperature rise data corresponding to the intermediate signal.
[0086] It can be understood that this embodiment determines the local maximum and local minimum in each intermediate signal; connects all the local maxima through an interpolation algorithm to obtain an upper envelope; connects all the local minima through an interpolation algorithm to obtain a lower envelope; calculates the average value of the upper envelope and the lower envelope to obtain the trend component of each intermediate signal; subtracts the trend component from the target battery temperature rise data to obtain the initial IMF component, which can provide an effective data basis for calculating the target IMF component and then calculating the recombined signal.
[0087] In addition, in some embodiments, the GRU neural network includes a first sub-GRU network and a second sub-GRU network, the battery temperature prediction model includes a first target sub-GRU network and a second target sub-GRU network, and step S30 includes but is not limited to the following steps:
[0088] Step S301, based on a preset data set division ratio, all first SOC-OCV curves, first entropy thermal coefficients, first charge and discharge current values, first voltage values, first state of charge, first heat generation, battery temperature rise data, residual signals, and recombined signals are divided into a training set, a test set, and a validation set;
[0089] Step S302, obtaining a first charge-discharge current value, a first voltage value, a first state of charge, and a first heat generation from a training set, and inputting the first charge-discharge current value, the first voltage value, the first state of charge, and the first heat generation into a first sub-GRU network to obtain a first output result;
[0090] Step S303, inputting the first charge and discharge current value, the first voltage value, the first state of charge and the first heat generation into the second sub-GRU network to obtain a second output result;
[0091] Step S304, comparing the residual signal with the first output result to obtain a first comparison result;
[0092] Step S305, comparing the residual signal with the first output result to obtain a first comparison result;
[0093] Step S306, calculating a first loss value based on the first loss function and the first comparison result, and calculating a second loss value based on the second loss function and the second comparison result;
[0094] Step S307, training the first sub-GRU network based on the first loss value and the first loss function to obtain a first intermediate network;
[0095] Step S308, training the second sub-GRU network based on the second loss value and the second loss function to obtain a second intermediate network;
[0096] Step S309, obtaining third sample data from the validation set and the test set, adjusting the first intermediate network based on the third sample data, and obtaining a first target sub-GRU network, wherein the third sample data has the same data type as the first sample data;
[0097] Step S310, obtaining fourth sample data from the validation set and the test set, adjusting the second intermediate network based on the fourth sample data, and obtaining a second target sub-GRU network, wherein the fourth sample data has the same data type as the second sample data.
[0098] It should be noted that the embodiments of the present application do not involve specific improvements to the first loss function and the second loss function, and those skilled in the art may select them according to actual needs.
[0099] It can be understood that the GRU neural network in the embodiment of the present application is obtained by changing the recurrent neural network (RNN). The GRU neural network solves the problems of gradient diffusion and gradient explosion that occur during the RNN training process, and simplifies the structure of the LSTM, so that the present application uses the trained GRU neural network, that is, the battery temperature prediction model, to predict the battery temperature with higher efficiency.
[0100] It should be noted that the embodiments of the present application do not limit the specific data set division ratio. The data volume ratio of the training set, test set and validation set of the present embodiment can be 7:2:1, and the technical personnel in this field can adjust it according to actual needs. In the three sets of training set, test set and validation set, each data set has 9 data types including the first SOC-OCV curve, the first entropy thermal coefficient, the first charge and discharge current value, the first voltage value, the first state of charge, the first heat generation, the battery temperature rise data, the residual signal and the recombined signal. The difference lies in the different data volume.
[0101] It should be noted that the first sub-GRU network and the second sub-GRU network of the embodiment of the present application are both single-output structures. The output results of the two GRU sub-neural networks are superimposed to obtain the final battery temperature prediction result. Therefore, the recombined signal and the residual signal are respectively used as the training targets (i.e., target label data) of the first sub-GRU network and the second sub-GRU network, and the corresponding loss values are calculated. Based on the corresponding loss functions and loss values, the two sub-GRU networks are trained and optimized respectively to obtain the first target sub-GRU network and the second target sub-GRU network, which can ensure the accuracy of the results output by the two target sub-GRU networks after training, and thus ensure the accuracy of the target temperature prediction results output by the trained battery temperature prediction model.
