Battery soc prediction method and device based on digital twinning, equipment and medium

Abstract

The present application belongs to the technical field of digital twinning, and discloses a battery SOC prediction method, device, equipment and medium based on digital twinning; the prediction method comprises: collecting current, voltage and environmental temperature measurement values of the battery; inputting the current and voltage measurement values of the battery into a Kalman filter for state estimation to obtain a Kalman filter state estimation value; inputting the current measurement value, voltage measurement value, environmental temperature measurement value and Kalman filter state estimation value of the battery into a pre-trained LTSM neural network to obtain a battery SOC prediction value. The present application adopts a modeling method of data driving and physical mechanism fusion, improves the accuracy of the battery energy storage system model, improves the prediction accuracy of the battery state of charge SOC, and provides data support for the health level management of the battery energy storage system.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of digital twinning, and particularly relates to a battery SOC prediction method and device based on digital twinning, equipment and a medium. BACKGROUND

[0002] Battery management is crucial to improve the safety, reliability and performance of the battery system. Specifically, accurate estimation of the battery state of charge SOC provides basic information on the battery charging and discharging level, and effective management thereof can help prolong the battery life. However, the battery charging and discharging is a nonlinear process with strong time-varying characteristics, and environmental factors also have an impact thereon. Therefore, effective modeling thereof has been a research focus.

[0003] However, the modeling method of the prior art still has the technical problem of low model accuracy due to the characteristics of the battery charging and discharging, resulting in low prediction accuracy of the battery state of charge SOC. SUMMARY

[0004] The application aims to provide a battery SOC prediction method and device based on digital twinning, which adopts a data-driven and physical mechanism fusion modeling method, improves the accuracy of the battery energy storage system model, improves the prediction accuracy of the battery state of charge SOC, and provides data support for the health level management of the battery energy storage system.

[0005] To achieve the above-mentioned purpose, the application adopts the following technical solutions:

[0006] In a first aspect, the application provides a battery SOC prediction method based on digital twinning, comprising:

[0007] collecting current, voltage and ambient temperature measurement values of the battery;

[0008] inputting the current, voltage and ambient temperature measurement values of the battery into a Kalman filter for state estimation to obtain a Kalman filter state estimation value;

[0009] inputting the current, voltage and ambient temperature measurement values of the battery into a Kalman filter for state estimation to obtain a Kalman filter state estimation value;

[0010] The application further improves that in the step of inputting the current, voltage and ambient temperature measurement values of the battery into a Kalman filter for state estimation to obtain a Kalman filter state estimation value, the state equation of the Kalman filter is:

[0011] (9)

[0012] T is the system sampling period; Qn is the system noise quantity; the state input vector: [ ]; is the battery state of charge SOC; represents the voltage on resistor R1 at time k; represents the voltage on resistor R2 at time k; the time constant is the product of resistor R1 and capacitor C1, is the product of resistor R2 and capacitor C2; the product of resistor and capacitor; is the battery current.

[0013] The further improvement in the step of inputting the current, voltage and ambient temperature measurement values of the battery into the Kalman filter for state estimation to obtain the Kalman filter state estimation value comprises: predicting the current and voltage measurement values of the battery through a state prediction equation and a covariance prediction equation, adjusting the prediction values through a Kalman gain adjustment equation, and correcting the adjusted values through a state update equation and a covariance update equation to obtain the Kalman filter state estimation value.

[0014] The state prediction equation, the covariance prediction equation, the Kalman gain adjustment equation, the state update equation and the covariance update equation are derived from the state space equation of the battery.

[0015] The further improvement in the step of inputting the current, voltage and ambient temperature measurement values of the battery into the Kalman filter for state estimation to obtain the Kalman filter state estimation value comprises: predicting the current, voltage and ambient temperature measurement values of the battery through a state prediction equation and a covariance prediction equation, adjusting the prediction values through a Kalman gain adjustment equation, and correcting the adjusted values through a state update equation and a covariance update equation to obtain the Kalman filter state estimation value.

