A method and system for estimating battery state of charge in the presence of missing and non-gaussian noise

Abstract

The application provides a kind of battery state of charge estimation method and system under the condition of missing and non-gaussian noise measurement, it is related to battery management technical field, the method comprises: obtaining battery current and terminal voltage sampling data;Based on the equivalent circuit model of battery, state space modeling is carried out, and the unified measurement equation containing Bernoulli random variable is obtained to represent the state of measurement data packet arrival;When data arrives, the actual terminal voltage sample value is used to perform measurement update, when data is lost, the virtual measurement value is generated based on the prior estimate value of battery polarization voltage to replace the actual terminal voltage sample value to perform measurement update;The battery state of charge estimation value is determined according to the measurement update result.The application introduces virtual measurement compensation mechanism, solves the problem of SOC estimation interruption and error accumulation when measurement data is randomly lost, maintains the continuity of estimation process, and significantly improves the accuracy and robustness of battery state of charge estimation under the condition of packet loss.

Description

Technical Field

[0001] This invention relates to the field of battery management technology, and in particular to a method and system for estimating the state of charge of a battery under measurement loss and non-Gaussian noise. Background Technology

[0002] State of Charge (SOC) represents the remaining capacity of a battery and is a core monitoring parameter of the Battery Management System (BMS). Since SOC cannot be directly measured, it is typically estimated indirectly based on available voltage and current data, using a battery equivalent circuit model and filtering algorithms. Current mainstream model-based estimation schemes, such as extended Kalman filtering and volumetric Kalman filtering, heavily rely on continuous and complete sensor sampling data. However, in practical automotive or energy storage applications, sensor measurement data is prone to random loss during transmission due to factors such as electromagnetic interference, vibration, or network communication limitations. When measurement data is missing, traditional filtering algorithms, unable to obtain effective terminal voltage measurements, are forced to interrupt the measurement update process. This causes SOC estimation errors to accumulate rapidly during the data loss period, severely impacting the accuracy of the BMS's assessment of remaining capacity and the system's reliability. Summary of the Invention

[0003] To address the aforementioned technical problems in the prior art, this invention provides a method and system for estimating the state of charge of a battery under measurement loss and non-Gaussian noise conditions.

[0004] The first aspect of the present invention provides a method for estimating the state of charge of a battery under measurement loss and non-Gaussian noise, comprising:

[0005] Acquire battery current sampling data and terminal voltage sampling data;

[0006] Based on the current sampling data and terminal voltage sampling data, the battery is modeled in state space using the battery equivalent circuit model to obtain the state equation and unified measurement equation; wherein, the unified measurement equation contains random variables that follow a Bernoulli distribution to characterize the arrival state of the measurement data packets.

[0007] When the random variable representing the measurement data packet arrives, the measurement update is performed using the actual terminal voltage sample value; when the random variable representing the measurement data packet is lost, a virtual measurement value is generated based on the prior estimate of the battery polarization voltage, and the virtual measurement value is used to replace the actual terminal voltage sample value to perform the measurement update.

[0008] The estimated state of charge of the battery is determined based on the measurement update results.

[0009] Furthermore, after acquiring the battery's current sampling data and terminal voltage sampling data, the process also includes:

[0010] The model parameters of the battery equivalent circuit model are identified online using the event-triggered least squares method with a forgetting factor. The model parameters include ohmic internal resistance, polarization internal resistance, and polarization capacitance.

[0011] When the random variable represents the loss of the measurement data packet, the parameter freezing mechanism is triggered to keep the model parameters at the current time consistent with the model parameters at the previous time.

[0012] When the random variable representing the measurement data packet arrives, the parameter update mechanism is triggered to iteratively update the model parameters based on the sampling data at the current moment.

[0013] Furthermore, determining the estimated state of charge of the battery based on the measurement update results specifically includes:

[0014] The identified model parameters are input into a joint estimator based on the maximum correlation entropy criterion and the square root generalized fifth-order capacitive Kalman filter;

[0015] In the measurement update phase of the joint estimator, a cost function based on the maximum correlation entropy criterion is constructed, and a diagonal weighting matrix is ​​determined based on the cost function;

[0016] The prior state covariance and measurement noise covariance are corrected online using the diagonal weighting matrix to suppress the influence of non-Gaussian noise on the state of charge estimation.

[0017] Output the corrected state of charge estimate.