[0102] Additionally, in some embodiments, Figure 1 Step S40 calculates the second state of charge and the second heat generation corresponding to the second battery to be tested according to the second SOC-OCV curve, the second entropy thermal coefficient, the second charge and discharge current value and the second voltage value, including but not limited to the following steps:
[0103] Step S41, determining the initial state of charge of the second battery to be tested according to the second SOC-OCV curve;
[0104] Step S42, calculating a second state of charge according to the second charge-discharge current value and the initial state of charge based on the ampere-hour integration method;
[0105] Step S43, based on the Bernadi heat generation equation, calculate the second heat generation according to the second charge and discharge current value, the second entropy thermal coefficient and the second voltage value.
[0106] It should be noted that, in this embodiment, the second state of charge is calculated based on the ampere-hour integration method according to the second charge-discharge current value and the initial state of charge, and is obtained according to the following formula:
[0107]
[0108] Among them, SOC0 is the initial state of charge, I is the second charge and discharge current value, t is time, C is the total capacity of the battery, and SOC is the second state of charge.
[0109] It should be noted that the second voltage value of this embodiment includes the open circuit voltage and the battery terminal voltage corresponding to the second battery to be tested. Based on the Bernadi heat generation equation, the second heat generation is calculated according to the second charge and discharge current value and the second entropy heat coefficient, and is obtained according to the following formula:
[0110]
[0111] Wherein, I is the second charge and discharge current value, E is the open circuit voltage, U is the battery terminal voltage, T is the Kelvin temperature of the second battery to be tested, is the second entropy thermal coefficient.
[0112] It can be understood that this embodiment calculates the second state of charge and the second heat generation corresponding to the second battery to be tested based on the second SOC-OCV curve, the second entropy thermal coefficient, the second charge and discharge current value and the second voltage value, which can provide an effective data basis for obtaining the target temperature prediction result, and this embodiment introduces heat generation as an input feature and training data of the prediction model, and the heat generation has a strong correlation with the target feature (i.e., the target battery temperature prediction result), so that the prediction result of the battery temperature prediction model is more accurate.
[0113] In addition, in some embodiments, the battery temperature prediction model includes an update gate and a reset gate, and the second state of charge and the second heat generation are input into the battery temperature prediction model to obtain a target temperature prediction result, which is obtained according to the following formula:
[0114] r t =σ(W r ·[h t-1 ,x t ]);
[0115] z t =σ(W r ·[h t-1 ,x t ]);
[0116]
[0117] h t =(1-z t )*h t-1 +z t *h t ;
[0118] y t =σ(W o ·h t +c);
[0119] Among them, x tis the second charge and discharge current value, the second voltage value, the second state of charge and the second heat generation value currently input, h t is the hidden state vector at time t, h t-1 is the hidden state vector at time t-1, σ is the first activation function in the hidden layer of the battery temperature prediction model, r t is the output of the reset gate at time t, z t is the output of the update gate at time t, W r is the weight matrix of the update gate, W o is the output weight matrix, c is the output bias, y t is the target temperature prediction result, is the candidate hidden state vector at time t, and tanh is the second activation function, which is used to map the input to the [-1,1] interval.
[0120] It can be understood that the battery temperature prediction model of this embodiment includes an update gate and a reset gate, and the input parameters, hidden states, and previous states are combined by the update gate and the reset gate to control the output information, thereby providing an effective model architecture basis for the output target temperature prediction result, and the battery temperature prediction model of this embodiment has the ability to capture long-term dependencies and can be applied to most sequence tasks. Compared with the traditional method, the present invention uses neural network technology to more accurately predict the battery temperature, thereby improving the reliability and safety of the system, and predicting the battery temperature through voltage and current data reduces the dependence on additional temperature sensors, reducing the complexity and cost of the system; the embodiment of the present application is not only suitable for battery temperature monitoring, but also can be extended to other fields that require real-time temperature monitoring, such as electric vehicles, power storage, etc. In addition, the battery temperature prediction method based on EEMD-GRU provided in the present application can be deployed in a battery management system (BMS), so that the current battery temperature can be known in real time, and the comparison with the normal working temperature range of the battery can timely warn the battery safety, and take corresponding safety measures to control and avoid thermal runaway.