[0016] (10)

[0017] Wherein: is the terminal voltage of the battery, input u (k) is the battery current; the system noise and the observation noise are respectively: w (k) and v (k).

[0018] The further improvement in the step of inputting the current, voltage and ambient temperature measurement values of the battery into the Kalman filter for state estimation to obtain the Kalman filter state estimation value comprises: predicting the current, voltage and ambient temperature measurement values of the battery through a state prediction equation and a covariance prediction equation, adjusting the prediction values through a Kalman gain adjustment equation, and correcting the adjusted values through a state update equation and a covariance update equation to obtain the Kalman filter state estimation value.

[0019] (6)

[0020] (7)

[0021] (8)

[0022] where I is the battery input current, is the terminal voltage, represents the open-circuit voltage with a nonlinear relationship with the battery SOC, is the ohmic resistance, and are the resistance and capacitance of the two RC circuits in the Thevenin equivalent circuit, are the voltage drops on the .

[0023] The further improvement of the present application is that in the step of inputting the current measurement value, voltage measurement value, environmental temperature measurement value and Kalman filter state estimation value of the battery into the pre-trained LSTM neural network to obtain the battery SOC prediction value, the LSTM neural network is composed of a cell state, a forgetting gate, an input gate and an output gate.

[0024] The update equations of the forgetting gate, the input gate, the cell state and the output gate are:

[0025] (1)

[0026] (2)

[0027] (3)

[0028] (4)

[0029] is the input data of the LSTM neural network at time t, including the battery voltage, current, environmental temperature and state estimation value of the Kalman filter; is a sigmoid function; represents the forgetting gate output; w and b represent the weight coefficients of each layer of neurons; i t is the output vector of the input gate; tanh is a tanh function; represents the convolution layer output; represents the output layer output; represents the final output of the LSTM neural network, which is the battery SOC prediction value.

[0030] The further improvement of the present application is that in the step of inputting the current measurement value, voltage measurement value, environmental temperature measurement value and Kalman filter state estimation value of the battery into the pre-trained LSTM neural network to obtain the battery SOC prediction value, the training method of the LSTM neural network comprises:

[0031] Obtain historical aging data of the battery in different working conditions, the historical aging data including current, voltage and ambient temperature; input the historical aging data as a time series into a Kalman filter to obtain a set of Kalman filter state estimation values;

[0032] Normalize the battery historical aging data and the Kalman filter state estimation values; normalize the current, voltage and ambient temperature data and the corresponding Kalman filter state estimation values as a set of input data; obtain a plurality of sets of input data, and divide the plurality of sets of input data into a training set and a test set;

[0033] Train the LSTM neural network using the training set to obtain a pre-trained LSTM neural network.

[0034] In a second aspect, the present application provides a battery SOC prediction device based on digital twinning, comprising:

[0035] A collection module configured to collect current, voltage and ambient temperature measurement values of the battery;

[0036] A state estimation module configured to input the current, voltage and ambient temperature measurement values of the battery into a Kalman filter for state estimation to obtain Kalman filter state estimation values;

[0037] A prediction module configured to input the current, voltage and ambient temperature measurement values of the battery and the Kalman filter state estimation values into a pre-trained LSTM neural network to obtain a battery SOC prediction value.

[0038] In a third aspect, the present application provides an electronic device comprising a processor and a memory, wherein the processor is configured to execute a computer program stored in the memory to implement the battery SOC prediction method based on digital twinning.

[0039] In a fourth aspect, the present application provides a computer readable storage medium storing at least one instruction, wherein the at least one instruction is executed by a processor to implement the battery SOC prediction method based on digital twinning.

[0040] Compared with the prior art, the present application has the following beneficial effects:

[0041] This invention provides a battery SOC prediction method, device, equipment, and medium based on digital twins. It collects measurements of the battery's current, voltage, and ambient temperature; inputs these measurements into a Kalman filter for state estimation, obtaining the Kalman filter state estimate; and inputs the battery's current, voltage, ambient temperature, and Kalman filter state estimate into a pre-trained LSTM neural network to obtain the predicted battery SOC. This invention, by sensing the measurable information of the modeling object and considering the nonlinear and stochastic nature of the battery energy storage system's dynamic processes, employs a data and mechanism fusion approach to construct a digital twin model of the complex object. This ensures the synchronization of the modeled object's state with the physical entity, achieving accurate estimation of state variables and providing a model foundation for subsequent battery management.