[0018] Furthermore, the measurement update is performed based on the square root generalized fifth-order capacitive Kalman filter algorithm, wherein the set of capacitive points... and corresponding weighting coefficients It is generated according to the generalized higher-order spherical-radial volume rule as follows:

[0019] ;

[0020] ;

[0021] In the formula, This represents the state dimension of the battery system.

[0022] Furthermore, the generation of virtual measurement values ​​based on prior estimates of battery polarization voltage specifically includes:

[0023] The prior estimate of the polarization voltage at the current moment is calculated using the following formula. :

[0024] ;

[0025] The virtual measurement value is synthesized according to the following formula. :

[0026] ;

[0027] in, This is the posterior estimate of the polarization voltage at the previous moment. The current at the previous moment, The current at the current moment, The sampling period is , and These are the polarization internal resistance, polarization capacitance, and ohmic internal resistance when measurement loss occurs. The prior estimate for the current moment. This is the open-circuit voltage fitting function.

[0028] Furthermore, during the measurement update process, the square root matrix of the state covariance is calculated using the following formula. Perform a recursive update:

[0029] ;

[0030] in, This represents the matrix triangular decomposition operation. This is the weighted and centered state volume point deviation matrix. This is the weighted and centered measurement volume point deviation matrix. This is the filter gain matrix at the current time. This is the square root matrix of the noise covariance.

[0031] A second aspect of the present invention provides a battery state-of-charge estimation system under measurement loss and non-Gaussian noise, comprising:

[0032] The data acquisition module is used to acquire the battery's current sampling data and terminal voltage sampling data;

[0033] The state-space modeling module is used to perform state-space modeling of the battery based on the current sampling data and terminal voltage sampling data, using the battery equivalent circuit model to obtain state equations and unified measurement equations; wherein, the unified measurement equations contain random variables that follow a Bernoulli distribution, used to characterize the arrival state of the measurement data packets.

[0034] The measurement update execution module is used to perform measurement update using the actual terminal voltage sample value when the measurement data packet represented by the random variable arrives; and to generate a virtual measurement value based on the prior estimate of the battery polarization voltage when the measurement data packet represented by the random variable is lost, and to perform measurement update by replacing the actual terminal voltage sample value with the virtual measurement value.

[0035] The state of charge determination module is used to determine the estimated state of charge of the battery based on the measurement update results.

[0036] A third aspect of the present invention provides an electronic device including a memory, a processor, and a program stored in the memory and running on the processor, wherein the processor executes the program to implement the steps in the measurement loss and non-Gaussian noise battery state-of-charge estimation method as described in the first aspect of the present invention.

[0037] A fourth aspect of the present invention provides a computer-readable storage medium having a program stored thereon that, when executed by a processor, implements the steps in the method for estimating the state of charge of a battery under measurement loss and non-Gaussian noise as described in the first aspect of the present invention.

[0038] A fifth aspect of the present invention provides a computer program product comprising software code, wherein the program in the software code performs the steps of the method for estimating the state of charge of a battery under measurement loss and non-Gaussian noise as described in the first aspect of the present invention.

[0039] Compared with the prior art, the battery state of charge estimation method and system provided by the present invention under measurement loss and non-Gaussian noise has the following beneficial effects: by introducing stochastic modeling of measurement arrival state and virtual measurement compensation mechanism, when the actual terminal voltage data is unavailable due to communication packet loss, the filtering algorithm can still obtain effective observations to drive the measurement update link, thereby maintaining the continuity of the SOC estimation process, avoiding the divergence of estimation error during data loss, and significantly improving the accuracy and robustness of battery state of charge estimation under measurement packet loss environment. Attached Figure Description

[0040] The accompanying drawings, which form part of this disclosure, are used to provide a further understanding of this disclosure. The illustrative embodiments of this disclosure and their descriptions are used to explain this disclosure and do not constitute an undue limitation of this disclosure.