[0121] refer to Figure 2 , Figure 2 : is a module schematic diagram of a battery temperature prediction device based on EEMD-GRU provided in another embodiment of the present application, the battery temperature prediction device includes:
[0122] The offline data acquisition module 210 is used to obtain the first SOC-OCV curve, the first entropy thermal coefficient, the first charge and discharge current value, the first voltage value, the first state of charge, the first heat generation and the battery temperature rise data corresponding to each first battery to be tested;
[0123] The temperature rise data decomposition module 220 is used to decompose the temperature rise data of each battery by a preset EEMD algorithm to obtain a target IMF component and a residual signal, and to superimpose all the target IMF components to obtain a recombined signal;
[0124] A model training module 230 is used to train a preset GRU neural network based on all the first SOC-OCV curves, the first entropy thermal coefficient, the first charge and discharge current values, the first voltage value, the first state of charge, the first heat generation, the battery temperature rise data, the residual signal and the recombined signal to obtain a battery temperature prediction model;
[0125] The online data acquisition module 240 is configured to obtain a second SOC-OCV curve, a second entropy thermal coefficient, and a second charge-discharge current value corresponding to the second battery to be tested, and calculate a second state of charge and a second heat generation value corresponding to the second battery to be tested according to the second SOC-OCV curve, the second entropy thermal coefficient, and the second charge-discharge current value, wherein the second battery to be tested is different from all the first batteries to be tested;
[0126] The temperature estimation module 250 is used to input the second state of charge and the second heat generation into the battery temperature prediction model to obtain a target temperature prediction result.
[0127] It should be noted that the specific implementation of the battery temperature prediction device 200 based on EEMD-GRU in this embodiment is basically the same as the specific implementation of the battery temperature prediction method based on EEMD-GRU mentioned above, and will not be repeated here.
[0128] like Figure 3 As shown, Figure 3 : is a structural diagram of a control device provided by an embodiment of the present application. The present invention also provides a control device 300, including:
[0129] The processor 310 may be implemented by a general-purpose central processing unit (CPU), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present application;
[0130] The memory 320 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 320 can store an operating system and other applications. When the technical solution provided in the embodiment of this specification is implemented by software or firmware, the relevant program code is stored in the memory 320, and the processor 310 calls and executes the battery temperature prediction method based on EEMD-GRU in the embodiment of this application;
[0131] Input / output interface 330, used to implement information input and output;
[0132] Communication interface 340, used to realize communication interaction between the apparatus and other devices, which can be realized by wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.);
[0133] bus 350 , which transmits information between the various components of the device (e.g., processor 310 , memory 320 , input / output interface 330 , and communication interface 340 );
[0134] The processor 310 , the memory 320 , the input / output interface 330 and the communication interface 340 are connected to each other in communication within the device via the bus 350 .
[0135] In addition, an embodiment of the present application further provides a storage medium, which is a computer-readable storage medium and stores a computer program. When the computer program is executed by a processor, the above-mentioned battery temperature prediction method based on EEMD-GRU is implemented.
[0136] As a non-transient computer-readable storage medium, the memory can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage devices. In some embodiments, the memory may optionally include a memory remotely arranged relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned networks include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof. The device embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and are implemented to be located in one place, or may also be distributed to multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the present embodiment.
[0137] It will be appreciated by those skilled in the art that all or some of the steps and systems in the disclosed method above may be implemented as software, firmware, hardware and appropriate combinations thereof. Some physical components or all physical components may be implemented as software executed by a processor, such as a central processing unit, a digital signal processor or a microprocessor, or may be implemented as hardware, or may be implemented as an integrated circuit, such as an application specific integrated circuit. Such software may be distributed on a computer-readable medium, which may include a computer storage medium (or a non-transitory medium) and a communication medium (or a temporary medium). As known to those skilled in the art, the term computer storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules or other data). Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tapes, disk storage or other magnetic storage devices, or any other medium that may be used to store desired information and may be accessed by a computer. Furthermore, it is well known to those skilled in the art that communication media typically include computer readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media.
[0138] The above is a specific description of the preferred implementation of the present invention, but the present invention is not limited to the above-mentioned implementation mode. Technical personnel familiar with the field can also make various equivalent deformations or substitutions under the shared conditions without violating the spirit of the present invention. These equivalent deformations or substitutions are all included in the scope defined by the claims of the present invention.
