CN115358484A - Battery pack capacity prediction method and related equipment - Google Patents
Battery pack capacity prediction method and related equipment Download PDFInfo
- Publication number
- CN115358484A CN115358484A CN202211113456.XA CN202211113456A CN115358484A CN 115358484 A CN115358484 A CN 115358484A CN 202211113456 A CN202211113456 A CN 202211113456A CN 115358484 A CN115358484 A CN 115358484A
- Authority
- CN
- China
- Prior art keywords
- data
- capacity prediction
- capacity
- temperature
- prediction model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
- Y02E60/00—Enabling technologies; Technologies with a potential or indirect contribution to GHG emissions mitigation
- Y02E60/10—Energy storage using batteries
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Computational Linguistics (AREA)
- Development Economics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Game Theory and Decision Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Entrepreneurship & Innovation (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application provides a battery pack capacity prediction method and related equipment, wherein the method comprises the following steps: acquiring operation data of a battery pack, operation temperature corresponding to the operation data and a preset calibration temperature; carrying out temperature calibration on the operation data according to the operation temperature and the calibration temperature to obtain target data; and inputting the target data into the trained capacity prediction model to obtain a target capacity prediction value of the battery pack. Before capacity prediction is carried out, temperature calibration is carried out on operation data to obtain target data. And then, the capacity is predicted by combining the target data and the trained capacity prediction model, so that the prediction precision of the capacity of the battery pack is improved to a certain extent.
Description
Technical Field
The present application relates to the field of battery technologies, and in particular, to a battery pack capacity prediction method and related devices.
Background
The Battery Pack (Battery Pack) is widely applied to various devices, and if the capacity of the Battery Pack can be accurately predicted, the Battery Pack has important significance and value for ensuring the healthy use of the Battery Pack and prolonging the service life of the Battery Pack. At present, a prediction method of battery pack capacity mainly calculates through a model, however, the current model usually does not consider the influence of the environment on the battery pack capacity degradation, so that accurate modeling cannot be realized, and the prediction precision of the battery pack capacity is inevitably influenced.
Disclosure of Invention
The embodiment of the application discloses a battery pack capacity prediction method and related equipment, and solves the problem of low battery pack capacity prediction precision.
The application provides a battery pack capacity prediction method, which comprises the following steps:
acquiring operation data of a battery pack, operation temperature corresponding to the operation data and a preset calibration temperature;
carrying out temperature calibration on the operation data according to the operation temperature and the calibration temperature to obtain target data;
and inputting the target data into a trained capacity prediction model to obtain a target capacity prediction value of the battery pack.
In some optional embodiments, the performing the temperature calibration on the operation data according to the operation temperature and the calibration temperature to obtain target data includes:
acquiring a preset temperature calibration model;
and carrying out temperature calibration on the operation data according to the temperature calibration model, the calibration temperature and the operation temperature to obtain the target data.
In some optional embodiments, before the obtaining the preset temperature calibration model, the method further comprises:
acquiring historical capacity data and historical temperature data corresponding to the historical capacity data;
and determining a temperature calibration model according to the historical capacity data and the historical temperature data.
In some optional embodiments, before the inputting the target data into the capacity prediction model to obtain the target capacity prediction value, the method further includes:
calculating an outlier corresponding to each data in the target data, and filtering out data corresponding to the outliers larger than a first threshold value to obtain filtering data;
performing dimensionality reduction calculation on the filtering data to obtain dimensionality reduction data;
correspondingly, the inputting the target data into the trained capacity prediction model to obtain the target capacity prediction value includes:
and inputting the dimensionality reduction data into the trained capacity prediction model to obtain a target capacity prediction value.
In some optional embodiments, before the obtaining operational data of the battery pack, the method comprises:
optimizing neurons of a hidden layer of the initial capacity prediction model by using a first optimization algorithm to obtain an optimized capacity prediction model;
acquiring training data and a sample label corresponding to the training data;
inputting the training data into the optimized capacity prediction model to obtain a sample capacity prediction value;
calculating a loss function value of the optimized capacity prediction model according to the sample capacity prediction value and the sample label;
if the loss function value is larger than a second threshold value, taking the loss function value as the input of the optimized capacity prediction model for back propagation, optimizing the weight and the bias of the neuron corresponding to the back propagation by using a second optimization algorithm, updating the weight and the bias of the optimized capacity prediction model, and returning to execute the step of inputting the training data into the optimized capacity prediction model;
and if the loss function value is smaller than or equal to a second threshold value, determining the current optimized capacity prediction model as a trained capacity prediction model.
In some optional embodiments, the optimizing the neuron number of the hidden layer of the initial volume prediction model by using the first optimization algorithm includes:
determining the state of each neuron in a hidden layer of the initial capacity prediction model based on a preset state probability value;
and removing the neurons with the state as the first state to obtain the optimized capacity prediction model.
In some optional embodiments, after the determining the current optimized capacity prediction model as the trained capacity prediction model, the method further comprises:
acquiring test data and a test label corresponding to the test data;
inputting the test data into the trained capacity prediction model to obtain a test capacity prediction value;
calculating an error value and a fitting degree of the trained capacity prediction model according to the test label and the test capacity prediction value;
and if the error value is larger than a third threshold value or the fitting degree is smaller than a fourth threshold value, returning to the step of acquiring the training data and the sample label corresponding to the training data.
The present application further provides a battery pack capacity prediction apparatus, the apparatus includes an acquisition module, a temperature calibration module, and a prediction module:
the acquisition module is used for acquiring the operation data of the battery pack, the operation temperature corresponding to the operation data and a preset calibration temperature;
the temperature calibration module is used for carrying out temperature calibration on the operation data according to the operation temperature and the calibration temperature to obtain target data;
and the prediction module is used for inputting the target data into the trained capacity prediction model to obtain a target capacity prediction value of the battery pack.