[0041] Figure 1 This is a flowchart of the method for estimating the state of charge of a battery under measurement loss and non-Gaussian noise provided in Embodiment 1 of the present invention;

[0042] Figure 2 This is a schematic diagram of the Thevenin equivalent circuit model provided in Embodiment 1 of the present invention;

[0043] Figure 3This is a block diagram of the battery state of charge estimation system under measurement loss and non-Gaussian noise provided in Embodiment 2 of the present invention. Detailed Implementation

[0044] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used in this invention have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0045] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, unless the context clearly indicates otherwise, the singular form is intended to include the plural form as well. Furthermore, it should be understood that the terms “comprising” and “having”, and any variations thereof, are intended to cover non-exclusive inclusion, for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0046] Where there is no conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0047] All data acquisition in this embodiment is carried out in accordance with laws and regulations and with user consent, and the data is used legally.

[0048] Example 1

[0049] like Figure 1 This embodiment provides a method for estimating the state of charge of a battery under measurement loss and non-Gaussian noise, including:

[0050] S1. Obtain the battery's current sampling data and terminal voltage sampling data.

[0051] S2. Based on the current sampling data and terminal voltage sampling data, the battery is modeled in state space using the battery equivalent circuit model to obtain the state equation and the unified measurement equation; wherein, the unified measurement equation contains random variables that follow a Bernoulli distribution to characterize the arrival state of the measurement data packets.

[0052] In this step, the Thevenin equivalent circuit model is selected as the basic model, such as... Figure 2 As shown, this model can efficiently characterize the dynamic characteristics of the battery using three lumped parameters: ohmic internal resistance, polarization internal resistance, and polarization capacitance. It can ensure the accuracy of the physical polarization characteristics while maintaining low computational complexity, thus meeting the real-time requirements of BMS for joint estimation of SOC and parameters.

[0053] Based on the Thevenin equivalent circuit model, and according to Kirchhoff's laws, its continuous domain physical equations are expressed as follows:

[0054] polarization voltage With terminal voltage The expression is:

[0055] (1)

[0056] in, and These are the derivatives of the battery polarization voltage and the polarization voltage, respectively. For load current, For ohmic internal resistance, For polarization internal resistance, Polarized capacitor, This refers to the battery terminal voltage. The open-circuit voltage of the battery has a non-linear relationship with the state of charge (SOC), which can be approximated as linear as follows: ,in, Denotes the order of a polynomial. Represents the polynomial fitting coefficients. Represents a variable.

[0057] Continuous transfer function The expression is:

[0058] (2)

[0059] in, The polarization time constant can be expressed as: , The difference between the battery terminal voltage and the open-circuit voltage Domain representation, can be expressed as .

[0060] Discretize the above model to obtain the discrete transfer function. The expression is:

[0061] (3)

[0062] in, The sampling period is Let be the coefficients, satisfying: , , .

[0063] To meet the recursive computation requirements of BMS, based on the Thevenin equivalent circuit model, the discrete state-space expression is obtained as follows:

[0064] (4)

[0065] in, For Coulomb efficiency, The available capacity of the battery. For process noise, This is a nonlinear measurement function.

[0066] Considering the possibility of measurement data loss during communication, a unified measurement equation is constructed as follows:

[0067] (5)

[0068] in, For the observed values, The arrival status of data packets is represented by random variables that follow a Bernoulli distribution. Virtual measurements synthesized based on prior physical information. To observe uncertain noise.

[0069] S3. When the random variable representing the measurement data packet arrives, the measurement update is performed using the actual terminal voltage sample value; when the random variable representing the measurement data packet is lost, a virtual measurement value is generated based on the prior estimate of the battery polarization voltage, and the virtual measurement value is used to replace the actual terminal voltage sample value to perform the measurement update.

[0070] In this step, virtual measurement values The calculation equation is as follows:

[0071] (6)

[0072] in, and To measure the polarization internal resistance and polarization capacitance when lost, These are prior estimates. The open-circuit voltage value fitted based on prior estimates. This is the posterior estimate of the polarization voltage at the previous moment. The current at the previous moment, The current at the current moment, The sampling period is , , These are the polarization internal resistance, polarization capacitance, and ohmic internal resistance when the measurement is lost.

[0073] This step constructs a unified measurement equation containing Bernoulli random variables and uses prior information from the battery physics model to generate virtual measurement values ​​for compensation when measurement data is lost. This allows the measurement update phase of the filtering algorithm to continue running during data loss, thereby maintaining the continuity of the SOC estimation process, effectively suppressing the divergence of estimation errors caused by missing data, and significantly improving the robustness and accuracy of SOC estimation under measurement loss conditions.

[0074] S4. Determine the estimated state of charge of the battery based on the measurement update results.