[0042] In this approach, the data-driven and mechanism-integrated modeling method is specifically manifested in two aspects: First, a dynamic model of the battery is constructed based on the Thevenin equivalent circuit, and a Kalman filter is used to estimate battery parameters, such as resistance, capacitance, rated capacity, and state of charge (SOC). Second, based on historical battery aging data, an LSTM neural network is used to construct a model of the battery energy storage system, and the model is trained. Finally, real-time measured battery voltage, current, and ambient temperature data, along with the system estimation parameters obtained from the Kalman filter, are used as inputs to the trained LSTM neural network, thereby obtaining the predicted SOC value of the battery energy storage system. This data-driven and physical mechanism-integrated modeling method improves the accuracy of the battery energy storage system model and the prediction accuracy of the battery SOC, providing data support for the health management of battery energy storage systems.

[0043] The method of this invention adopts a data-driven and mechanism-integrated approach, leveraging the memory function of LSTM neural networks for the charging and discharging behavior of batteries under different operating conditions and time environments. It also has the advantage of directly fitting battery dynamic behavior with data without building a nonlinear system physical model. Furthermore, by combining Kalman filtering for closed-loop management of system errors, the open-loop risk of LSTM neural networks in terms of error constraints is reduced. This ensures the overall accuracy of battery dynamic process modeling, improves the accuracy of its State of Charge (SOC) prediction, and enhances the level of battery health management.

[0044] This paper proposes a method to model battery energy storage systems by fusing LSTM neural networks with Kalman filters. Firstly, the LSTM neural network, with its memory function, can remember the dynamic uncertainties of the battery under various operating conditions. Secondly, using the state estimates from the Kalman filter as training data for the LSTM neural network reduces the open-loop risk of errors, thereby improving the overall accuracy of the battery energy storage system model and enhancing the robustness of the method. Attached Figure Description

[0045] The accompanying drawings, which form a part of this specification, are included to provide a further understanding of the application, and are incorporated into and constitute a part of this specification. The embodiments of the application, and their

[0046] Figure 1 A schematic diagram of a battery SOC prediction method based on digital twinning according to the present application;

[0047] Figure 2 A schematic diagram of an LSTM neural network structure;

[0048] Figure 3 A Thevenin equivalent circuit of a battery;

[0049] Figure 4 A schematic diagram of a battery state of charge SOC estimation step based on a Kalman filter;

[0050] Figure 5 A input-output relationship diagram of a Kalman filter;

[0051] Figure 6 A flowchart of a battery SOC prediction method based on digital twinning according to the present application;

[0052] Figure 7 A structure block diagram of a battery SOC prediction device based on digital twinning according to the present application;

[0053] Figure 8 A structure block diagram of an electronic device according to the present application. DETAILED DESCRIPTION

[0054] The application will be described in greater detail with reference to the accompanying drawings, in which embodiments of the application are shown. As long as there is no conflict, the embodiments of the application and the features in the embodiments can be combined with each other.

[0055] The following detailed description is exemplary in nature and is intended to provide further description of the application. Unless otherwise defined, all technical terms used herein are to be understood as commonly understood by one of ordinary skill in the art to which the application pertains. The terminology used in the description of the application herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the example embodiments according to the application.

[0056] Example 1

[0057] Referring to Figures 1 to 6 The present application provides a battery SOC prediction method based on digital twinning, which comprises the following steps:

[0058] S1, collecting current, voltage and ambient temperature measurement values of the battery;

[0059] S2, input the current, voltage and ambient temperature measurement values of the battery into a Kalman filter for state estimation to obtain a Kalman filter state estimation value; the Kalman filter state estimation value includes a battery SOC estimation value and a battery voltage estimation value and measurement value error;

[0060] S3, input the current measurement value, voltage measurement value, ambient temperature measurement value and Kalman filter state estimation value of the battery into a pre-trained LSTM neural network to obtain a battery SOC prediction value.