The present application further provides an electronic device comprising a processor and a memory, the processor being configured to implement the battery pack capacity prediction method when executing a computer program stored in the memory.
The present application also provides a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements the battery pack capacity prediction method.
In the battery pack capacity prediction method provided by the application, considering that the temperature can generate large influence on the capacity of the battery pack, before capacity prediction is carried out, temperature calibration can be carried out on operation data according to the operation temperature corresponding to the operation data and a preset calibration temperature, and the influence caused by temperature difference is reduced in a temperature calibration mode to obtain target data. And then, capacity prediction is carried out by combining the target data and the trained capacity prediction model, and the influence of temperature on a target capacity prediction value is reduced, so that the prediction precision of the capacity of the battery pack is improved to a certain extent.
Drawings
Fig. 1 is a schematic view of an application scenario of a battery pack capacity prediction method provided in an embodiment of the present application.
Fig. 2 is a flowchart of a battery pack capacity prediction method according to an embodiment of the present application.
Fig. 3 is a graph of a temperature calibration model provided in an embodiment of the present application.
Fig. 4 is a flowchart of training a capacity prediction model according to an embodiment of the present application.
Fig. 5 is a schematic structural diagram of a neural network provided in an embodiment of the present application.
Fig. 6 is a flowchart of an expansion process performed on target data according to an embodiment of the present application.
Fig. 7 is a block diagram of a battery pack capacity prediction apparatus according to an embodiment of the present application.
Detailed Description
For ease of understanding, some descriptions of concepts related to the embodiments of the present application are given by way of illustration and reference.
In the present application, "at least one" means one or more, "and" a plurality "means two or more. "and/or" describes an association relationship of associated objects, meaning that three relationships may exist, e.g., A and/or B may represent: a exists alone, A and B exist simultaneously, and B exists alone, wherein A and B can be singular or plural. The terms "first," "second," "third," "fourth," and the like in the description and in the claims and drawings of the present application, if any, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order.
The battery pack capacity prediction method provided by the embodiment of the application can be applied to electronic equipment with the battery pack. The electronic device may include any one of a mobile phone, a tablet computer, a notebook computer, an intelligent voice interaction device, an intelligent home device, an in-vehicle device, an Augmented Reality (AR)/Virtual Reality (VR) device, an ultra-mobile personal computer (UMPC), a netbook, a Personal Digital Assistant (PDA), a home energy storage device, a mobile energy storage device, and a self-mobile device.
In order to better understand the battery pack capacity prediction method and the related device provided in the embodiment of the present application, first, an application scenario of the battery pack capacity prediction method of the present application is described below.
Fig. 1 is a schematic view of an application scenario of a battery pack capacity prediction method provided in an embodiment of the present application. The method for predicting the capacity of the battery pack provided by the embodiment of the application is applied to the electronic device 1, and the electronic device 1 includes, but is not limited to, a memory 12, at least one processor 13 and a data acquisition unit 14 which are communicatively connected with each other through a communication bus 11. The data acquisition unit 14 may be a data acquisition device with a data acquisition function, and is configured to acquire operation data of the battery pack, and store the acquired operation data in the memory 12, so that the processor 13 performs corresponding data processing.
Fig. 1 is only an example of the electronic device 1, and does not constitute a limitation to the electronic device 1, and in practical applications, the electronic device 1 may include more or less components than those shown in the drawings, or some components may be combined, or different components may be replaced, for example, the electronic device 1 may further include an input and output device, a network access device, and the like.
In order to solve the technical problem that the prediction accuracy of the battery pack capacity is low due to environmental influences, the embodiment of the application provides a battery pack capacity prediction method, which can perform temperature calibration on operation data of a battery pack, and combines consideration of temperature factors, so that the prediction accuracy of the battery pack capacity is guaranteed. As shown in fig. 2, fig. 2 is a flowchart illustrating a method for predicting the capacity of a battery pack according to an embodiment of the present application.
The battery pack capacity prediction method is applied to electronic equipment (such as the electronic equipment 1 in fig. 1). The order of the steps in the flow chart may be changed and some steps may be omitted according to different needs.
And step 21, acquiring the operation data of the battery pack, the operation temperature corresponding to the operation data and a preset calibration temperature.
In the embodiment of the application, the battery pack is charged and discharged within a preset time period, and data of the battery pack during the charging and discharging period can be directly acquired as operation data. For example: the operation data may include one or more of parameters of an operation temperature, a charging voltage, a State of Charge (SOC) during charging, and a charging time of the battery pack. Here, when the SOC =1, it indicates that the battery is fully charged, and when the SOC =0, it indicates that the battery is exhausted.
And/or the operational data may be indirectly calculated data. For example: the voltage increment of the single battery, the charging capacity increment corresponding to the voltage increment of the single battery, the SOC increment of the battery pack, the charging capacity increment corresponding to the SOC increment of the battery pack, the charging time increment, the SOC variation of single charging of a user and the like can be obtained by calculating the directly acquired charging voltage, the charging state in the charging process and the charging time.
The operation temperature corresponding to the operation data is the temperature of the battery pack when the operation data is acquired. The calibration temperature is a preset reference temperature, and the calibration temperature can be set according to the suitable use temperature of the battery pack, for example, the calibration temperature can be 20 ℃, 25 ℃, 30 ℃ or other temperature values.
And step 22, carrying out temperature calibration on the operation data according to the operation temperature and the calibration temperature to obtain target data.
In the embodiment of the present application, a mathematical model may be trained based on historical data to obtain a temperature calibration model capable of achieving temperature tuning. Specifically, historical capacity data and temperature data corresponding to the historical capacity data are acquired, for example, the historical capacity data may include one or more of historical charging voltage and state of charge during historical charging.