Claims
1. A battery temperature prediction method based on EEMD-GRU, It is characterized in that include: Obtaining a first SOC-OCV curve, a first entropy thermal coefficient, a first charge and discharge current value, a first voltage value, a first state of charge, a first heat generation, and battery temperature rise data corresponding to each first battery to be tested; Decomposing each of the battery temperature rise data by a preset EEMD algorithm to obtain a target IMF component and a residual signal, and superimposing all of the target IMF components to obtain a recombined signal; A preset GRU neural network is trained based on all of the first SOC-OCV curve, the first entropy thermal coefficient, the first charge and discharge current value, the first voltage value, the first state of charge, the first heat generation, the battery temperature rise data, the residual signal, and the recombined signal to obtain a battery temperature prediction model; Obtaining a second SOC-OCV curve, a second entropy thermal coefficient, a second charge and discharge current value, and a second voltage value corresponding to a second battery to be tested, and calculating a second state of charge and a second heat generation corresponding to the second battery to be tested according to the second SOC-OCV curve, the second entropy thermal coefficient, the second charge and discharge current value, and the second voltage value, wherein the second battery to be tested is different from all the first batteries to be tested; The second charge and discharge current value, the second voltage value, the second state of charge, and the second heat generation are input into the battery temperature prediction model to obtain a target temperature prediction result.
2. The battery temperature prediction method based on EEMD-GRU according to claim 1, It is characterized in that Decomposing each of the battery temperature rise data by a preset EEMD algorithm to obtain a target IMF component and a residual signal includes: Based on a preset number of times, Gaussian white noise is added to each of the battery temperature rise data for multiple times to obtain multiple intermediate signals, wherein the amount of noise added to each of the intermediate signals of each of the battery temperature rise data is different from each other; Performing EMD decomposition on each of the intermediate signals to obtain a corresponding initial IMF component; An average value of the initial IMF components corresponding to each of the intermediate signals is taken to obtain the target IMF components and the residual signal corresponding to each of the battery temperature rise data.
3. The battery temperature prediction method based on EEMD-GRU according to claim 2, It is characterized in that Perform EMD decomposition on each of the intermediate signals to obtain corresponding initial IMF components, including: determining local maxima and local minima in each of the intermediate signals; Connecting all the local maxima by an interpolation algorithm to obtain an upper envelope; Connecting all the local minimum values by the interpolation algorithm to obtain a lower envelope; Calculating the average value of the upper envelope and the lower envelope to obtain the trend component of each of the intermediate signals; The target battery temperature rise data is subtracted from the trend component to obtain the initial IMF component, wherein the target battery temperature rise data is the original battery temperature rise data corresponding to the intermediate signal.
4. The battery temperature prediction method based on EEMD-GRU according to claim 1, It is characterized in that The GRU neural network includes a first sub-GRU network and a second sub-GRU network, and the battery temperature prediction model includes a first target sub-GRU network and a second target sub-GRU network. The preset GRU neural network is trained based on all of the first SOC-OCV curves, the first entropy thermal coefficient, the first charge and discharge current value, the first voltage value, the first state of charge, the first heat generation, the battery temperature rise data, the residual signal, and the recombined signal to obtain a battery temperature prediction model, including: Based on a preset data set division ratio, all of the first SOC-OCV curve, the first entropy thermal coefficient, the first charge and discharge current value, the first voltage value, the first state of charge, the first heat generation, the battery temperature rise data, the residual signal, and the recombined signal are divided into a training set, a test set, and a validation set; Acquire the first charge-discharge current value, the first voltage value, the first state of charge, and the first heat generation from the training set, and input the first charge-discharge current value, the first voltage value, the first state of charge, and the first heat generation into the first sub-GRU network to obtain a first output result; Inputting the first charge-discharge current value, the first voltage value, the first state of charge, and the first heat generation into the second sub-GRU network to obtain a second output result; Comparing the residual signal with the first output result to obtain a first comparison result; Comparing the recombinant signal with the second output result to obtain a second comparison result; Calculate a first loss value based on a first loss function and the first comparison result, and calculate a second loss value based on a second loss function and the second comparison result; Training the first sub-GRU network based on the first loss value and the first loss function to obtain a first intermediate network; Training the second sub-GRU network based on the second loss value and the second loss function to obtain a second intermediate network; Acquire third sample data from the validation set and the test set, and adjust the first intermediate network based on the third sample data to obtain the first target sub-GRU network, wherein the third sample data has the same data type as the first sample data; Acquire fourth sample data from the validation set and the test set, adjust the second intermediate network based on the fourth sample data, and obtain the second target sub-GRU network, wherein the fourth sample data has the same data type as the second sample data.