[0061] Referring to Figure 2 The LSTM neural network established by the application mainly consists of a cell state, a forgetting gate, an input gate and an output gate.

[0062] The forgetting gate: determines which redundant information of the upper layer to discard. A value of 0 to 1 is generated through the forgetting gate according to the output value of the last moment and the current input, to determine whether to let the information learned at the last moment pass or partially pass. Formula (1) represents the forgetting gate update.

[0063] The input gate and the cell state: used to update the input information of the network, and the output information of the forgetting gate is used to update the cell state together. Formulas (2) and (3) represent the input gate and cell state update, respectively.

[0064] The output gate: used to help process the network information to obtain the output of the cell structure. Formula (4) represents the output gate update.

[0065] (1)

[0066] (2)

[0067] (3)

[0068] (4)

[0069] is the input data of the LSTM neural network at t moment, including the battery voltage, current, ambient temperature and state estimation value of the Kalman filter; is a sigmoid function; represents the output of the forgetting gate; w and b represent the weight coefficients of the neurons of each layer; i t is the output vector of the input gate; tanh is a tanh function; represents the output of the convolution layer; represents the output of the output layer; represents the final output of the LSTM neural network, i.e. the battery SOC prediction value.

[0070] Based on the above LSTM neural network training, the mapping relationship of battery voltage, current, environmental temperature, Kalman filter state estimation value and battery state of charge SOC is obtained f .

[0071] (5)

[0072] The training of the LSTM neural network comprises the following steps:

[0073] 1. Prepare training data.

[0074] The data source is two parts; the first part: the historical aging data of the battery under different working conditions, specifically including current, voltage, and environmental temperature; the second part: the battery SOC estimation value based on Kalman filtering.

[0075] The current, voltage and environmental temperature data in the historical aging data of the battery under different working conditions are input into the Kalman filter as time series, and a group of Kalman filter state estimation values are obtained, including battery SOC estimation value and battery voltage estimation value and measurement error.

[0076] 2. Normalize the current, voltage and environmental temperature data in the battery historical aging data and the Kalman filter state estimation value; the normalized current, voltage, environmental temperature data and the corresponding Kalman filter state estimation value are used as a group of input data; obtain several groups of input data, and divide them into training set and test set according to the ratio of 7:3.

[0077] 3. Train the LSTM neural network with the training set until the prediction error is less than the set threshold, or stop after reaching the maximum iteration number, obtain the pre-trained LSTM neural network. And test the pre-trained LSTM neural network with the test set.

[0078] The present application is based on Figure 3 The Davinian equivalent circuit is shown, and a battery dynamics model based on the Davinian equivalent circuit is established:

[0079] (6)

[0080] (7)

[0081] (8)

[0082] Wherein, I is the input current of the battery, is the terminal voltage, represents the open circuit voltage with nonlinear relationship with the battery SOC, is the ohmic resistance, and The resistance and capacitance of the two RC circuits, respectively The voltage drop on the resistance.

[0083] Battery energy storage system related parameter estimation based on Kalman filter:

[0084] The input of Kalman filter: battery measured voltage, current. The battery energy storage system is a typical nonlinear system, and its state estimation is affected by environmental noise and measurement accuracy. Kalman filter is a common method to achieve optimal state estimation by filtering noise. The battery dynamics model can be constructed by the above Thevenin equivalent circuit, and the state equation of the battery energy storage system Kalman filter can be obtained:

[0085] (9)

[0086] T is the system sampling period; Q n is the system noise quantity; the state input vector: [ ]; is the battery state of charge SOC at time k; represents the voltage on the resistance R1 at time k; represents the voltage on the resistance R2 at time k; the time constant is the product of resistance R1 and capacitance C1, is the product of resistance R2 and capacitance C2, the product of resistance and capacitance; is the battery current. Based on this, the state space equation of the battery energy storage system can be obtained:

[0087] (10)

[0088] Where: is the terminal voltage of the battery, the input u (k) is the battery current; the system noise and observation noise are: w (k) and v (k).