The determining a temperature calibration model according to the historical capacity data and the historical temperature data specifically includes: and calculating the historical capacity data and the temperature data corresponding to the historical capacity data to obtain a capacity retention rate, and generating a fitting formula of the capacity retention rate and the temperature data, namely a temperature calibration model, according to the relation between the capacity retention rate and the temperature data.
For example, in some examples, historical capacity data of the same battery pack at different temperatures may be obtained, and certain temperature data may be determined as a calibrated temperature (e.g., 25 ℃), and historical capacity data corresponding to the calibrated temperature may be determined as reference capacity data. Then, the historical capacity data corresponding to the different temperature data can be divided by the reference capacity data to obtain the capacity retention rates corresponding to the different temperature data. And then, performing curve fitting according to the capacity retention rates corresponding to different temperature data to obtain a fitting formula of the capacity retention rates and the temperature data, namely a temperature calibration model.
The expression form of the temperature calibration model can be any one of function types such as a linear function, a multiple function, an exponential function, a power function and the like, and the specific expression form can be set according to actual requirements. For example, in some examples, the expression for the temperature calibration model may be:
in the above equation, Q (i) represents the i-th operation data before temperature calibration, Q' (i) represents the i-th operation data after temperature calibration, and T i Representing the operating temperature value corresponding to the ith operating data.
This is illustrated in connection with the graph of the temperature calibration model shown in fig. 3. The graph shown in fig. 3 reflects the relationship between the historical temperature and the capacity retention rate, and a fitting formula of the capacity retention rate and the temperature data can be obtained from the graph.
According to the determined temperature calibration model, the calibration temperature and the operation temperature corresponding to the operation data, the temperature calibration can be performed on the parameters (such as charging capacity increment) related to the capacity in the operation data, and the operation data at different operation temperatures are mapped to the calibration temperature through the temperature calibration model to obtain target data.
In the embodiment of the application, the influence of the temperature on the capacity prediction can be reduced to a certain extent by carrying out temperature calibration on the operation data.
And 23, inputting the target data into the trained capacity prediction model to obtain a target capacity prediction value of the battery pack.
After the target data is obtained, the target data may be input into the trained capacity prediction model to obtain a target capacity prediction value output by the capacity prediction model.
The target capacity prediction value may be understood as a capacity prediction value of the battery pack at a calibration temperature.
The specific type of the capacity prediction model can be set according to actual requirements. For example, the capacity prediction model may include any one or a combination of Long Short-Term Memory (LSTM) Network, recurrent Neural Network (RNN), convolutional Neural Network (CNN), and other types of models.
After the initial capacity prediction model is acquired, the initial capacity prediction model needs to be trained so that the capacity prediction model can accurately predict the capacity of the battery pack.
When training the initial capacity prediction model, a large number of network parameters are usually required to be trained. If the structure of the neural network is too complex and the amount of the training sample data is small, an overfitting situation is easily caused in the training process. Therefore, in the embodiment of the present application, in order to avoid overfitting during the training process while ensuring the training efficiency of the model, the initial capacity prediction model may be trained based on an optimization algorithm. For a detailed training process of the capacity prediction model, reference may be made to the following detailed description of the steps of the flow chart shown in fig. 4.
Fig. 4 is a flowchart of training a capacity prediction model according to an embodiment of the present application, and as shown in fig. 4, the training of the capacity prediction model includes the following steps:
and 41, optimizing the neurons of the hidden layer of the initial capacity prediction model by using a first optimization algorithm to obtain an optimized capacity prediction model.
And optimizing the neurons of the hidden layer of the initial capacity prediction model by using a first optimization algorithm to obtain an optimized capacity prediction model. The first optimization algorithm may be a Dropout algorithm, which means that during the neural network training process, neurons of a partial hidden layer are randomly removed, and all inputs and outputs of corresponding neurons are removed at the same time. During each training, a fixed proportion of neurons were randomly removed among all neurons.
Optimizing the neurons of the hidden layer of the initial capacity prediction model by using a first optimization algorithm, which specifically comprises the following steps: and determining the state of each neuron in the hidden layer of the initial capacity prediction model based on a preset state probability value, removing the neuron with the state being the first state, and obtaining a neural network structure corresponding to the optimized capacity prediction model.
And determining the states of all neurons in the hidden layer of the initial capacity prediction model according to the preset state probability value, wherein the states comprise the first state of the neuron of the hidden layer smaller than the preset state probability value and other states of the neuron of the hidden layer larger than the preset state probability value. In the process of each training, removing the neuron with the state of the neuron in the hidden layer as the first state, and taking the neurons in other states as the neurons to be trained of the optimized capacity prediction model.
Fig. 5 is a schematic diagram of a neural network structure provided in an embodiment of the present application. As shown in fig. 5, (a) the scenario in fig. 5 shows the neural network structure before neuron optimization for the hidden layer, and (b) the scenario in fig. 5 shows the neural network structure after neuron optimization for the hidden layer.
In a specific embodiment, assume that the initial capacity prediction model has n neurons (i.e., the solid line o symbol in fig. 5), n being a natural number greater than or equal to 1. Each neuron is provided with a random state value, and the value range of the state value is in the range of [0,1 ]. Comparing the state value of the neuron of the hidden layer with the preset state probability value P according to the preset state probability value P, if the state value of the neuron of the hidden layer is smaller than the preset state probability value P, removing the neuron smaller than the preset state probability value P (namely, a dotted line O symbol containing x symbols), and keeping the neuron larger than the preset state probability value P.
In another specific embodiment, assume that the initial volume prediction model has n neurons. Each neuron can determine the state of the neuron according to the random state probability value P. For example, assuming a value of 0.4 for P, each neuron has a 40% probability of being in the first state and a 60% probability of being in the other state. After the state of each neuron is determined, the neuron corresponding to the first state may be removed, leaving the other neurons.