[0075] S5. After acquiring the battery's current sampling data and terminal voltage sampling data, the model parameters of the battery equivalent circuit model are identified online using the event-triggered least squares method with a forgetting factor, and a parameter freezing mechanism is triggered when measurement is lost.

[0076] To achieve accurate dynamic tracking of lithium-ion battery model parameters that may suffer from measurement loss, this step employs an event-triggered forgetting factor least squares method. The main process is as follows:

[0077] The discrete domain regression equation is as follows:

[0078] (7)

[0079] in, For the input vector, The parameter vector to be identified.

[0080] The forgetting factor equation is:

[0081] (8)

[0082] in, Forgetting factor, When measurement data is lost At this time, the forgetting factor is forcibly set to 1, and a parameter freezing strategy is implemented to maintain the consistency of the model parameters at the current time step with those at the previous time step; when the measurement data is recovered, When the forgetting factor returns to its initial value, the parameter update mechanism is triggered, and the parameters continue to be updated.

[0083] The least squares parameter identification equation for the forgetting factor is:

[0084] (9)

[0085] in, For gain, Let be the error covariance matrix. for The estimated value.

[0086] By solving the parameter mapping relationship, the identified battery model parameters can be obtained, including ohmic internal resistance, polarization internal resistance, and polarization capacitance:

[0087] (10)

[0088] in, To identify the ohmic internal resistance, To identify polarization resistance, To identify polarized capacitors.

[0089] This step associates the arrival status of measurement data with the forgetting factor. When data is lost, the forgetting factor is forced to 1 to freeze the parameters, thus avoiding incorrect model parameter updates due to the lack of effective terminal voltage information. When the data is recovered, the forgetting factor is restored to its initial value and continues to recursively identify the parameters. This enables stable online tracking of battery model parameters in the case of packet loss, further ensuring the accuracy of SOC estimation.

[0090] S6. Input the identified model parameters into the joint estimator based on the maximum correlation entropy criterion and the square root generalized fifth-order capacitive Kalman filter, perform measurement updates, and output the corrected state of charge estimate.

[0091] To suppress the impact of non-Gaussian noise and model linearization error on SOC estimation based on the battery model parameter identification results, this step employs a joint estimation algorithm combining the maximum correlation entropy criterion and the square root generalized fifth-order capacitive Kalman filter. Consider a discrete nonlinear system of the following form:

[0092] (11)

[0093] in, For state vectors, For measurement vectors, For the input vector, and Given a nonlinear function; For process noise, To measure noise, and Let be the process noise covariance and the measurement noise covariance, and assume they are independent of each other.

[0094] First, the volume points and weights for the square root generalized fifth-order capacitive Kalman filter are generated. Consider the real number domain. The integral over the given area is calculated using the generalized fifth-order nadir integral:

[0095] (12)

[0096] in, , , They are , , The weights corresponding to the trajectory, For the system state dimension, calculate according to equation (12) Integrating these equations yields the system of moment matching equations:

[0097] (13)

[0098] in,

[0099] (16)

[0100] (17)

[0101] (18)

[0102] (19)

[0103] Solving the above system of equations yields a unique solution:

[0104] (20)

[0105] Transforming the generalized fifth-order ductile integral formula into a standard Gaussian distribution, we obtain:

[0106] (twenty one)

[0107] Corresponding volume point and weight They are respectively:

[0108] (twenty two)

[0109] (twenty three)

[0110] Next, the time update is performed using the square root generalized fifth-order capacitive Kalman filter. Assume the posterior state estimate at time k-1 is... The corresponding square root matrix of state covariance is ,satisfy , .

[0111] Generate volume points:

[0112] (twenty four)

[0113] Propagation volume point:

[0114] (25)

[0115] Predicted state:

[0116] (26)

[0117] Predicted state covariance square root matrix:

[0118] (27)

[0119] in, , , Let be the prior state covariance matrix, and let the weighted centering matrix be:

[0120] (28)

[0121] Then, the measurement update is performed using the square root generalized fifth-order capacitive Kalman filter.