[0089] Please refer to Figure 4 , the steps of battery state of charge SOC estimation based on Kalman filter are as follows: mainly including prediction, adjustment and correction three parts. Among them, the state prediction and update equation, covariance prediction and update equation, Kalman gain adjustment equation are derived from the state space equation of the battery energy storage system.

[0090] Please refer to Figure 5As shown, the input of the Kalman filter is the current measurement value, voltage measurement value of the battery, and the output is the Kalman filter state estimation value; the Kalman filter state estimation value includes the battery SOC estimation value and the error between the battery voltage estimation value and the measurement value; the Kalman filter state estimation value is input as part of the prediction of the LSTM neural network model.

[0091] In actual use, the current, voltage and environmental temperature measurement values of the battery are actually collected; the current, voltage and environmental temperature measurement values of the battery are input into the Kalman filter for state estimation to obtain the Kalman filter state estimation value; the current measurement value, voltage measurement value and environmental temperature measurement value of the battery and the Kalman filter state estimation value are input into the pre-trained LSTM neural network to obtain the battery state of charge prediction value.

[0092] The application utilizes the data-driven and physical mechanism fusion mode to construct a digital twin model of the battery energy storage system, uses the state estimation value of the Kalman filter as the input of the LSTM neural network, and finally realizes the SOC output of the battery.

[0093] Embodiment 2

[0094] Please refer to Figure 7 As shown, the application provides a battery SOC prediction device based on digital twinning, which comprises:

[0095] The acquisition module is used for acquiring the current, voltage and environmental temperature measurement values of the battery;

[0096] The state estimation module is used for inputting the current, voltage and environmental temperature measurement values of the battery into the Kalman filter for state estimation to obtain the Kalman filter state estimation value;

[0097] The prediction module is used for inputting the current measurement value, voltage measurement value and environmental temperature measurement value of the battery and the Kalman filter state estimation value into the pre-trained LSTM neural network to obtain the battery SOC prediction value.

[0098] In the state estimation module, the state equation of the Kalman filter is:

[0099] (9)

[0100] T is the system sampling period; Q n is the system noise quantity; the state input vector is: =[ ]; SOC is the battery state of charge at k time; V1 represents the voltage on the resistor R1 at k time; V2 represents the voltage on the resistor R2 at k time; the time constant is the product of resistance R1 and capacitance C1, is the product of resistance R2 and capacitance C2, the product of resistance and capacitance; is the battery current.

[0101] The specific execution steps of the state estimation module include: predicting the current and voltage measurement values of the battery through a state prediction equation and a covariance prediction equation, adjusting the predicted values through a Kalman gain adjustment equation, and correcting the adjusted values through a state update equation and a covariance update equation to obtain a Kalman filter state estimation value;

[0102] The state prediction equation, the covariance prediction equation, the Kalman gain adjustment equation, the state update equation, and the covariance update equation are all derived from a state space equation of the battery

[0103] The state space equation of the battery is:

[0104] (10)

[0105] Wherein: is the terminal voltage of the battery, input u (k) is the battery current; the system noise and the observation noise are: w (k) and v (k);

[0106] The state equation of the Kalman filter is obtained from a battery dynamics model based on the Thevenin equivalent circuit; the battery dynamics model based on the Thevenin equivalent circuit is:

[0107] (6)

[0108] (7)

[0109] (8)

[0110] Wherein, I is the input current of the battery, is the terminal voltage, represents the open-circuit voltage having a nonlinear relationship with the battery SOC, is the ohmic resistance, and are the resistance and capacitance of two RC circuits in the Thevenin equivalent circuit, respectively, are the voltage drops on .