In each training process, the neurons reserved in the neurons in the hidden layer can be obtained according to the preset state probability value P, each training is equivalent to extracting a sub-network from the original network, and the neurons to be trained in the sub-network are the reserved neurons. The generalization capability of the optimized capacity prediction model can be effectively improved by utilizing the first optimization algorithm, and overfitting is avoided.
And 42, acquiring training data and sample labels corresponding to the training data.
The training data and the sample labels corresponding to the training data can be obtained from historical operating data of the battery pack, and the sample labels corresponding to the training data are sample label values of the training data, namely real capacity values corresponding to the training data.
In addition, in some embodiments, in order to reduce the influence of temperature on the model training process, the training data and the sample labels may be subjected to temperature correction after the training data and the sample labels are acquired, and model training may be performed using the training data and the sample labels after the temperature correction.
And 43, inputting the training data into the optimized capacity prediction model to obtain a sample capacity prediction value.
Inputting training data into the optimized capacity prediction model, executing a forward propagation process, and calculating a sample capacity prediction value, wherein a specific calculation formula is as follows:
in the formula (I), the compound is shown in the specification,
and f represents a sample capacity predicted value, and f represents one forward propagation of the optimized capacity prediction model.
And step 44, calculating a loss function value of the optimized capacity prediction model according to the sample capacity prediction value and the sample label.
Based on the sample label corresponding to the training data, the training effect of the optimized capacity prediction model is judged, and the method specifically comprises the following steps: inputting training data into the optimized capacity prediction model, and calculating a sample capacity prediction value output by the optimized capacity prediction model in the forward propagation process
Further calculating the difference value between the sample capacity predicted value and the sample label to obtain a loss function value of the optimized capacity prediction model, wherein the value of the calculated loss function is represented by the following specific formula:
in the formula, N represents the length of training data (or the number of training data), t represents the number of training data, and y represents the number of training data t Representing the sample label value corresponding to the tth training data,
and the predicted value of the sample capacity corresponding to the t-th training data is shown.
And judging whether the optimized capacity prediction model is trained according to the loss function value of the sample capacity prediction value.
And step 45, if the loss function value is larger than the second threshold value, taking the loss function value as the input of the optimized capacity prediction model for back propagation, optimizing the weight and the bias of the neuron corresponding to the back propagation by using a second optimization algorithm, updating the weight and the bias of the optimized capacity prediction model, and returning to the step 43 to input the training data into the optimized capacity prediction model.
Specifically, when it is determined that the loss function value is greater than the second threshold, it may be determined that the optimized capacity prediction model does not reach a convergence state, a back propagation process is continuously performed, the loss function value is used as an input for performing back propagation, and a second optimization algorithm is used to optimize the weight and bias of a neuron corresponding to the back propagation current training, where the second optimization algorithm may be an Adam algorithm, and the Adam algorithm is a gradient optimization algorithm.
And optimizing the weight and the bias of the neuron corresponding to the current back propagation training by using a second optimization algorithm, which specifically comprises the following steps: in the back propagation process, the gradient at the t step is calculated, and the specific formula is as follows:
in the formula, theta represents a network parameter to be optimized in the capacity prediction model, J (theta) represents an objective function of theta,
denotes the gradient of θ, g t Denotes theta t-1 Gradient of (a), theta t-1 The network parameters of step (t-1) are indicated.
Accordingly, the exponential moving average m of the gradient at step t is calculated t And the exponential moving average v of the square of the gradient t The concrete formula is as follows:
in the formula, beta 1 And beta 2 Is an exponential decay factor, beta 1 For controlling the weight distribution, beta 2 For controlling the influence of the gradient squared.
Due to m 0 And v 0 Will be initialized to 0 during the calculation process, which will make m at the beginning of the phase t And v t All of them are biased to 0, so that it is necessary to perform deviation correction to reduce the influence of deviation on the initial training stage, and the deviation correction formula is as follows:
in the formula (I), the compound is shown in the specification,
and
is m t And v t Modified value after correction, finally utilized
And
updating the network parameters to be optimized, wherein the concrete formula is as follows:
in the formula, theta t And (4) the network parameter of the t step, alpha is the initial learning rate, and epsilon is a smoothing index.
The values of the relevant parameters in the formulas can be set according to actual requirements, for example, in an example, the values of the parameters can be as follows: beta is a 1 ＝0.9，β 2 ＝0.999，α＝0.001，ε＝10 -8 。
The weights and biases of the neurons corresponding to the current training can be updated according to the Adam algorithm described above. After the weights and biases have been updated, the process can return to step 43 to continue the forward propagation process until the optimized capacity prediction model reaches a converged state.
And step 46, if the loss function value is less than or equal to the second threshold value, determining the current optimized capacity prediction model as a trained capacity prediction model.
Upon determining that the loss function value is less than or equal to the second threshold value, it may be determined that the optimized capacity prediction model has reached a converged state. At this time, the current optimized capacity prediction model may be determined as a trained capacity prediction model.
In addition, in some embodiments, when it is determined that the loss function value is less than or equal to the second threshold, it may be further determined whether the optimized capacity prediction model reaches a preset number of training times. One forward propagation process and one backward propagation process will be performed as one training time. If the number of current training times is less than the preset number of training times, optimizing the weight and bias of the neuron corresponding to the back propagation by using a second optimization algorithm, updating the weight and bias of the optimized capacity prediction model, and returning to step 43 to input training data into the optimized capacity prediction model; and if the current training times are equal to the preset training times, judging that the model training is finished, and determining the current optimized capacity prediction model as a trained capacity prediction model.
In the embodiment, according to the optimization of the first optimization algorithm and the second optimization algorithm, the generalization capability of the trained capacity prediction model is effectively improved, and the prediction efficiency and the prediction precision of the capacity prediction model are improved.
In some embodiments, after determining the current optimized capacity prediction model as the trained capacity prediction model, the testing of the trained capacity prediction model may also verify the predictive effect of the trained capacity prediction model.