[0122] Generate volume points:

[0123] (29)

[0124] It is based on the current prior mean The scalar value is calculated in this iteration as shown in (6). The volume point equation for the propagation of the measurement equation is:

[0125] (30)

[0126] Predicted measurements:

[0127] (31)

[0128] Predicted measurement covariance square root matrix:

[0129] (32)

[0130] in, , , Let the measurement covariance matrix be given, and the weighted centering matrix be:

[0131] (33)

[0132] Cross-covariance matrix:

[0133] (34)

[0134] Kalman gain:

[0135] (35)

[0136] Status Update:

[0137] (36)

[0138] Square root update of the state covariance matrix:

[0139] (37)

[0140] Finally, the maximum correlation entropy criterion is introduced to suppress the influence of non-Gaussian noise. As a measure of local similarity, the maximum correlation entropy criterion can effectively handle heavy-tailed distributions or impulse noise. For any two random variables... and The joint distribution function is The relevant entropy is defined as:

[0141] (38)

[0142] in, Represents the expectation operator. The Mercel kernel is translation-invariant and is typically described by a Gaussian kernel:

[0143] (39)

[0144] in, For kernel bandwidth, the error is .because For unknowns, sample approximations can be used:

[0145] (40)

[0146] in, , From Extracted from One sample. When hour, It can achieve its maximum value. Therefore, a Taylor expansion of the Gaussian kernel is performed:

[0147] (41)

[0148] Where X and Y represent the measured observations and the model-predicted observations, respectively; x and y are the single values ​​(sample values) at their corresponding times.

[0149] It can be seen that the correlation entropy can be regarded as the weighted sum of all even-order moments of two random variables, and the higher-order moment information can be adjusted by the kernel bandwidth parameter.

[0150] Equation (11) can be transformed into the following form:

[0151] (42)

[0152] in, , The covariance matrix is:

[0153] (43)

[0154] in, , , Each by , , The Cholesky decomposition is performed to obtain the result. Multiply both sides of equation (42) by the left side. We can obtain:

[0155] (44)

[0156] in, , , .

[0157] Define a cost function based on the MCC criterion as follows:

[0158] (45)

[0159] In the formula, for The One element, for The OK, and These are the state values. and measurement values The dimension. Under MCC The solution can be expressed as follows:

[0160] (46)

[0161] The optimal estimate can be obtained by solving for the extreme value, let ,available The optimal solution:

[0162] (47)

[0163] definition We can obtain:

[0164] (48)

[0165] in,

[0166] (49)

[0167] Equation (47) can be further written as:

[0168] (50)

[0169] Therefore, the prior state covariance and the measurement noise covariance are:

[0170] (51)

[0171] (52)

[0172] The modified covariance matrix is ​​substituted into the measurement update process of the square root generalized fifth-order capacitive Kalman filter, replacing the original... and This enables robust estimation of the State of Charge (SOC) in non-Gaussian noise environments, ultimately outputting a corrected estimate of the state of charge.

[0173] In this step, the square root generalized fifth-order capacitive Kalman filter uses the generalized fifth-order spherical-radial capacitive rule to generate capacitive points and weights. Compared with the traditional third-order capacitive Kalman filter, its high-order moment matching characteristics can more accurately approximate the strong nonlinear characteristics of the battery, providing a more accurate numerical integration basis for SOC estimation. At the same time, the square root recursive update mechanism ensures the symmetric positive definiteness of the covariance matrix, improving the numerical stability of the filter.

[0174] Building upon this foundation, a maximum correlation entropy criterion is introduced to construct a cost function and derive a diagonal weighting matrix, enabling online correction of the prior state covariance and measurement noise covariance. Its beneficial effects are as follows: the maximum correlation entropy criterion can effectively capture the higher-order moment information of the error signal; in non-Gaussian noise environments (such as heavy-tailed noise or impulse noise), it adaptively reduces the weight of anomalous innovations, thereby suppressing the interference of non-Gaussian noise on state estimation. This overcomes the limitation of traditional Kalman filtering, which is only applicable to Gaussian white noise, and significantly improves the reliability of SOC estimation for battery management systems in complex electromagnetic environments.

[0175] Example 2

[0176] like Figure 3 This embodiment provides a battery state of charge estimation system under measurement loss and non-Gaussian noise conditions, including:

[0177] The data acquisition module is used to acquire the battery's current sampling data and terminal voltage sampling data;

[0178] The state-space modeling module is used to perform state-space modeling of the battery based on the current sampling data and terminal voltage sampling data, using the battery equivalent circuit model to obtain state equations and unified measurement equations; wherein, the unified measurement equations contain random variables that follow a Bernoulli distribution, used to characterize the arrival state of the measurement data packets.