[0111] In the prediction module, the LSTM neural network is composed of a cell state, a forget gate, an input gate, and an output gate;

[0112] The update equations of the forget gate, the input gate, the cell state and the output gate are as follows:

[0113] (1)

[0114] (2)

[0115] (3)

[0116] (4)

[0117] is the input data of the LSTM neural network at time t, including the battery voltage, current, ambient temperature and state estimation value of the Kalman filter; is a sigmoid function; represents the output of the forget gate; w and b represent the weight coefficients of the neurons in each layer; i t is the output vector of the input gate; tanh is a tanh function; represents the output of the convolutional layer; represents the output of the output layer; represents the final output of the LSTM neural network, which is the battery SOC prediction value.

[0118] In the prediction module, the training method of the LSTM neural network comprises:

[0119] Obtain historical aging data of the battery under different working conditions, wherein the historical aging data includes current, voltage and ambient temperature; input the historical aging data as a time series into the Kalman filter to obtain a set of state estimation values of the Kalman filter;

[0120] Normalize the battery historical aging data and the state estimation values of the Kalman filter; normalize the current, voltage and ambient temperature data and the corresponding state estimation values of the Kalman filter as a set of input data; obtain a plurality of sets of input data, and divide the plurality of sets of input data into a training set and a test set;

[0121] Train the LSTM neural network using the training set to obtain a pre-trained LSTM neural network.

[0122] Embodiment 3

[0123] Please refer to Figure 8 The application also provides an electronic device 100 based on the battery SOC prediction method based on digital twinning; the electronic device 100 comprises a memory 101, at least one processor 102, a computer program 103 stored in the memory 101 and executable on the at least one processor 102, and at least one communication bus 104.

[0124] The memory 101 can be used to store the computer program 103, and the processor 102 can realize the steps of the battery SOC prediction method based on digital twinning by running or executing the computer program stored in the memory 101 and calling the data stored in the memory 101. The memory 101 can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application program required by a function (such as a sound playing function, an image playing function, etc.), and the like; and the data storage area can store data (such as audio data) created according to the use of the electronic device 100. In addition, the memory 101 can include a 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 card, at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device.

[0125] The at least one processor 102 can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The processor 102 can be a microprocessor or any conventional processor, etc. The processor 102 is the control center of the electronic device 100, and is connected to various parts of the electronic device 100 through various interfaces and lines.

[0126] The memory 101 in the electronic device 100 stores a plurality of instructions to realize a battery SOC prediction method based on digital twinning, and the processor 102 can execute the plurality of instructions to realize:

[0127] Collecting current, voltage and environmental temperature measurement values of the battery;

[0128] Inputting the current, voltage and environmental temperature measurement values of the battery into a Kalman filter for state estimation to obtain a Kalman filter state estimation value;

[0129] The current measurement value, voltage measurement value, environment temperature measurement value, Kalman filter state estimation value of the battery are input into the pre-trained LSTM neural network to obtain a battery SOC prediction value.

[0130] Embodiment 4

[0131] The modules / units integrated in the electronic device 100, if realized in the form of software function units and sold or used as independent products, can be stored in a computer readable storage medium. Based on such understanding, all or part of the processes in the above-mentioned embodiment methods can also be completed by a computer program instructing related hardware, and the computer program can be stored in a computer readable storage medium. The computer program can implement the steps of each method embodiment described above when executed by a processor. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or some intermediate forms, etc. The computer readable medium can include any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, and read-only memory (ROM).

[0132] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer usable program code.

[0133] The present application is described with reference to flowcharts and / or block diagrams of the method, device (system), and computer program product according to the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of the flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device produce a device that implements the functions specified in the flowcharts and / or block diagrams. Figure 1 The device that implements the functions specified in one or more flows and / or blocks Figure 1 The device that implements the functions specified in one or more flows and / or blocks

[0134] These computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer readable memory produce an article of manufacture including instructions which implement the flow Figure 1 The functions specified in the flow or flows and / or blocks Figure 1 The functions specified in the flow or flows and / or blocks

[0135] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flow Figure 1 The functions specified in the flow or flows and / or blocks Figure 1 The functions specified in the flow or flows and / or blocks

[0136] Finally, it should be noted that the above-mentioned embodiments are merely used to illustrate the technical solutions of the present application, rather than limit the present application. Although the present application has been described in detail with reference to the above-mentioned embodiments, those skilled in the art should understand that the specific embodiments of the present application can be modified or replaced equivalently without departing from the spirit and scope of the present application, and any modification or replacement without departing from the spirit and scope of the present application should be covered in the protection scope of the claims of the present application.