And obtaining the test data and the test label corresponding to the test data from the historical operating data, wherein the test label corresponding to the test data is the test label value of the test data, namely the real capacity value corresponding to the test data. Inputting the test data into the trained capacity prediction model to obtain a test capacity prediction value, calculating an error value and a fitting degree of the trained capacity prediction model according to the test label and the test capacity prediction value, and returning to step 42 to obtain the training data and a sample label corresponding to the training data if the error value is greater than a third threshold value or the fitting degree is less than a fourth threshold value.
Calculating the error value and the fitting degree of the trained capacity prediction model according to the test label and the test capacity prediction value, and specifically comprising the following steps: the prediction effect of the trained capacity prediction model can be evaluated by calculating the mean square error and the mean absolute error according to the test label and the test capacity prediction value, and the prediction effect of the trained capacity prediction model can be evaluated by calculating the fitting degree according to the test label and the test capacity prediction value, wherein the specific calculation formula is as follows:
where RMSE represents the root mean square error, MAE represents the mean absolute error, R 2 Representing the degree of fit, m represents the test dataK denotes the serial number of the test data, y k Indicating the test tag value corresponding to the kth test data,
represents a predicted value of the capacity of the kth test data,
represents the mean volume of the test data.
If the mean square error and the average absolute error are larger than the third threshold or the fitting degree is smaller than the fourth threshold, the trained capacity prediction model does not reach the expected prediction accuracy, the capacity prediction model needs to be trained again, the training data and the sample labels corresponding to the training data are obtained again, and the parameters of the model are adjusted again to improve the prediction accuracy of the capacity prediction model.
In practical application, the root mean square error and the average absolute value error need to be smaller than a third threshold, the third threshold can be set according to practical requirements, and the error values of different types can correspond to the same or different third thresholds. For example, the third threshold for root mean square error may be 4% and the third threshold for average absolute error may be 2%.
In an exemplary scenario, the performance of the trained capacity prediction model may be randomly selected from 20 ternary lithium batteries with different aging degrees. The rated capacity of each of the two selected battery packs is 24Ah, the real capacity of each of the two selected battery packs is 22.18Ah and 22.92Ah, the operation data of the two battery packs are acquired through the data acquisition unit respectively, the data acquisition unit acquires the operation data of the two battery packs at the acquisition frequency of 10HZ, the increment of 3.8V-4.1V, namely 0.3V, can be selected as the voltage increment of a single battery, and the increment of 50% -80%, namely 30% of SOC increment of the battery packs can be selected as the SOC increment of the battery packs.
And inputting the operation data into the trained capacity prediction model to obtain capacity prediction values corresponding to the two battery packs. And calculating the mean square error, the average absolute error and the fitting degree corresponding to the two battery packs according to the initial capacities of the two battery packs and the predicted capacity values of the two battery packs to obtain the results shown in the tables 1 and 2.
Tables 1 and 2 are evaluations of the 22.18Ah battery and 22.92Ah battery according to the root mean square error, the average absolute value error, and the degree of fit, table 1 is the result of predicted estimation of the capacity of the 22.18Ah battery, and table 2 is the result of predicted estimation of the capacity of the 22.92Ah battery.
Tables 1-22.18Ah Battery pack Capacity prediction evaluation results
| RMSE(％) | MAE(％) | R 2 |
| 2.32％ | 1.57 | 0.96 |
Table 2-22.92Ah battery pack capacity prediction evaluation results
| RMSE(％) | MAE(％) | R 2 |
| 1.95 | 1.32 | 0.97 |
The evaluation results show the effectiveness of the capacity prediction model provided by the embodiment of the application.
In the embodiment of the application, the accuracy of capacity prediction of the capacity prediction model is guaranteed through training and testing (including verification of training effect) of the capacity prediction model, and in the training process of the capacity prediction model, the first optimization algorithm and the second optimization algorithm are combined, so that overfitting of the capacity prediction model is avoided, and meanwhile, the calculation efficiency of the capacity prediction model is optimized.
In a specific embodiment, the obtaining of the actual capacity value includes: under the condition that the charging current and the state of charge in the charging process can be accurately acquired, the maximum available capacity of the battery pack at the current moment is calculated to serve as the real capacity, and the model is trained or the accuracy of the model is evaluated according to the real capacity.
The method for calculating the maximum available capacity of the battery pack at the current moment specifically comprises the following steps: acquiring the charging current of any time period from the operation data, calculating the actual fragment charging capacity of the battery pack by using an ampere-hour integration method, and then taking the fragment charging capacity and the charged state variation of the time period as a ratio to obtain the maximum available capacity of the battery pack at the current moment, wherein the specific calculation formula is as follows:
in the formula, ca part Indicating partial charge capacity, I (t) indicating charge current at time, t SOC＝20 And t SOC＝90 Respectively representing the start time and end time of the charging segment, ca m Represents the maximum available capacity of the battery pack at the present time, and Δ SOC represents the variation of the SOC of the charging section.
In the embodiment of the present application, the target data of step 22 may be further processed, and specifically, refer to the flowchart shown in fig. 6. Fig. 6 is a flowchart of an expansion process performed on target data according to an embodiment of the present application, and as shown in fig. 6, the target data obtained in step 22 is further processed as follows:
and 61, calculating an outlier corresponding to each data in the target data, and filtering out data corresponding to the outliers larger than a first threshold value to obtain filtered data.
The target data is data calibrated by a preset temperature calibration model, partial invalid data possibly exists in the target data and can influence a capacity prediction result, and the invalid data is data obviously abnormal.
Therefore, in order to reduce the influence of invalid data on the capacity prediction result, a Local Outlier Factor (LOF) algorithm may be used to process the target data and filter out data in which an anomaly obviously exists.