[0179] The measurement update execution module is used to perform measurement update using the actual terminal voltage sample value when the measurement data packet represented by the random variable arrives; and to generate a virtual measurement value based on the prior estimate of the battery polarization voltage when the measurement data packet represented by the random variable is lost, and to perform measurement update by replacing the actual terminal voltage sample value with the virtual measurement value.

[0180] The state of charge determination module is used to determine the estimated state of charge of the battery based on the measurement update results.

[0181] The above modules work together to achieve high-precision prediction and low-cost scheduling of vehicle supply and demand in the Internet of Vehicles environment.

[0182] Example 3

[0183] Embodiment 3 of the present invention provides an electronic device.

[0184] An electronic device includes a memory, a processor, and a program stored in the memory and running on the processor. The processor includes, but is not limited to, at least one of a central processing unit (CPU), a graphics processing unit (GPU), a neural network processor (NPU), a tensor processor (TPU), or an artificial intelligence acceleration chip. The program is used to implement the steps in the method for estimating the state of charge of a battery under measurement loss and non-Gaussian noise as described in Embodiment 1 of the present invention when executing the program.

[0185] The detailed steps are the same as those provided in Example 1 for estimating the state of charge of a battery under measurement loss and non-Gaussian noise, and will not be repeated here.

[0186] Example 4

[0187] Embodiment 4 of the present invention provides a computer-readable storage medium.

[0188] A computer-readable storage medium having a program stored thereon, which, when executed by a processor, implements the steps in the method for estimating the state of charge of a battery under measurement loss and non-Gaussian noise as described in Embodiment 1 of the present invention.

[0189] The detailed steps are the same as those provided in Example 1 for estimating the state of charge of a battery under measurement loss and non-Gaussian noise, and will not be repeated here.

[0190] Example 5

[0191] Embodiment 5 of the present invention provides a computer program product.

[0192] A computer program product includes software code, wherein the program in the software code performs the steps in the method for estimating the state of charge of a battery under measurement loss and non-Gaussian noise as described in Embodiment 1 of the present invention.

[0193] The detailed steps are the same as those provided in Example 1 for estimating the state of charge of a battery under measurement loss and non-Gaussian noise, and will not be repeated here.

[0194] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention 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. The solutions in the embodiments of the present invention can be implemented using various computer languages. For example, in one implementation, the methods and systems can be developed based on deep learning frameworks (such as TensorFlow, PyTorch, etc.) and using the Python language. Those skilled in the art will understand that other suitable programming languages ​​or tools can also be used for implementation without departing from the core ideas of the present invention.

[0195] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0196] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0197] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0198] The above description is merely a preferred embodiment of this practice and is not intended to limit the scope of this practice. Various modifications and variations can be made to this practice by those skilled in the art. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of this practice should be included within the protection scope of this practice.

Claims

1. A method for estimating the state of charge of a battery under measurement loss and non-Gaussian noise, characterized in that, include: Acquire battery current sampling data and terminal voltage sampling data; Based on the current sampling data and terminal voltage sampling data, the battery is modeled in state space using the battery equivalent circuit model to obtain the state equation and unified measurement equation; wherein, the unified measurement equation contains random variables that follow a Bernoulli distribution to characterize the arrival state of the measurement data packets. When the measurement data packet, represented by the random variable, arrives, a measurement update is performed using the actual terminal voltage sample value. When the measurement data packet, represented by the random variable, is lost, a virtual measurement value is generated based on the prior estimate of the battery polarization voltage, and this virtual measurement value replaces the actual terminal voltage sample value to perform a measurement update. The measurement update is performed based on the square root generalized fifth-order capacitive Kalman filter algorithm, where the set of capacitive points... and corresponding weighting coefficients It is generated according to the generalized higher-order spherical-radial volume rule as follows: ； ； In the formula, This refers to the state dimension of the battery system. The generation of virtual measurements based on prior estimates of battery polarization voltage specifically includes: The prior estimate of the polarization voltage at the current moment is calculated using the following formula. : ； The virtual measurement value is synthesized according to the following formula. : ； in, This is the posterior estimate of the polarization voltage at the previous moment. The current at the previous moment, The current at the current moment, The sampling period is , and These are the polarization internal resistance, polarization capacitance, and ohmic internal resistance when measurement loss occurs. The prior estimate for the current moment. This is the open-circuit voltage fitting function; The process of determining the battery's state of charge (SOC) estimate based on measurement update results includes: inputting the identified model parameters into a joint estimator based on the maximum correlation entropy criterion and the square root generalized fifth-order capacitive Kalman filter; constructing a cost function based on the maximum correlation entropy criterion during the measurement update phase of the joint estimator, and determining a diagonal weighting matrix based on the cost function; using the diagonal weighting matrix to perform online correction on the prior state covariance and measurement noise covariance to suppress the influence of non-Gaussian noise on the SOC estimate; and outputting the corrected SOC estimate.

2. The method according to claim 1, characterized in that, After acquiring the battery's current sampling data and terminal voltage sampling data, the following is also included: The model parameters of the battery equivalent circuit model are identified online using the event-triggered least squares method with a forgetting factor. The model parameters include ohmic internal resistance, polarization internal resistance, and polarization capacitance. When the random variable represents the loss of the measurement data packet, the parameter freezing mechanism is triggered to keep the model parameters at the current time consistent with the model parameters at the previous time. When the random variable representing the measurement data packet arrives, the parameter update mechanism is triggered to iteratively update the model parameters based on the sampling data at the current moment.

3. The method according to claim 1, characterized in that, During the measurement update process, the square root matrix of the state covariance is calculated using the following formula. Perform a recursive update: ； in, This represents the matrix triangular decomposition operation. This is the weighted and centered state volume point deviation matrix. This is the weighted and centered measurement volume point deviation matrix. This is the filter gain matrix at the current time. This is the square root matrix of the noise covariance.

4. A battery state-of-charge estimation system under measurement loss and non-Gaussian noise, characterized in that, include: The data acquisition module is used to acquire the battery's current sampling data and terminal voltage sampling data; The state-space modeling module is used to perform state-space modeling of the battery based on the current sampling data and terminal voltage sampling data, using the battery equivalent circuit model to obtain state equations and unified measurement equations; wherein, the unified measurement equations contain random variables that follow a Bernoulli distribution, used to characterize the arrival state of the measurement data packets. The measurement update execution module is used to perform measurement updates using the actual terminal voltage sample value when the measurement data packet represented by the random variable arrives; and to generate a virtual measurement value based on a priori estimate of the battery polarization voltage when the measurement data packet represented by the random variable is lost, and to replace the actual terminal voltage sample value with the virtual measurement value to perform measurement updates; the measurement update is performed based on the square root generalized fifth-order capacitive Kalman filter algorithm, wherein the set of capacitive points... and corresponding weighting coefficients It is generated according to the generalized higher-order spherical-radial volume rule as follows: ； ； In the formula, This refers to the state dimension of the battery system. The generation of virtual measurements based on prior estimates of battery polarization voltage specifically includes: The prior estimate of the polarization voltage at the current moment is calculated using the following formula. : ； The virtual measurement value is synthesized according to the following formula. : ； in, This is the posterior estimate of the polarization voltage at the previous moment. The current at the previous moment, The current at the current moment, The sampling period is , and These are the polarization internal resistance, polarization capacitance, and ohmic internal resistance when measurement loss occurs. The prior estimate for the current moment. This is the open-circuit voltage fitting function; The state of charge (SOC) determination module is used to determine the estimated SOC of the battery based on measurement update results. Specifically, it includes: inputting the identified model parameters into a joint estimator based on the maximum correlation entropy criterion and the square root generalized fifth-order capacitive Kalman filter; constructing a cost function based on the maximum correlation entropy criterion during the measurement update phase of the joint estimator, and determining a diagonal weighting matrix based on the cost function; using the diagonal weighting matrix to perform online correction on the prior state covariance and measurement noise covariance to suppress the influence of non-Gaussian noise on the SOC estimation; and outputting the corrected SOC estimate.

5. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the program, it implements the steps of the method for estimating the state of charge of a battery under measurement loss and non-Gaussian noise as described in any one of claims 1 to 3.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps of the method for estimating the state of charge of a battery under measurement loss and non-Gaussian noise as described in any one of claims 1 to 3.

7. A computer program product, comprising software code, characterized in that, The program in the software code executes the steps of the method for estimating the state of charge of a battery under measurement loss and non-Gaussian noise as described in any one of claims 1 to 3.

Images