Claims

1. A battery SOC prediction method based on digital twinning, characterized in that, The method comprises the following steps: collecting current, voltage and ambient temperature measurements of the battery; inputting the current, voltage and ambient temperature measurements of the battery into a Kalman filter for state estimation to obtain Kalman filter state estimation values; inputting the current, voltage and ambient temperature measurements of the battery, the Kalman filter state estimation values into a pre-trained LSTM neural network to obtain battery SOC prediction values; the state equation of the Kalman filter is obtained from a battery dynamics model based on a Thevenin equivalent circuit; the battery dynamics model based on the Thevenin equivalent circuit is: （6） （7） （8） where I is the battery input current, V is the terminal voltage, OCV is the open circuit voltage which has a non-linear relationship with the battery SOC, R is the ohmic resistance, and R and C are the resistance and capacitance of the two RC circuits in the Thevenin equivalent circuit, respectively, the voltage drop across R. the training method of the LSTM neural network comprises: obtaining historical aging data of the battery under different working conditions, the historical aging data comprising current, voltage and ambient temperature; inputting the historical aging data as a time sequence into a Kalman filter to obtain a set of Kalman filter state estimation values; normalizing the battery historical aging data and the Kalman filter state estimation values; normalizing the current, voltage and ambient temperature data and the corresponding Kalman filter state estimation values as a set of input data; obtaining a plurality of sets of input data, and dividing the plurality of sets of input data into a training set and a test set; training the LSTM neural network using the training set to obtain a pre-trained LSTM neural network; in the step of inputting the current, voltage and ambient temperature measurements of the battery into the Kalman filter for state estimation to obtain Kalman filter state estimation values, the state equation of the Kalman filter is: （9） T is the system sampling period; Q n is the system noise quantity; the state input vector: [ ]; is the battery state of charge SOC at time k; denotes the voltage over resistor R1 at time k; denotes the voltage over resistor R2 at time k; the time constant is the product of resistor R1 and capacitor C1, is the product of resistor R2 and capacitor C2 the product of resistor and capacitor; is the battery current; in the step of inputting the current, voltage and ambient temperature measurements of the battery, the Kalman filter state estimation values into the pre-trained LSTM neural network to obtain battery SOC prediction values, the LSTM neural network is composed of a cell state, a forget gate, an input gate and an output gate; the update equations of the forget gate, the input gate, the cell state and the output gate are: （1） （2） （3） （4） is the input data of the LSTM neural network at time t, including the battery voltage, current, ambient temperature and state estimation value of the Kalman filter; is a sigmoid function; represents the output of the forgetting gate; w and b represent the weight coefficients of the neurons in each layer; i t is the output vector of the input gate; tanh is a tanh function; represents the output of the convolutional layer; represents the output of the output layer; represents the final output of the LSTM neural network, the battery SOC prediction value; in the step of inputting the current, voltage and ambient temperature measurements of the battery into the Kalman filter for state estimation to obtain Kalman filter state estimation values, specifically comprising: predicting the current and voltage measurements of the battery through a state prediction equation and a covariance prediction equation, adjusting the prediction values through a Kalman gain adjustment equation, and correcting the adjusted values through a state update equation and a covariance update equation to obtain the Kalman filter state estimation values; wherein the state prediction equation, the covariance prediction equation, the Kalman gain adjustment equation, the state update equation and the covariance update equation are derived from the state space equation of the battery. 2.The digital-twin-based battery SOC prediction method of claim 1, wherein, in the step of inputting the current, voltage and ambient temperature measurements of the battery into the Kalman filter for state estimation to obtain Kalman filter state estimation values, the state space equation of the battery is: （10） where: V is the terminal voltage of the battery, input u (k) is the battery current; the system noise and observation noise are: w (k) and v (k).