Calculating an outlier corresponding to each data in the target data, comparing the outlier with a first threshold value, taking the data corresponding to the outlier larger than the first threshold value as invalid data, and filtering the invalid data to obtain filtered data. The formula for calculating the outlier is specifically:
in the formula, LOF k (p) represents the outlier of point p, N k (p) represents a k-distance neighborhood of point p, point o being a point in the neighborhood, lrd k (p)、lrd k (o) indicates the local achievable densities of point p and point o, respectively.
And step 62, performing dimensionality reduction calculation on the filtering data to obtain dimensionality reduction data.
The high-dimensional data samples have sparsity, so that the data characteristics of the model are difficult to find, and in order to train the model better, the high-dimensional data needs to be subjected to dimensionality reduction by adopting a dimensionality reduction technology, wherein the dimensionality reduction technology can be principal component analysis, linear discriminant analysis and the like.
In the embodiment of the application, in order to avoid sparsity of the filtering data, the filtering data is subjected to dimensionality reduction processing to obtain the processed dimensionality reduction data.
First, the filtered data is centralized, assuming that the filtered data is an n-dimensional data set D = (x) (1) ,x (2) ,...,x (m) ) The data set D includes a plurality of filter data, and all the filter data in the data set D are centralized, and the specific formula is as follows:
in the formula, m represents the total number of the filtered data, and i represents the label of the current filtered data.
Secondly, after centralizing all the filtering data, calculating the covariance matrix XX of the filtering data T Performing characteristic value decomposition on the covariance matrix to obtain characteristic values which are sorted from large to small, and selecting characteristic vectors corresponding to the first n' characteristic values
And standardizing the selected eigenvectors to construct an eigenvector matrix theta.
Finally, the data x is filtered for each of the data sets D (i) Respectively projecting the data to the constructed eigenvector matrix theta, and converting each filtered data into z (i) ＝θ T x (i) Obtaining a dimension-reduced data set D' = (z) (1) ,z (2) ,...,z (m) ) I.e. dimension reduction data.
In order to improve the prediction accuracy of the model, in the embodiment of the present application, the dimension reduction data may be input into the trained capacity prediction model to obtain the target capacity prediction value, and specifically includes: data set D 'corresponding to dimension reduction data'＝(z (1) ,z (2) ,...,z (m) ) Initial feature vector X = { X) converted to input trained volume prediction model 1 ,x 2 ,…,x m And after the initial characteristic vector is input into the trained capacity prediction model, outputting a set Y = { C of target capacity prediction values 1 ,C 2 ,…,C m C denotes a target capacity prediction value, and m denotes the number of battery packs to be predicted.
In conclusion, the method and the device have the advantages that the temperature calibration mode is carried out on the operation data, so that the influence of the temperature on the capacity of the battery pack is reduced, and the accuracy of capacity prediction can be improved to a certain extent. In addition, when the capacity prediction model is trained, the capacity prediction model is optimized through the first optimization algorithm and the second optimization algorithm, the prediction accuracy of the capacity prediction model is improved, the capacity prediction model can fully learn the nonlinear mapping relation between various operation data (such as voltage, current, temperature, constant-current charging time and the like of a battery) and the capacity of a battery pack in the aging process of the battery, rapid and accurate capacity prediction is provided, and the requirement for predicting the capacity in real time is met. The first optimization algorithm avoids overfitting of the capacity prediction model, the second optimization algorithm improves the calculation efficiency of the capacity prediction model, and compared with a common neural network model training method, the method provided by the application has the advantages of being long in training time and few in parameter adjustment.
Fig. 7 is a block diagram of a battery pack capacity prediction apparatus 7 according to an embodiment of the present application.
In some embodiments, the battery capacity prediction device 7 may include a plurality of functional modules composed of computer program segments. The computer programs of the various program segments in the battery capacity prediction means 7 may be stored in a memory of the electronic device and executed by at least one processor to perform the functions of battery capacity prediction (described in detail in fig. 1).
In the present embodiment, the battery pack capacity prediction device 7 may be divided into a plurality of functional blocks according to the functions performed by the device. The functional module may include: an acquisition module 710, a temperature calibration module 720, and a prediction module 730. A module as referred to herein is a sequence of computer program segments capable of being executed by at least one processor and of performing a fixed function and stored in a memory. In the embodiment of the present application, the definition of the battery capacity prediction device 7 may refer to the definition of the battery capacity prediction method above, and details are not described herein again.
The obtaining module 710 is configured to obtain operation data of a battery pack, an operation temperature corresponding to the operation data, and a preset calibration temperature.
The temperature calibration module 720 is configured to perform temperature calibration on the operation data according to the operation temperature and the calibration temperature to obtain target data.
The prediction module 730 is configured to input the target data into a trained capacity prediction model to obtain a target capacity prediction value of the battery pack.
In some optional embodiments, the temperature calibration module 720 is further configured to:
acquiring a preset temperature calibration model, a calibration temperature and an operation temperature corresponding to the operation data;
and carrying out temperature calibration on the operation data according to the temperature calibration model, the calibration temperature and the operation temperature to obtain the target data.
In some optional embodiments, the temperature calibration module 720 is further configured to:
acquiring historical capacity data and historical temperature data corresponding to the historical capacity data;
and determining a temperature calibration model according to the historical capacity data and the historical temperature data.
In some optional embodiments, the obtaining module 710 is further configured to calculate an outlier corresponding to each data in the target data, and filter out data corresponding to the outlier larger than a first threshold to obtain filtered data;
performing dimensionality reduction calculation on the filtering data to obtain dimensionality reduction data;
correspondingly, the inputting the target data into the trained capacity prediction model to obtain the target capacity prediction value includes:
and inputting the dimension reduction data into the trained capacity prediction model to obtain a target capacity prediction value of the battery pack.