5. The battery temperature prediction method based on EEMD-GRU according to claim 1, It is characterized in that Calculating a second state of charge and a second heat generation corresponding to the second battery to be tested according to the second SOC-OCV curve, the second entropy thermal coefficient, the second charge and discharge current value, and the second voltage value, including: Determining an initial state of charge of the second battery to be tested according to the second SOC-OCV curve; Calculating the second state of charge according to the second charge-discharge current value and the initial state of charge based on the ampere-hour integration method; Based on the Bernadi heat generation equation, the second heat generation amount is calculated according to the second charge and discharge current value, the second entropy thermal coefficient and the second voltage value.
6. The battery temperature prediction method based on EEMD-GRU according to claim 5, It is characterized in that The second state of charge is calculated based on the ampere-hour integration method according to the second charge-discharge current value and the initial state of charge, and is obtained according to the following formula: Among them, SOC 0 is the initial state of charge, I is the second charge and discharge current value, t is time, C is the total capacity of the battery, and SOC is the second state of charge; The second voltage value includes the battery terminal voltage and the open circuit voltage corresponding to the second battery to be tested. The second heat generation is calculated based on the Bernadi heat generation equation according to the second charge and discharge current value, the second entropy thermal coefficient and the second voltage value, and is obtained according to the following formula: Wherein, I is the second charge and discharge current value, E is the open circuit voltage, U is the battery terminal voltage, and T is the Kelvin temperature of the second battery to be tested. is the second entropy thermal coefficient.
7. The battery temperature prediction method based on EEMD-GRU according to claim 1, It is characterized in that The battery temperature prediction model includes an update gate and a reset gate. The second charge and discharge current value, the second voltage value, the second state of charge, and the second heat generation are input into the battery temperature prediction model to obtain a target temperature prediction result, which is obtained according to the following formula: r t =σ(W r ·[h t-1 ,x t ]); z t =σ(W r ·[h t-1 ,x t ]); h t =(1-z t )*h t-1 +z t *h t ; y t =σ(W o ·h t +c); Among them, x t is the second charge and discharge current value, the second voltage value, the second state of charge and the second heat generation currently input, h t is the hidden state vector at time t, h t-1 is the hidden state vector at time t-1, σ is the first activation function in the hidden layer of the battery temperature prediction model, r t is the output of the reset gate at time t, z t is the output of the update gate at time t, W r is the weight matrix of the update gate, W o is the output weight matrix, c is the output bias, y t is the target temperature prediction result, is the candidate hidden state vector at time t, and tanh is the second activation function.
8. A battery temperature prediction device based on EEMD-GRU, It is characterized in that include: An offline data acquisition module, used to obtain a first SOC-OCV curve, a first entropy thermal coefficient, a first charge and discharge current value, a first voltage value, a first state of charge, a first heat generation, and battery temperature rise data corresponding to each first battery to be tested; A temperature rise data decomposition module, used to decompose each of the battery temperature rise data by a preset EEMD algorithm to obtain a target IMF component and a residual signal, and to superimpose all of the target IMF components to obtain a recombined signal; A model training module, used for training a preset GRU neural network based on all of the first SOC-OCV curve, the first entropy thermal coefficient, the first charge and discharge current value, the first voltage value, the first state of charge, the first heat generation, the battery temperature rise data, the residual signal and the recombined signal to obtain a battery temperature prediction model; an online data acquisition module, for obtaining a second SOC-OCV curve, a second entropy thermal coefficient, a second charge and discharge current value, and a second voltage value corresponding to a second battery to be tested, and calculating a second state of charge and a second heat generation corresponding to the second battery to be tested according to the second SOC-OCV curve, the second entropy thermal coefficient, the second charge and discharge current value, and the second voltage value, wherein the second battery to be tested is different from all the first batteries to be tested; The temperature estimation module is used to input the second charge and discharge current value, the second voltage value, the second state of charge and the second heat generation into the battery temperature prediction model to obtain a target temperature prediction result.
9. A control device, It is characterized in that The invention comprises at least one control processor and a memory for communicating with the at least one control processor; the memory stores instructions executable by the at least one control processor, and the instructions are executed by the at least one control processor so that the at least one control processor can execute the battery temperature prediction method based on EEMD-GRU as described in any one of claims 1 to 7.
10. A computer-readable storage medium, It is characterized in that The computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to enable a computer to execute the battery temperature prediction method based on EEMD-GRU as described in any one of claims 1 to 7.