3. A battery SOC prediction apparatus based on digital twinning, characterized by, The method comprises the following steps: a collecting module for collecting current, voltage and ambient temperature measurements of the battery; a state estimation module for inputting the current, voltage and ambient temperature measurements of the battery into a Kalman filter for state estimation to obtain Kalman filter state estimation values; a Kalman filter state estimation module for inputting the current, voltage and ambient temperature measurements of the battery into a Kalman filter for state estimation to obtain Kalman filter state estimation values; The prediction module is configured to input the current measurement value, the voltage measurement value, the ambient temperature measurement value and the Kalman filter state estimation value of the battery into a pre-trained LSTM neural network to obtain a battery SOC prediction value. The state equation of the Kalman filter is obtained from a battery dynamics model based on a Thevenin equivalent circuit. （6） （7） （8） where I is the battery input current, is the terminal voltage, is the open circuit voltage which has a non-linear relationship with the battery SOC, is the ohmic resistance, and are the resistance and capacitance of the two RC circuits in the Thevenin equivalent circuit, are the voltage drops across the resistors, respectively. The training method of the LSTM neural network comprises the following steps: Obtain historical aging data of the battery under different working conditions, wherein the historical aging data comprises current, voltage and ambient temperature; input the historical aging data as a time sequence into a Kalman filter to obtain a set of Kalman filter state estimation values; Normalize the battery historical aging data and the Kalman filter state estimation values; normalize the current, voltage and ambient temperature data and the corresponding Kalman filter state estimation values as a set of input data; obtain a plurality of sets of input data, and divide the plurality of sets of input data into a training set and a test set; Train the LSTM neural network using the training set to obtain a pre-trained LSTM neural network; In the step of inputting the current, voltage and ambient temperature measurement values of the battery into the Kalman filter for state estimation to obtain a Kalman filter state estimation value, the state equation of the Kalman filter is: （9） T is the system sampling period; Q n is the system noise quantity; the state input vector: [ ]; is the battery state of charge SOC at time k; denotes the voltage over resistor R1 at time k; denotes the voltage over resistor R2 at time k; the time constant is the product of resistor R1 and capacitor C1, is the product of resistor R2 and capacitor C2 the product of resistor and capacitor; is the battery current; In the step of inputting the current measurement value, the voltage measurement value, the ambient temperature measurement value and the Kalman filter state estimation value of the battery into a pre-trained LSTM neural network to obtain a battery SOC prediction value, the LSTM neural network comprises a cell state, a forget gate, an input gate and an output gate; The update equations of the forget gate, the input gate, the cell state and the output gate are as follows: （1） （2） （3） （4） is the input data of the LSTM neural network at time t, including the battery voltage, current, ambient temperature and state estimation value of the Kalman filter; is a sigmoid function; represents the output of the forget gate; w and b represent the weight coefficients of the neurons in each layer; i t is the output vector of the input gate; tanh is a tanh function; represents the output of the convolutional layer; represents the output of the output layer; represents the final output of the LSTM neural network, the battery SOC prediction value; In the step of inputting the current, voltage and ambient temperature measurement values of the battery into the Kalman filter for state estimation to obtain a Kalman filter state estimation value, the step specifically comprises: predicting the current and voltage measurement values of the battery through a state prediction equation and a covariance prediction equation, adjusting the prediction values through a Kalman gain adjustment equation, and correcting the adjusted values through a state update equation and a covariance update equation to obtain the Kalman filter state estimation value. The state prediction equation, the covariance prediction equation, the Kalman gain adjustment equation, the state update equation and the covariance update equation are derived from the state space equation of the battery.

4. An electronic device, comprising: The computer readable storage medium stores at least one instruction, and the at least one instruction is executed by the processor to implement the battery SOC prediction method based on digital twinning as claimed in any one of claims 1 to 2.

5. A computer readable storage medium, characterized in that, The computer readable storage medium stores at least one instruction, and the at least one instruction is executed by the processor to implement the battery SOC prediction method based on digital twinning as claimed in any one of claims 1 to 2.

Images