In some optional embodiments, the temperature calibration module 720 is further configured to:
optimizing neurons of a hidden layer of the initial capacity prediction model by using a first optimization algorithm to obtain an optimized capacity prediction model;
acquiring training data and a sample label corresponding to the training data;
inputting the training data into the optimized capacity prediction model to obtain a sample capacity prediction value;
calculating a loss function value of the optimized capacity prediction model according to the sample capacity prediction value and the sample label;
if the loss function value is larger than a second threshold value, taking the loss function value as the input of the optimized capacity prediction model for back propagation, optimizing the weight and the bias of the neuron corresponding to the back propagation by using a second optimization algorithm, updating the weight and the bias of the optimized capacity prediction model, and returning to execute the input of the training data into the optimized capacity prediction model;
and if the loss function value is smaller than or equal to a second threshold value, determining the current optimized capacity prediction model as a trained capacity prediction model.
In some optional embodiments, the optimizing the neuron number of the hidden layer of the initial volume prediction model by using the first optimization algorithm includes:
determining the state of each neuron in a hidden layer of the initial capacity prediction model based on a preset state probability value;
and removing the neurons with the state as the first state to obtain the optimized capacity prediction model.
In some optional embodiments, after determining the current optimized capacity prediction model as the trained capacity prediction model, the method further includes:
acquiring test data and a test label corresponding to the test data;
inputting the test data into the trained capacity prediction model to obtain a test capacity prediction value;
calculating an error value and a fitting degree of the trained capacity prediction model according to the test label and the test capacity prediction value;
and if the error value is larger than a third threshold value or the fitting degree is smaller than a fourth threshold value, returning to the step of acquiring the training data and the sample label corresponding to the training data.
Referring to fig. 1, in the present embodiment, the memory 12 may be an internal memory of the electronic device 1, that is, a memory built in the electronic device 1. In other embodiments, the memory 12 may also be an external memory of the electronic device 1, i.e. a memory externally connected to the electronic device 1.
In some embodiments, the memory 12 is used for storing program codes and various data, and realizes high-speed and automatic access to programs or data during the operation of the electronic device 1.
The memory 12 may include random access memory and may also include non-volatile memory such as a hard disk, a memory, a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash memory Card (Flash Card), at least one magnetic disk storage device, a Flash memory device, or other volatile solid state storage device.
In an embodiment, the Processor 13 may be a Central Processing Unit (CPU), other general purpose Processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA) or other Programmable logic device, a discrete Gate or transistor logic device, a discrete hardware component, or the like. A general purpose processor may be a microprocessor or the processor may be any other conventional processor or the like.
The program code and various data in the memory 12 may be stored in a computer-readable storage medium if they are implemented in the form of software functional units and sold or used as separate products. Based on such understanding, all or part of the processes in the methods of the embodiments, such as the method for predicting the capacity of the battery pack, may also be implemented by a computer program, which may be stored in a computer-readable storage medium, and when the computer program is executed by a processor, the steps of the embodiments of the methods may be implemented. Wherein the computer program comprises computer program code, which may be in the form of source code, object code, an executable file or some intermediate form, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, removable hard disk, magnetic disk, optical disk, computer Memory, read-Only Memory (ROM), or the like.
It is understood that the above described module division is a logical function division, and there may be other division ways in actual implementation. In addition, functional modules in the embodiments of the present application may be integrated into the same processing unit, or each module may exist alone physically, or two or more modules are integrated into the same unit. The integrated module can be realized in a hardware form, and can also be realized in a form of hardware and a software functional module.
Finally, it should be noted that the above embodiments are only used for illustrating the technical solutions of the present application and not for limiting, and although the present application is described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that modifications or equivalent substitutions can be made to the technical solutions of the present application without departing from the spirit and scope of the technical solutions of the present application.
Claims (10)
1. A battery capacity prediction method, the method comprising:
acquiring operation data of a battery pack, operation temperature corresponding to the operation data and a preset calibration temperature;
carrying out temperature calibration on the operation data according to the operation temperature and the calibration temperature to obtain target data;
and inputting the target data into a trained capacity prediction model to obtain a target capacity prediction value of the battery pack.
2. The battery pack capacity prediction method of claim 1, wherein the temperature calibrating the operating data according to the operating temperature and the calibration temperature to obtain target data comprises:
acquiring a preset temperature calibration model;
and carrying out temperature calibration on the operation data according to the temperature calibration model, the calibration temperature and the operation temperature to obtain the target data.
3. The battery pack capacity prediction method of claim 2, wherein prior to the obtaining a preset temperature calibration model, the method further comprises:
acquiring historical capacity data and historical temperature data corresponding to the historical capacity data;
and determining a temperature calibration model according to the historical capacity data and the historical temperature data.
4. The battery pack capacity prediction method according to any one of claims 1 to 3, wherein before the inputting the target data into a capacity prediction model to obtain a target capacity prediction value of the battery pack, the method further comprises:
calculating an outlier corresponding to each data in the target data, and filtering out data corresponding to the outliers larger than a first threshold value to obtain filtering data;
performing dimensionality reduction calculation on the filtering data to obtain dimensionality reduction data;
correspondingly, the inputting the target data into the trained capacity prediction model to obtain the target capacity prediction value of the battery pack includes:
and inputting the dimensionality reduction data into the trained capacity prediction model to obtain a target capacity prediction value of the battery pack.
5. The battery pack capacity prediction method according to claim 1, wherein before the obtaining of the operation data of the battery pack, the method comprises:
optimizing neurons of a hidden layer of the initial capacity prediction model by using a first optimization algorithm to obtain an optimized capacity prediction model;
acquiring training data and a sample label corresponding to the training data;
inputting the training data into the optimized capacity prediction model to obtain a sample capacity prediction value;
calculating a loss function value of the optimized capacity prediction model according to the sample capacity prediction value and the sample label;
if the loss function value is larger than a second threshold value, taking the loss function value as the input of the optimized capacity prediction model for back propagation, optimizing the weight and the bias of the neuron corresponding to the back propagation by using a second optimization algorithm, updating the weight and the bias of the optimized capacity prediction model, and returning to execute the step of inputting the training data into the optimized capacity prediction model;
and if the loss function value is smaller than or equal to a second threshold value, determining the current optimized capacity prediction model as a trained capacity prediction model.
6. The battery pack capacity prediction method of claim 5, wherein the optimizing the neuron number of the hidden layer of the initial capacity prediction model using a first optimization algorithm comprises:
determining the state of each neuron in a hidden layer of the initial capacity prediction model based on a preset state probability value;
and removing the neurons with the state as the first state to obtain the optimized capacity prediction model.
7. The battery pack capacity prediction method of claim 5, wherein after the determining the current optimized capacity prediction model as a trained capacity prediction model, the method further comprises:
acquiring test data and a test label corresponding to the test data;
inputting the test data into the trained capacity prediction model to obtain a test capacity prediction value;
calculating an error value and a fitting degree of the trained capacity prediction model according to the test label and the test capacity prediction value;
and if the error value is larger than a third threshold value or the fitting degree is smaller than a fourth threshold value, returning to the step of acquiring the training data and the sample label corresponding to the training data.
8. An apparatus for predicting battery capacity, the apparatus comprising an acquisition module, a temperature calibration module, and a prediction module:
the acquisition module is used for acquiring the operation data of the battery pack, the operation temperature corresponding to the operation data and a preset calibration temperature;
the temperature calibration module is used for carrying out temperature calibration on the operation data according to the operation temperature and the calibration temperature to obtain target data;
and the prediction module is used for inputting the target data into the trained capacity prediction model to obtain a target capacity prediction value of the battery pack.
9. An electronic device, characterized in that the electronic device comprises a processor and a memory, the processor being configured to execute a computer program stored in the memory to implement the battery capacity prediction method according to any one of claims 1 to 7.
10. A computer-readable storage medium storing at least one instruction which, when executed by a processor, implements a battery capacity prediction method as claimed in any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211113456.XA CN115358484A (en) | 2022-09-14 | 2022-09-14 | Battery pack capacity prediction method and related equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211113456.XA CN115358484A (en) | 2022-09-14 | 2022-09-14 | Battery pack capacity prediction method and related equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115358484A true CN115358484A (en) | 2022-11-18 |
Family
ID=84006059
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211113456.XA Pending CN115358484A (en) | 2022-09-14 | 2022-09-14 | Battery pack capacity prediction method and related equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115358484A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117505811A (en) * | 2024-01-08 | 2024-02-06 | 北京适创科技有限公司 | Die temperature control method and related device |
-
2022
- 2022-09-14 CN CN202211113456.XA patent/CN115358484A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117505811A (en) * | 2024-01-08 | 2024-02-06 | 北京适创科技有限公司 | Die temperature control method and related device |
| CN117505811B (en) * | 2024-01-08 | 2024-04-05 | 北京适创科技有限公司 | Die temperature control method and related device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Li et al. | Lithium-ion battery capacity estimation—A pruned convolutional neural network approach assisted with transfer learning | |
| Sui et al. | A review of non-probabilistic machine learning-based state of health estimation techniques for Lithium-ion battery | |
| Xiao et al. | Accurate state-of-charge estimation approach for lithium-ion batteries by gated recurrent unit with ensemble optimizer | |
| Che et al. | Data efficient health prognostic for batteries based on sequential information-driven probabilistic neural network | |
| CN112946484B (en) | SOC estimation method, system, terminal equipment and readable storage medium based on BP neural network | |
| Liu et al. | A hierarchical and flexible data-driven method for online state-of-health estimation of Li-ion battery | |
| CN114325450A (en) | Lithium ion battery health state prediction method based on CNN-BilSTM-AT hybrid model | |
| JP2023520970A (en) | Lithium battery SOC estimation method, apparatus, and computer-readable storage medium | |
| CN113065283A (en) | Battery life prediction method, system, electronic device and storage medium | |
| CN112834927A (en) | Lithium battery residual life prediction method, system, device and medium | |
| CN115201686B (en) | Lithium ion battery health state assessment method under incomplete charge and discharge data | |
| Takyi-Aninakwa et al. | A hybrid probabilistic correction model for the state of charge estimation of lithium-ion batteries considering dynamic currents and temperatures | |
| CN114705990A (en) | Battery cluster state of charge estimation method and system, electronic equipment and storage medium | |
| CN115358484A (en) | Battery pack capacity prediction method and related equipment | |
| CN116106761A (en) | Real-time lithium ion battery electric quantity estimation method based on typical correlation analysis | |
| CN113495214B (en) | Super-capacitor state-of-charge estimation method based on temperature change model | |
| CN111260015A (en) | Lithium ion battery model parameter identification method based on chaotic cat swarm algorithm | |
| Qian et al. | A CNN-SAM-LSTM hybrid neural network for multi-state estimation of lithium-ion batteries under dynamical operating conditions | |
| Di Capua et al. | Using genetic programming to learn behavioral models of lithium batteries | |
| Dong et al. | State of health estimation and remaining useful life estimation for Li-ion batteries based on a hybrid kernel function relevance vector machine | |
| CN117110880A (en) | Battery multi-state joint estimation method for cloud-edge end cooperation | |
| CN114610595A (en) | Method, device, equipment and storage medium for identifying model performance influence factors | |
| KR20230028017A (en) | A model-based state-of-charge estimation device for lithium-ion batteries that is robust to time-varying load current situations and method thereof | |
| CN113344245A (en) | Hybrid deep learning short-term prediction model, method, storage medium, and computing device | |
| Di Capua et al. | A behavioral model for lithium batteries based on genetic programming |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |