A model-based approach to sensor fault detection in battery management systems

Abstract

The present invention relates to the field of battery management technology, and specifically to a model-based battery management system sensor fault detection method, comprising: establishing a second-order RC equivalent circuit model with a hysteresis effect as a lithium battery equivalent circuit model; utilizing an extended Kalman filter algorithm based on aging state correction to estimate the SOC value at each moment, and determine the estimated capacity and capacity residual at each moment; utilizing a power integration method to estimate the SOE value at each moment, and determine the estimated energy and energy residual at each moment; comparing the capacity residual at each moment with a preset capacity threshold, and comparing the energy residual at each moment with a preset energy threshold, to determine which sensor has failed. The present invention considers the battery aging state, adds a forgetting parameter to correct the Kalman gain matrix, and improves the accuracy of the estimated SOC value; based on the SOC and SOE indicators, the estimated capacity and estimated energy are obtained, thereby improving the accuracy of sensor fault detection.

Description

Technical Field

[0001] The present invention relates to the technical field of battery management, and in particular to a model-based sensor fault detection method for a battery management system. Background Art

[0002] In recent years, the technology field related to electric vehicles has become a hot topic. Since power batteries are an indispensable part of electric vehicles, lithium batteries, as the main component of power batteries, have also received widespread attention in the market. Compared with similar products, lithium batteries have greater energy density. In addition, lithium batteries have inherent advantages such as high power, long life, and low maintenance costs.

[0003] The battery management system (BMS) in power batteries is crucial to ensuring the reliability, efficiency and life of lithium batteries. It is an important component to ensure the performance and safety of power batteries. Among them, the state of charge (SOC), state of health (SOH) and state of energy (SOE) of lithium batteries are key functions of BMS and need to be monitored online. Once an unchecked fault occurs in a lithium battery, it may cause irreversible and serious safety hazards, and even cause catastrophic damage under extreme conditions. Therefore, it is necessary to determine whether the lithium battery has a fault based on real-time measurement data of the lithium battery's state of charge (SOC), state of health (SOH) and state of energy (SOE). However, since the state estimation and fault detection of lithium batteries can only be performed based on the actual measured data of the battery management system sensors, it is of great significance to study the fault detection of battery management system sensors in electric vehicles.

[0004] The existing model-based fault diagnosis method for battery management system sensors has the following defects: (1) It does not take into account the hysteresis phenomenon in the battery charging and discharging process, resulting in low accuracy of the battery equivalent circuit model, which in turn leads to low accuracy of parameter identification and a problem of large terminal voltage estimation value; (2) In the existing technology, SOC is often used as an indicator for battery management system sensor fault detection. This indicator performs well in the experimental process, but in actual applications, only considering a single SOC indicator to detect battery management system sensor faults will result in low fault detection accuracy; at the same time, since lithium batteries are affected by the aging process and the external environment in actual applications, the total capacity will fluctuate, resulting in low accuracy of SOC indicator estimation, resulting in untimely and inaccurate fault detection, thereby affecting the battery life. Summary of the Invention

[0005] To this end, the technical problem to be solved by the present invention is to overcome the problem that the existing technology does not take into account the hysteresis phenomenon in the battery charging and discharging process, resulting in low accuracy of the battery equivalent circuit model, which in turn leads to low accuracy of parameter identification and a large terminal voltage estimation value; only using the SOC indicator for battery management system sensor fault detection leads to low fault detection accuracy, and the existing estimation accuracy of the SOC indicator is not high, resulting in untimely and inaccurate fault detection, thereby affecting the battery service life.

[0006] To solve the above technical problems, the present invention provides a model-based battery management system sensor fault detection method, comprising:

[0007] A second-order RC equivalent circuit model with hysteresis effect is established as the lithium battery equivalent circuit model, and the discrete state equation of the lithium battery equivalent circuit model is determined; based on the forgetting factor recursive least squares method, the parameters of the lithium battery equivalent circuit model are identified to obtain the parameter values of each component in the lithium battery equivalent circuit model, and based on the discrete state equation of the lithium battery equivalent circuit model, the terminal voltage value, the first polarization voltage value, and the second polarization voltage value at each moment are obtained;

[0008] The total capacity of the lithium battery is calibrated to obtain the actual capacity of the lithium battery; based on the lithium battery equivalent circuit model, the state of charge, the first polarization voltage, and the second polarization voltage are used as the first state variable, and combined with the actual capacity of the lithium battery, a lithium battery state space equation based on the SOC is obtained, and the extended Kalman filter algorithm is used to sequentially obtain the estimated value of the first state variable at each moment, thereby sequentially obtaining the estimated SOC value at each moment; wherein, the Kalman gain matrix at each moment is corrected based on the forgetting parameter to obtain the target Kalman gain matrix at each moment;

[0009] Based on the lithium battery equivalent circuit model, the energy state, the first polarization voltage, and the second polarization voltage are used as the second state variable. Combined with the actual capacity of the lithium battery, the lithium battery state space equation based on SOE is determined. The power integration method is used to sequentially obtain the estimated value of the second state variable at each moment, thereby sequentially obtaining the estimated value of SOE at each moment.

[0010] Determine the estimated capacity at the current moment based on the actual capacity of the lithium battery, the actual discharge time, the current value at the current moment, and the estimated SOC value at the current moment; obtain the capacity residual at the current moment based on the difference between the actual capacity of the lithium battery and the estimated capacity at the current moment; determine the reference energy of the lithium battery based on the rated voltage and actual capacity of the lithium battery; determine the estimated energy at the current moment based on the reference energy of the lithium battery, the actual discharge time, the current value at the current moment, the terminal voltage value at the current moment, and the estimated SOE value at the current moment; obtain the energy residual at the current moment based on the difference between the reference energy of the lithium battery and the estimated energy at the current moment;

[0011] If the capacity residual at the current moment exceeds the preset capacity threshold, the current sensor in the battery management system fails; if the capacity residual at the current moment does not exceed the preset capacity threshold, the current sensor in the battery management system does not fail; if the energy residual at the current moment exceeds the preset energy threshold, the voltage sensor in the battery management system fails; if the energy residual does not exceed the preset energy threshold, the voltage sensor in the battery management system does not fail.

[0012] Preferably, based on a second-order RC equivalent circuit model with hysteresis effect, the circuit equation of the lithium battery equivalent circuit model is determined as:

[0013]

[0014] U d =U oc -U1-U2-IR0-U h

[0015] Among them, U d Indicates terminal voltage; U oc Indicates the open circuit voltage; U1 indicates the first polarization voltage, that is, the voltage across the first polarization capacitor; U2 indicates the second polarization voltage, that is, the voltage across the second polarization capacitor; U h represents the hysteresis voltage; C1 represents the parameter value of the first polarization capacitor; C2 represents the parameter value of the second polarization capacitor; R1 represents the parameter value of the first polarization resistor; R2 represents the parameter value of the second polarization resistor; R0 represents the parameter value of the lithium battery internal resistance; I represents the current; t represents the time;

[0016] Set the sampling period to T, and the discrete state equation of the lithium battery equivalent circuit model is:

[0017]

[0018]

[0019] U h (t+1)=e -|σI(t)|TU h (t)+(1-e -|σI(t)|T )U H

[0020] Wherein, U1(t+1) represents the first polarization voltage value at time (t+1); U1(t) represents the first polarization voltage value at time t; U2(t+1) represents the second polarization voltage value at time (t+1); U2(t) represents the second polarization voltage value at time t; I(t) represents the current value at time t; σ represents the attenuation factor; U h (t+1) represents the hysteresis voltage value at time (t+1); U h (t) represents the hysteresis voltage value at time t; U H Indicates the maximum hysteresis voltage value.

[0021] Preferably, the total capacity of the lithium battery is calibrated to obtain the actual capacity of the lithium battery, which is expressed as:

[0022]

[0023] Among them, Q N Indicates the actual capacity of the lithium battery; Q α Indicates the lithium battery capacity at the start of capacity calibration; Q β Indicates the lithium battery capacity at the end of capacity calibration; t α Indicates the time when capacity calibration starts; t β Indicates the end time of capacity calibration; SOC α Indicates the estimated SOC value at the start of capacity calibration; SOC β It represents the estimated value at the end of capacity calibration; I(t) represents the current value at time t; and t represents time.

[0024] Preferably, based on the lithium battery equivalent circuit model, the state of charge, the first polarization voltage, and the second polarization voltage are used as the first state variable, and combined with the actual capacity of the lithium battery, a lithium battery state space equation based on SOC is obtained, which is expressed as follows:

[0025]

[0026]

[0027] Among them, the first state variable at time t is SOC(t+1) represents the estimated SOC value at time (t+1); SOC(t) represents the estimated SOC value at time t; U1(t+1) represents the first polarization voltage value at time (t+1); U1(t) represents the first polarization voltage value at time t; U2(t+1) represents the second polarization voltage value at time (t+1); U2(t) represents the second polarization voltage value at time t; I(t) represents the current value at time t and serves as the input quantity at time t; U d (t) represents the terminal voltage value at time t and is used as the output at time t; g1(·) represents the first measurement function; U h (t) represents the hysteresis voltage value at time t; C1 represents the parameter value of the first polarization capacitor; C2 represents the parameter value of the second polarization capacitor; R1 represents the parameter value of the first polarization resistor; R2 represents the parameter value of the second polarization resistor; R0 represents the parameter value of the lithium battery internal resistance; T represents the sampling period; w t represents the process noise at time t; v t represents the measurement noise at time t.

[0028] Preferably, the extended Kalman filter algorithm is used to sequentially obtain an estimated value of the first state variable at each moment, thereby sequentially obtaining an estimated SOC value at each moment; wherein, the Kalman gain matrix at each moment is corrected based on the forgetting parameter to obtain a target Kalman gain matrix at each moment, including:

[0029] According to the lithium battery state space equation based on SOC, the state equation and measurement equation of the nonlinear system are determined, and their expressions are:

[0030]

[0031] in, represents the first state variable at time t; represents the first state variable at time (t-1); f(·) represents the transfer function; i t-1 Represents the current value at time (t-1) and serves as the input quantity at time (t-1); y t =[U d (t)] represents the output at time t; g1(·) represents the first measurement function; w t represents the process noise at time t; v t represents the measurement noise at time t;

[0032] The recursive process of the extended Kalman filter algorithm based on aging state correction includes the following steps:

[0033] Step 1: Calculate the estimated value of the first state variable at time t and its corresponding error covariance matrix, which is expressed as:

[0034]

[0035] in, represents the predicted value of the first state variable at time t; represents the estimated value of the first state variable at time (t-1); represents the predicted value of the error covariance matrix at time t; P t-1 represents the estimated value of the error covariance matrix at time (t-1); A represents the state transfer matrix; W t-1 represents the process noise covariance matrix at time (t-1);

[0036] Step 2: Calculate the Kalman gain matrix at time t, which is expressed as:

[0037]

[0038] Among them, K t represents the Kalman gain matrix at time t; C represents the measurement matrix; V t represents the measurement noise covariance matrix at time t;

[0039] The Kalman gain matrix at time t is corrected based on the forgetting parameter to obtain the target Kalman gain matrix at time t, which is expressed as:

[0040]

[0041] Among them, K′ t represents the target Kalman gain matrix at time t; λ represents the forgetting parameter;

[0042] Step 3: According to the target Kalman gain matrix at time t, the estimated value of the first state variable at time t and its corresponding error covariance matrix are corrected. The expression is:

[0043]

[0044] in, represents the estimated value of the first state variable at time t; z k Indicates process quantity; P t represents the estimated value of the error covariance matrix at time t; based on the estimated value of the first state variable at time t, the estimated SOC value at time t is obtained;

[0045] Based on steps 1 to 3, the iterations are continuously repeated until a preset number of iterations is reached, and the estimated value of the first state variable at each moment is obtained in sequence, thereby obtaining the estimated value of the SOC at each moment in sequence.

[0046] Preferably, based on the lithium battery equivalent circuit model, the energy state, the first polarization voltage, and the second polarization voltage are used as the second state variable, the current is used as the input variable, and the terminal voltage is used as the output variable. In combination with the actual capacity of the lithium battery, the lithium battery state space equation based on SOE is determined, and its expression is:

[0047]

[0048] Among them, the second state variable at time t SOE(t+1) represents the estimated SOE value at time (t+1); SOE(t) represents the estimated SOE value at time t; U1(t+1) represents the first polarization voltage value at time (t+1); U1(t) represents the first polarization voltage value at time t; U2(t+1) represents the second polarization voltage value at time (t+1); U2(t) represents the second polarization voltage value at time t; I(t) represents the current value at time t and serves as the input quantity at time t; U d (t) represents the terminal voltage value at time t and is used as the output at time t; g2(·) represents the second measurement function; U h (t) represents the hysteresis voltage value at time t; C1 represents the parameter value of the first polarization capacitor; C2 represents the parameter value of the second polarization capacitor; R1 represents the parameter value of the first polarization resistor; R2 represents the parameter value of the second polarization resistor; R0 represents the parameter value of the lithium battery internal resistance; T represents the sampling period; w t represents the process noise at time t; v t represents the measurement noise at time t.

[0049] Preferably, determining the estimated capacity at the current moment according to the actual capacity of the lithium battery, the actual discharge duration, the current value at the current moment, and the SOC estimation value at the current moment; and obtaining the capacity residual at the current moment according to the difference between the actual capacity of the lithium battery and the estimated capacity at the current moment includes:

[0050] Determine the estimated capacity at time t based on the actual capacity of the lithium battery, the actual discharge time, the current at time t, and the estimated SOC value at time t Its expression is:

[0051]

[0052] Among them, Q N Indicates the actual capacity of the lithium battery; M indicates the actual discharge time of the lithium battery; I t Indicates the current at time t; SOC t represents the estimated SOC value at time t;

[0053] According to the actual capacity of the lithium battery and the estimated capacity at time t The difference between them determines the capacity residual at time t Its expression is:

[0054] Preferably, determining the reference energy of the lithium battery according to the rated voltage and actual capacity of the lithium battery; determining the estimated energy at the current moment according to the reference energy, actual discharge time, current value, terminal voltage value at the current moment, and SOE estimated value at the current moment; and obtaining the energy residual at the current moment according to the difference between the reference energy of the lithium battery and the estimated energy at the current moment include:

[0055] Determine the reference energy D of the lithium battery based on the rated voltage and actual capacity of the lithium battery. N , its expression is: D N =Q N ·U N ;

[0056] Among them, Q N Indicates the actual capacity of the lithium battery; U N Indicates the rated voltage of the lithium battery;

[0057] Determine the estimated energy at time t based on the reference energy of the lithium battery, the actual discharge time, the current at time t, the terminal voltage at time t, and the estimated SOE value at time t. Its expression is:

[0058]

[0059] Where M represents the actual discharge time of the lithium battery; I t represents the current at time t; Represents the terminal voltage at time t; SOE t represents the estimated value of SOE at time t;

[0060] According to the reference energy of the lithium battery and the estimated energy at time t The difference between them determines the energy residual at time t Its expression is:

[0061] Preferably, the parameter identification of the lithium battery equivalent circuit model is performed based on the forgetting factor recursive least squares method to obtain the parameter values of each component in the lithium battery equivalent circuit model, which includes:

[0062] S21: Based on the circuit equation of the lithium battery equivalent circuit model, determine the differential equation after discretization of the lithium battery equivalent circuit model, which is expressed as:

[0063] Y(k)=U oc (k)-U s (k)

[0064]

[0065] θ=[θ1 θ2 θ3 θ4 θ5] T

[0066] Where Y(k) represents the output of the k-th sampling lithium battery equivalent circuit model, Y(k) = U d (k), U d (k) represents the terminal voltage value of the k-th sampling lithium battery equivalent circuit model; Represents the input quantity of the k-th sampling lithium battery equivalent circuit model; U oc (k), U oc (k-1), U oc (k-2) represents the open circuit voltage value of the k-th, (k-1)-th, and (k-2)-th sampling lithium battery equivalent circuit model; U s (k), U s (k-1), U s (k-2) represents the voltage difference of the k-th, (k-1)-th, and (k-2)-th sampling lithium battery equivalent circuit model, U s (k)=U1(k)+U2(k)+U h (k); U1(k) represents the first polarization voltage value of the k-th sampling lithium battery equivalent circuit model; U2(k) represents the second polarization voltage value of the k-th sampling lithium battery equivalent circuit model; U h (k) represents the hysteresis voltage value of the k-th sampling lithium battery equivalent circuit model; I(k), I(k-1), and I(k-2) represent the current values of the k-th, (k-1)-th, and (k-2)-th sampling lithium battery equivalent circuit model, respectively; θ represents the parameter vector of the lithium battery equivalent circuit model;

[0067] S22: Based on the forgetting factor recursive least squares method, perform parameter identification on the lithium battery equivalent circuit model. The specific steps include:

[0068] S221: Based on the differential equation after discretization of the lithium battery equivalent circuit model, determine the least squares equation of the lithium battery equivalent circuit model, which is expressed as:

[0069]

[0070] Wherein, Y(k) represents the output of the k-th sampling lithium battery equivalent circuit model; F(k) represents the expected error of the k-th sampling lithium battery equivalent circuit model; represents the input quantity of the k-th sampling lithium battery equivalent circuit model; θ(k) represents the parameter vector of the k-th sampling lithium battery equivalent circuit model; Represents the estimated value of the parameter vector of the (k-1)th sampling lithium battery equivalent circuit model;

[0071] S222: Determine the initial value of the parameter vector of the lithium battery equivalent circuit model and the initial value of the covariance matrix of the lithium battery equivalent circuit model;

[0072] S223: Calculating the estimated error of the k-th sampling lithium battery equivalent circuit model;

[0073] S224: Calculate the algorithm gain of the k-th sampling, the estimated value of the parameter vector of the k-th sampling lithium battery equivalent circuit model, and the covariance matrix of the k-th sampling lithium battery equivalent circuit model, and their expressions are:

[0074]

[0075] Where G(k) represents the algorithm gain of the kth sampling; Represents the estimated value of the parameter vector of the k-th sampling lithium battery equivalent circuit model; Indicates the The estimated value of the parameter vector of the k-th sampling lithium battery equivalent circuit model; γ represents the forgetting factor; H(k) represents the covariance matrix of the k-th sampling lithium battery equivalent circuit model; H(k-1) represents the covariance matrix of the (k-1)-th sampling lithium battery equivalent circuit model; E represents the identity matrix;

[0076] S225: The number of cycles is increased by 1, and steps 3 to 5 are executed repeatedly until the preset number of cycles is reached, thereby obtaining the optimal estimated value of each directly identified parameter in the parameter vector of the lithium battery equivalent circuit model;

[0077] S226: Based on the optimal estimated values of each directly identified parameter in the parameter vector of the lithium battery equivalent circuit model, the intermediate parameters in the intermediate parameter set S are expressed as:

[0078]

[0079] S2=S1·θ2

[0080]

[0081]

[0082] Wherein, S1, S2, S3, S4, and S5 represent the first intermediate parameter value, the second intermediate parameter value, the third intermediate parameter value, the fourth intermediate parameter value, and the fifth intermediate parameter value, respectively; θ1, θ2, θ3, θ4, and θ5 represent the optimal estimated value of the first direct identification parameter, the optimal estimated value of the second direct identification parameter, the optimal estimated value of the third direct identification parameter, the optimal estimated value of the fourth direct identification parameter, and the optimal estimated value of the fifth direct identification parameter, respectively; T represents the sampling period;

[0083] S227: Based on the optimal estimated values of the parameters in the parameter vector of the lithium battery equivalent circuit model and the intermediate parameters in the intermediate parameter set S, the parameter values of the components in the lithium battery equivalent circuit model are obtained, which are:

[0084]

[0085] Among them, C1 represents the parameter value of the first polarization capacitor; C2 represents the parameter value of the second polarization capacitor; R1 represents the parameter value of the first polarization resistor; R2 represents the parameter value of the second polarization resistor; R0 represents the parameter value of the internal resistance of the lithium battery; τ1 and τ2 represent the first intermediate variable and the second intermediate variable respectively.

[0086] Preferably, the preset capacity threshold is ±20% of the actual capacity of the lithium battery; and the preset energy threshold is ±20% of the reference energy of the lithium battery.

[0087] The above technical solution of the present invention has the following beneficial effects compared with the prior art:

[0088] The present invention discloses a model-based battery management system sensor fault detection method. By considering the hysteresis phenomenon in the charging and discharging process of the lithium battery, a more accurate second-order RC equivalent circuit model with hysteresis effect is established as the lithium battery equivalent circuit model, laying the foundation for accurate data testing. Based on the forgetting factor recursive least squares method, the lithium battery equivalent circuit model is parameter identified to obtain the parameter values of all components in the lithium battery equivalent circuit model, and then the terminal voltage value at each moment in the lithium battery equivalent circuit model is obtained, thereby improving the accuracy of the terminal voltage estimation. By calibrating the lithium battery capacity, the actual capacity of the lithium battery is obtained, the consumption of the lithium battery capacity is reduced, and thus the error in the SOC and SOE estimation is reduced. With the state of charge (SOC), the first polarization voltage, and the second polarization voltage as the first state variable, a lithium battery state space equation based on SOC is obtained, and the extended Kalman filter algorithm is used to obtain the estimated value of the first state variable at each moment in turn, thereby obtaining the SOC at each moment. Estimated value. In this process, adding a forgetting parameter to correct the Kalman gain matrix can improve the accuracy of the estimated value of the first state variable, thereby improving the accuracy of the SOC value estimation of the battery; using the power integration method, dynamically estimate the SOE value of the battery; define the ratio of the power at different times to the SOC as the estimated capacity, and define the ratio of the energy at different times to the SOE as the estimated energy. Based on the relationship between the difference between the estimated capacity and the reference capacity and the capacity threshold, and the relationship between the difference between the estimated energy and the reference energy and the energy threshold, determine whether the voltage sensor and / or current sensor in the battery management system is faulty. Since the estimated capacity will change with time and current, and the estimated energy will change with time, current and voltage, the estimated capacity and estimated energy are obtained based on the SOC index and the SOE index, which can improve the accuracy of voltage sensor and / or current sensor fault detection, thereby improving the safety and reliability of the battery management system, and thus ensuring the use efficiency and service life of the lithium battery. BRIEF DESCRIPTION OF THE DRAWINGS

[0089] In order to make the content of the present invention more clearly understood, the present invention is further described in detail below based on specific embodiments of the present invention in conjunction with the accompanying drawings, wherein:

[0090] Figure 1 This is a flow chart of a model-based battery management system sensor fault detection method provided by the present invention;

[0091] Figure 2 This is the equivalent circuit model diagram of a lithium battery;

[0092] Figure 3 It is a flow chart of parameter identification of lithium battery equivalent circuit model;

[0093] Figure 4It is a schematic diagram of SOC and SOE estimation;

[0094] Figure 5 It is a schematic diagram of the battery management system sensor fault diagnosis and separation strategy. DETAILED DESCRIPTION

[0095] The present invention will be further described below with reference to the accompanying drawings and specific embodiments so that those skilled in the art can better understand the present invention and implement it. However, the embodiments are not intended to limit the present invention.

[0096] Reference Figure 1 As shown, the flow chart of a model-based battery management system sensor fault detection method provided by the present invention specifically includes:

[0097] S1: Based on the hysteresis phenomenon in the charging and discharging process of lithium batteries, a second-order RC equivalent circuit model with hysteresis effect is established as the lithium battery equivalent circuit model, such as Figure 2 As shown;

[0098] Based on the second-order RC equivalent circuit model with hysteresis effect, the circuit equation of the lithium battery equivalent circuit model is determined as:

[0099]

[0100] U d =U oc -U1-U2-IR0-U h

[0101] Among them, U d Indicates terminal voltage; U oc Indicates the open circuit voltage; U1 indicates the first polarization voltage, that is, the voltage across the first polarization capacitor; U2 indicates the second polarization voltage, that is, the voltage across the second polarization capacitor; U h represents the hysteresis voltage; C1 represents the parameter value of the first polarization capacitor; C2 represents the parameter value of the second polarization capacitor; R1 represents the parameter value of the first polarization resistor; R2 represents the parameter value of the second polarization resistor; R0 represents the parameter value of the lithium battery internal resistance; I represents the current; t represents the time;

[0102] In view of the fact that the sampling of voltage and current is discrete in actual use, that is, the input of the SOC estimation algorithm is also discrete, it is necessary to discretize the above mathematical model and obtain the discrete state equation of the lithium battery equivalent circuit model as follows:

[0103]

[0104] U h (t+1)=e -|σI(t)|T U h(t)+(1-e -|σI(t)|T )U H

[0105] Wherein, U1(t+1) represents the first polarization voltage value at time (t+1); U1(t) represents the first polarization voltage value at time t; U2(t+1) represents the second polarization voltage value at time (t+1); U2(t) represents the second polarization voltage value at time t; I(t) represents the current value at time t; σ represents the attenuation factor; U h (t+1) represents the hysteresis voltage value at time (t+1); U h (t) represents the hysteresis voltage value at time t; U H Indicates the maximum hysteresis voltage value; T indicates the sampling period;

[0106] In S1, for the lithium battery equivalent circuit model, due to the existence of polarization, the open circuit voltage is the terminal voltage after the charge and discharge are completed and the battery is stationary for a long enough time, which can be approximately regarded as a monotonic function of the SOC; the open circuit voltage has a variety of different fitting methods, such as polynomial fitting, combined model, etc.; the internal resistance of the lithium battery is the ohmic internal resistance, which is the contact resistance of the power battery electrode material, electrolyte, diaphragm resistance and various parts; the RC network uses polarization internal resistance and polarization capacitance to describe the polarization phenomenon of the battery, and Kirchhoff's current law is used to write the polarization voltage state equation of each RC network. Theoretically, the order of the equivalent circuit model can be very large, but considering that the computational complexity of the third order and above is large and may not necessarily improve the model accuracy, the second-order RC equivalent circuit model is selected. At the same time, considering the hysteresis phenomenon in the battery charging and discharging process, this phenomenon is brought into the above second-order RC model, so as to further optimize the model to obtain a second-order RC equivalent circuit model with hysteresis. It is worth noting that considering the existence of hysteresis voltage will cause the original linear differential equation model to become nonlinear, so the subsequent steps need to be carried out under a nonlinear system.

[0107] S2: Based on the forgetting factor recursive least squares method, the lithium battery equivalent circuit model is parameter identified to obtain the parameter values of each component in the lithium battery equivalent circuit model, and based on the discrete state equation of the lithium battery equivalent circuit model, the terminal voltage value, the first polarization voltage value, and the second polarization voltage value at each moment are obtained, including:

[0108] Reference Figure 3It can be seen that the use of the forgetting factor recursive least squares method (FFRLS) for model parameter identification is to regard the lithium battery in the working state as a dynamic system, with the current as the system input and the terminal voltage as the system output. System identification is performed on such a single input single output (SIS0) system to obtain the coefficients of the differential equation containing the model parameters, thereby deducing the parameter values of each component in the lithium battery equivalent circuit model. Among them, the parameter values of each component in the lithium battery equivalent circuit can be calculated by directly identifying the identification results of the parameters θ1 to θ5.

[0109] S21: Based on the circuit equation of the lithium battery equivalent circuit model, determine the differential equation after discretization of the lithium battery equivalent circuit model, which is expressed as:

[0110] Y(k)=U oc (k)-U s (k)

[0111]

[0112] θ=[θ1 θ2 θ3 θ4 θ5] T

[0113] Where Y(k) represents the output of the k-th sampling lithium battery equivalent circuit model, Y(k) = U d (k), U d (k) represents the terminal voltage value of the k-th sampling lithium battery equivalent circuit model; Represents the input quantity of the k-th sampling lithium battery equivalent circuit model; U oc (k), U oc (k-1), U oc (k-2) represents the open circuit voltage value of the k-th, (k-1)-th, and (k-2)-th sampling lithium battery equivalent circuit model; U s (k), U s (k-1), U s (k-2) represents the voltage difference of the k-th, (k-1)-th, and (k-2)-th sampling lithium battery equivalent circuit model, U s (k)=U1(k)+U2(k)+U h (k); U1(k) represents the first polarization voltage value of the k-th sampling lithium battery equivalent circuit model; U2(k) represents the second polarization voltage value of the k-th sampling lithium battery equivalent circuit model; U h(k) represents the hysteresis voltage value of the k-th sampling lithium battery equivalent circuit model; I(k), I(k-1), and I(k-2) represent the current values of the k-th, (k-1)-th, and (k-2)-th sampling lithium battery equivalent circuit model, respectively; θ represents the parameter vector of the lithium battery equivalent circuit model; the time corresponding to the k-th sampling is time t;

[0114] S22: Based on the forgetting factor recursive least squares method, perform parameter identification on the lithium battery equivalent circuit model. The specific steps include:

[0115] S221: Based on the differential equation after discretization of the lithium battery equivalent circuit model, determine the least squares equation of the lithium battery equivalent circuit model, which is expressed as:

[0116]

[0117] Wherein, Y(k) represents the output of the k-th sampling lithium battery equivalent circuit model; F(k) represents the expected error of the k-th sampling lithium battery equivalent circuit model; represents the input quantity of the k-th sampling lithium battery equivalent circuit model; θ(k) represents the parameter vector of the k-th sampling lithium battery equivalent circuit model; Represents the estimated value of the parameter vector of the (k-1)th sampling lithium battery equivalent circuit model;

[0118] S222: Determine the initial value of the parameter vector of the lithium battery equivalent circuit model and the initial value of the covariance matrix of the lithium battery equivalent circuit model based on pre-assigned values based on experience;

[0119] S223: Calculating the estimated error of the k-th sampling lithium battery equivalent circuit model;

[0120] S224: Calculate the algorithm gain of the k-th sampling, the estimated value of the parameter vector of the k-th sampling lithium battery equivalent circuit model, and the covariance matrix of the k-th sampling lithium battery equivalent circuit model, and their expressions are:

[0121]

[0122] Where G(k) represents the algorithm gain of the kth sampling; Represents the estimated value of the parameter vector of the k-th sampling lithium battery equivalent circuit model; Indicates the The estimated value of the parameter vector of the k-th sampling lithium battery equivalent circuit model; γ represents the forgetting factor; H(k) represents the covariance matrix of the k-th sampling lithium battery equivalent circuit model; H(k-1) represents the covariance matrix of the (k-1)-th sampling lithium battery equivalent circuit model; E represents the identity matrix; In general, γ = 0.95 ~ 1, and γ = 0.98 is taken here;

[0123] S225: The number of cycles is increased by 1, and steps 3 to 5 are executed repeatedly until the preset number of cycles is reached, thereby obtaining the optimal estimated value of each directly identified parameter in the parameter vector of the lithium battery equivalent circuit model;

[0124] S226: Based on the optimal estimated values of each directly identified parameter in the parameter vector of the lithium battery equivalent circuit model, the intermediate parameters in the intermediate parameter set S are expressed as:

[0125]

[0126] S2=S1·θ2

[0127]

[0128] Wherein, S1, S2, S3, S4, and S5 represent the first intermediate parameter value, the second intermediate parameter value, the third intermediate parameter value, the fourth intermediate parameter value, and the fifth intermediate parameter value, respectively; θ1, θ2, θ3, θ4, and θ5 represent the optimal estimated value of the first direct identification parameter, the optimal estimated value of the second direct identification parameter, the optimal estimated value of the third direct identification parameter, the optimal estimated value of the fourth direct identification parameter, and the optimal estimated value of the fifth direct identification parameter, respectively; T represents the sampling period;

[0129] S227: Based on the optimal estimated values of the parameters in the parameter vector of the lithium battery equivalent circuit model and the intermediate parameters in the intermediate parameter set S, the parameter values of the components in the lithium battery equivalent circuit model are obtained, which are:

[0130]

[0131] Wherein, C1 represents the parameter value of the first polarization capacitor; C2 represents the parameter value of the second polarization capacitor; R1 represents the parameter value of the first polarization resistor; R2 represents the parameter value of the second polarization resistor; R0 represents the parameter value of the lithium battery internal resistance; τ1 and τ2 represent the first intermediate variable and the second intermediate variable respectively;

[0132] In S2, there are currently two main methods for obtaining lithium-ion battery parameters: offline identification and online identification. Offline identification mainly uses offline measured and stored battery external characteristic data such as voltage, current, and temperature under specific test conditions to update and calibrate model parameters, such as constant current discharge and HPPC testing. Online identification uses real-time data collected under real-time operating conditions of the power battery to update model parameters in real time.

[0133] Reference Figure 3 As shown, before the parameter identification of the lithium battery equivalent circuit model based on the recursive least squares method based on the forgetting factor is performed, the following is also included:

[0134] Under the HPPC discharge experimental conditions, multiple groups of stage discharge tests were conducted on the sample lithium battery to determine the relationship between the open circuit voltage Uoc and SOC of the lithium battery. The curve fitting was performed using the MATLAB Curve Fitting toolbox to obtain a 7th-order polynomial fitting function of Uoc and SOC, which is expressed as follows:

[0135] U oc (SOC) = p1SOC 7 +p2SOC 6 +p3SOC 5 +p4SOC 4 +p5SOC 3 +p6SOC 2

[0136] +p7SOC+p8

[0137] Among them, p1~p8 represent the coefficients of the fitting function; at the same time, the parameter values obtained by offline data identification are used as the initial parameter values of each component to facilitate subsequent online parameter identification;

[0138] Further references Figure 3 The parameter identification results obtained by using the forgetting factor recursive least squares method are brought into the second-order RC equivalent circuit model based on the hysteresis effect for verification. It can be seen from the results that the identification results after combining the two have higher accuracy;

[0139] S3: Calibrate the total capacity of the lithium battery to obtain the actual capacity of the lithium battery; based on the lithium battery equivalent circuit model, use the state of charge (SOC), the first polarization voltage, and the second polarization voltage as the first state variable, combined with the actual capacity of the lithium battery, obtain the lithium battery state space equation based on SOC, and use the extended Kalman filter algorithm to sequentially obtain the estimated value of the first state variable at each moment, thereby sequentially obtaining the estimated SOC value at each moment, including:

[0140] S31: The total capacity of the lithium battery is calibrated in a pulse discharge experiment to obtain the actual capacity of the lithium battery. The expression is:

[0141]

[0142] Among them, Q N Indicates the actual capacity of the lithium battery; Q α Indicates the lithium battery capacity at the start of capacity calibration; Q β Indicates the lithium battery capacity at the end of capacity calibration; t α Indicates the time when capacity calibration starts; t β Indicates the end time of capacity calibration; SOC αIndicates the estimated SOC value at the start of capacity calibration; SOC β It represents the estimated value at the end of capacity calibration; I(t) represents the current value at time t; t represents time;

[0143] Among them, during the charging and discharging operation of the lithium battery, the SOC value at the previous moment and the next moment will cause an error due to the change in capacity; during the battery discharge process, after the rated capacity is calibrated, the error caused by the change in capacity between the SOC value at the previous moment and the next moment can be reduced;

[0144] S32: Based on the lithium battery equivalent circuit model, with SOC, the first polarization voltage, and the second polarization voltage as the first state variable, current as the input variable, and terminal voltage as the output variable, combined with the actual capacity of the lithium battery, the lithium battery state space equation based on SOC is obtained, and its expression is:

[0145]

[0146] Among them, the first state variable at time t is SOC(t+1) represents the estimated SOC value at time (t+1); SOC(t) represents the estimated SOC value at time t; U1(t+1) represents the first polarization voltage value at time (t+1); U1(t) represents the first polarization voltage value at time t; U2(t+1) represents the second polarization voltage value at time (t+1); U2(t) represents the second polarization voltage value at time t; I(t) represents the current value at time t and serves as the input quantity at time t; U d (t) represents the terminal voltage value at time t and is used as the output at time t; g1(·) represents the first measurement function; U h (t) represents the hysteresis voltage value at time t; C1 represents the parameter value of the first polarization capacitor; C2 represents the parameter value of the second polarization capacitor; R1 represents the parameter value of the first polarization resistor; R2 represents the parameter value of the second polarization resistor; R0 represents the parameter value of the lithium battery internal resistance; T represents the sampling period; w t represents the process noise at time t; v t represents the measurement noise at time t;

[0147] S33: Using the extended Kalman filter algorithm, sequentially obtain an estimated value of the first state variable at each moment, thereby sequentially obtaining an estimated SOC value at each moment; wherein, the Kalman gain matrix at each moment is corrected based on the forgetting parameter to obtain a target Kalman gain matrix at each moment, including:

[0148] According to the lithium battery state space equation based on SOC, the state equation and measurement equation of the nonlinear system are determined, and their expressions are:

[0149]

[0150] in, represents the first state variable at time t; represents the first state variable at time (t-1); f(·) represents the transfer function; i t-1 Represents the current value at time (t-1) and serves as the input quantity at time (t-1); y t =[U d (t)] represents the output at time t; g1(·) represents the first measurement function; w t represents the process noise at time t; v t represents the measurement noise at time t;

[0151] Define the Jacobian matrices A, B, and C, and perform linearization on the state equation and measurement equation of the nonlinear system to obtain:

[0152]

[0153] Among them, the Jacobian matrices A, B, and C are defined, and their expressions are:

[0154]

[0155] Among them, Jacobian matrix A represents the state transfer matrix; Jacobian matrix B represents the state coefficient matrix; Jacobian matrix C represents the measurement matrix;

[0156] The recursive process of the extended Kalman filter algorithm based on aging state correction includes the following steps:

[0157] Step 1: Calculate the estimated value of the first state variable at time t and its corresponding error covariance matrix, which is expressed as:

[0158]

[0159]

[0160] in, represents the predicted value of the first state variable at time t; represents the estimated value of the first state variable at time (t-1); represents the predicted value of the error covariance matrix at time t; P t-1 represents the estimated value of the error covariance matrix at time (t-1); A represents the state transfer matrix; W t-1 represents the process noise covariance matrix at time (t-1);

[0161] Step 2: Calculate the Kalman gain matrix at time t, which is expressed as:

[0162]

[0163] Among them, K t represents the Kalman gain matrix at time t; C represents the measurement matrix; V t represents the measurement noise covariance matrix at time t;

[0164] The Kalman gain matrix at time t is corrected based on the forgetting parameter to obtain the target Kalman gain matrix at time t, which is expressed as:

[0165]

[0166] Among them, K′ t represents the target Kalman gain matrix at time t; λ represents the forgetting parameter;

[0167] Step 3: According to the target Kalman gain matrix at time t, the estimated value of the first state variable at time t and its corresponding error covariance matrix are corrected. The expression is:

[0168]

[0169] in, represents the estimated value of the first state variable at time t; z k Indicates process quantity; P t represents the estimated value of the error covariance matrix at time t; based on the estimated value of the first state variable at time t, the estimated SOC value at time t is obtained;

[0170] Based on steps 1 to 3, the iterations are continuously repeated until a preset number of iterations is reached, and an estimated value of the first state variable at each moment is obtained in sequence, thereby obtaining an estimated SOC value at each moment in sequence;

[0171] Among them, the total capacity of a single battery will change due to the aging process and the external environment. Therefore, before using the extended Kalman filter algorithm to estimate the SOC, a total capacity correction link needs to be added. In addition, the battery aging needs to be considered when using the extended Kalman filter algorithm to estimate the SOC. That is, after calibrating the capacity of the lithium battery, the forgetting parameter λ needs to be added to the extended Kalman filter algorithm to correct the Kalman gain. This can reduce the influence of the model's state observer on the estimation result, thereby greatly improving the accuracy of the SOC estimation. In summary, the present invention uses the extended Kalman filter algorithm (XEKF) based on aging state correction to estimate the battery SOC in real time.

[0172] S4: Based on the lithium battery equivalent circuit model, the state of energy (SOE), the first polarization voltage, and the second polarization voltage are used as the second state variable. Combined with the actual capacity of the lithium battery, the lithium battery state space equation based on SOE is determined. Its expression is:

[0173]

[0174] Among them, the second state variable at time t SOE(t+1) represents the estimated SOE value at time (t+1); SOE(t) represents the estimated SOE value at time t; U1(t+1) represents the first polarization voltage value at time (t+1); U1(t) represents the first polarization voltage value at time t; U2(t+1) represents the second polarization voltage value at time (t+1); U2(t) represents the second polarization voltage value at time t; I(t) represents the current value at time t and serves as the input quantity at time t; U d (t) represents the terminal voltage value at time t and is used as the output at time t; g2(·) represents the second measurement function; U h (t) represents the hysteresis voltage value at time t; C1 represents the parameter value of the first polarization capacitor; C2 represents the parameter value of the second polarization capacitor; R1 represents the parameter value of the first polarization resistor; R2 represents the parameter value of the second polarization resistor; R0 represents the parameter value of the lithium battery internal resistance; T represents the sampling period; w t represents the process noise at time t; v t represents the measurement noise at time t;

[0175] Using the power integration method, the estimated value of the second state variable at each moment is obtained in sequence, thereby obtaining the estimated value of SOE at each moment in sequence;

[0176] Among them, the advantage of the power integration method is that its principle is easy to understand and the calculation is simple. It can not only ignore the internal electrical characteristics of the battery, but also dynamically estimate the SOE value of the battery, which is suitable for the lithium battery equivalent circuit model of the present invention.

[0177] Among them, before determining the lithium battery state space equation based on SOE, based on the definition of SOE, the expression of SOE can be obtained as follows:

[0178]

[0179] Among them, SOE t represents the estimated SOE value at time t; SOE0 represents the initial value of SOE in the lithium battery; η represents the charging and discharging influence factor of the lithium battery; I t Indicates the current value at time t; U t Indicates the terminal voltage value at time t;

[0180] Among them, reference Figure 4 As shown, when estimating SOC and SOE in S3 and S4, the input is current and the output is voltage. The extended Kalman filter algorithm (XEKF) based on aging state correction is used to estimate SOC, and the power integration method is used to estimate SOE.

[0181] S5: Determine the estimated capacity at time t based on the actual capacity of the lithium battery, the actual discharge time, the current value at time t, and the estimated SOC value at time t Its expression is:

[0182]

[0183] Among them, Q N Indicates the actual capacity of the lithium battery; M indicates the actual discharge time of the lithium battery; I t Indicates the current value at time t, that is, the discharge current value at time t; SOC t represents the estimated SOC value at time t;

[0184] According to the actual capacity of the lithium battery and the estimated capacity at time t The difference between them determines the capacity residual at time t Its expression is:

[0185] Determine the reference energy D of the lithium battery based on the rated voltage and actual capacity of the lithium battery. N , its expression is: D N =Q N ·U N ;

[0186] Among them, Q N Indicates the actual capacity of the lithium battery; U N Indicates the rated voltage of the lithium battery;

[0187] Determine the estimated energy at time t based on the reference energy of the lithium battery, the actual discharge time, the current value at time t, the terminal voltage value at time t, and the SOE estimated value at time t. Its expression is:

[0188]

[0189] Where M represents the actual discharge time of the lithium battery; I t represents the current value at time t; Indicates the terminal voltage value at time t, that is, the discharge voltage value at time t; SOE t represents the estimated value of SOE at time t;

[0190] According to the reference energy of the lithium battery and the estimated energy at time t The difference between them determines the energy residual at time t Its expression is:

[0191] Among them, the actual capacity of the lithium battery Q N It can be used as a reference capacity, and the estimated capacity will change with time and current, and the two will generate residuals. Similarly, the reference energy is also a constant, while the estimated energy will change with time and current and voltage changes, and the two generate residuals

[0192] S6: Reference Figure 5 As shown, the fault signal is injected, and the capacity residual at time t is Compare it with the preset capacity threshold and calculate the energy residual at time t Compare the energy with the preset threshold to further determine which sensor has failed, including:

[0193] If the capacity residual at time t If the capacity residual at time t exceeds the preset capacity threshold, the current sensor in the battery management system fails. If the energy residual at time t does not exceed the preset capacity threshold, the current sensor in the battery management system will not fail. If the energy residual at time t exceeds the preset energy threshold, the voltage sensor in the battery management system fails; If the preset energy threshold is not exceeded, the voltage sensor in the battery management system does not fail;

[0194] The preset capacity threshold is ±20% of the actual capacity of the lithium battery; and the preset energy threshold is ±20% of the reference energy of the lithium battery.

[0195] In order to verify the effectiveness and accuracy of this patented method, a battery parameter identification and SOC, SOE estimation simulation platform was established in Matlab software, and fault signals were imported to verify the effectiveness of this patented method under HPPC discharge test conditions.

[0196] Obviously, the above embodiments are merely examples for clarity of explanation and are not intended to limit the implementation methods. Those skilled in the art will appreciate that other variations or modifications can be made based on the above description. It is not necessary and impossible to enumerate all implementation methods here. Obvious variations or modifications arising therefrom remain within the scope of protection of the present invention.

Claims

1. A model-based battery management system sensor fault detection method, characterized in that: include: The second-order RC equivalent circuit model with hysteresis effect is established as the lithium battery equivalent circuit model: ； ； ； in, is the terminal voltage; is the open circuit voltage; 、 are the first polarization voltage and the second polarization voltage respectively; is the hysteresis voltage; 、 are the parameter values of the first polarized capacitor and the second polarized capacitor respectively; 、 、 are the parameter values of the first polarization resistance, the second polarization resistance, and the internal resistance of the lithium battery respectively; Indicates current; Indicates time; and determining the discrete state equation of the lithium battery equivalent circuit model; performing parameter identification on the lithium battery equivalent circuit model based on the forgetting factor recursive least squares method to obtain the parameter value of each component in the lithium battery equivalent circuit model, and obtaining the terminal voltage value, the first polarization voltage value, and the second polarization voltage value at each moment based on the discrete state equation of the lithium battery equivalent circuit model; The total capacity of the lithium battery is calibrated to obtain the actual capacity of the lithium battery; based on the lithium battery equivalent circuit model, the state of charge, the first polarization voltage, and the second polarization voltage are used as the first state variable, combined with the actual capacity of the lithium battery, the lithium battery state space equation based on SOC is obtained, and the extended Kalman filter algorithm is used to obtain the estimated value of the first state variable at each moment and the estimated SOC value at each moment; wherein, based on the forgetting parameter The Kalman gain matrix at the moment is corrected to obtain The target Kalman gain matrix at time for: ； in, represents the forgetting parameter; express The predicted value of the error covariance matrix at time t; represents the measurement matrix; express The measurement noise covariance matrix at time t; Based on the lithium battery equivalent circuit model, the energy state, first polarization voltage, and second polarization voltage are used as the second state variable. Combined with the actual capacity of the lithium battery, the lithium battery state space equation based on SOE is determined. The power integration method is used to obtain the estimated value of the second state variable at each moment and the estimated value of SOE at each moment. Determine the estimated capacity at the current moment based on the actual capacity of the lithium battery, the actual discharge time, the current value at the current moment, and the estimated SOC value; obtain the capacity residual at the current moment based on the difference between the actual capacity of the lithium battery and the estimated capacity at the current moment; determine the reference energy of the lithium battery based on the rated voltage and actual capacity of the lithium battery; determine the estimated energy at the current moment based on the reference energy of the lithium battery, the actual discharge time, the current value at the current moment, the terminal voltage value, and the estimated SOE value; obtain the energy residual at the current moment based on the difference between the reference energy of the lithium battery and the estimated energy at the current moment; If the capacity residual at the current moment exceeds the preset capacity threshold, the current sensor in the battery management system fails; if the capacity residual at the current moment does not exceed the preset capacity threshold, the current sensor in the battery management system does not fail; if the energy residual at the current moment exceeds the preset energy threshold, the voltage sensor in the battery management system fails; if the energy residual does not exceed the preset energy threshold, the voltage sensor in the battery management system does not fail.

2. A model-based battery management system sensor fault detection method according to claim 1, characterized in that: Set the sampling period to , the discrete state equation of the lithium battery equivalent circuit model is: ； ； ； in, express The first polarization voltage value at the moment; express The first polarization voltage value at the moment; express The second polarization voltage value at the moment; express The second polarization voltage value at the moment; express Current value at the moment; represents the attenuation factor; express The hysteresis voltage value at the moment; express The hysteresis voltage value at the moment; Indicates the maximum hysteresis voltage value.

3. The model-based battery management system sensor fault detection method according to claim 1, characterized in that: The total capacity of the lithium battery is calibrated to obtain the actual capacity of the lithium battery, which is expressed as: ； in, Indicates the actual capacity of the lithium battery; Indicates the lithium battery capacity at the start of capacity calibration; Indicates the lithium battery capacity at the end of capacity calibration; Indicates the time when capacity calibration starts; Indicates the end time of capacity calibration; Indicates the estimated SOC value at the start of capacity calibration; Indicates the estimated value at the end of capacity calibration; express Current value at the moment; Indicates time.

4. The model-based battery management system sensor fault detection method according to claim 1, characterized in that: Based on the lithium battery equivalent circuit model, the state of charge, the first polarization voltage, and the second polarization voltage are used as the first state variable, and combined with the actual capacity of the lithium battery, the lithium battery state space equation based on SOC is obtained, and its expression is: ； ； in, The first state variable at time ; express SOC estimation value at the moment; express SOC estimation value at the moment; express The first polarization voltage value at the moment; express The first polarization voltage value at the moment; express The second polarization voltage value at the moment; express The second polarization voltage value at the moment; express The current value at the moment, and as The input amount at the time; express The terminal voltage value at the moment, and as Output at a given moment; represents the first measurement function; express The hysteresis voltage value at the moment; represents the parameter value of the first polarization capacitance; Indicates the parameter value of the second polarization capacitance; A parameter value representing the first polarization resistance; Indicates the parameter value of the second polarization resistance; Parameter value indicating the internal resistance of lithium battery; Indicates the sampling period; express The process noise at each moment; express The measurement noise at the moment.

5. A model-based battery management system sensor fault detection method according to claim 4, characterized in that: The extended Kalman filter algorithm is used to sequentially obtain an estimated value of the first state variable at each moment, thereby sequentially obtaining an estimated SOC value at each moment; wherein, the Kalman gain matrix at each moment is corrected based on the forgetting parameter to obtain a target Kalman gain matrix at each moment, including: According to the lithium battery state space equation based on SOC, the state equation and measurement equation of the nonlinear system are determined, and their expressions are: ； ； in, express The first state variable at time t; express The first state variable at time t; represents the transfer function; express The current value at the moment, and as The input amount at the time; express Output at a given moment; represents the first measurement function; express The process noise at each moment; express The measurement noise at each moment; The recursive process of the extended Kalman filter algorithm based on aging state correction includes the following steps: Step 1: Calculation The estimated value of the first state variable at time t and its corresponding error covariance matrix are expressed as follows: ； ； in, express The predicted value of the first state variable at time t; express The estimated value of the first state variable at time t; express The predicted value of the error covariance matrix at time t; express The estimated value of the error covariance matrix at time t; represents the state transition matrix; express The process noise covariance matrix at time t; Step 2: Calculation The Kalman gain matrix at time t is expressed as: ； in, express The Kalman gain matrix at time t; represents the measurement matrix; express The measurement noise covariance matrix at time t; Based on the forgetting parameter The Kalman gain matrix at the moment is corrected to obtain The target Kalman gain matrix at time t; Step 3: According to The target Kalman gain matrix at the moment, corrected The estimated value of the first state variable at time t and its corresponding error covariance matrix are expressed as follows: ； ； in, express The estimated value of the first state variable at time t; Indicates process quantity; express The estimated value of the error covariance matrix at time ; based on The estimated value of the first state variable at time , is obtained SOC estimation value at the moment; Based on steps 1 to 3, the iterations are continuously repeated until a preset number of iterations is reached, and the estimated value of the first state variable at each moment is obtained in sequence, thereby obtaining the estimated value of the SOC at each moment in sequence.

6. The model-based battery management system sensor fault detection method according to claim 1, characterized in that: Based on the lithium battery equivalent circuit model, the energy state, the first polarization voltage, and the second polarization voltage are used as the second state variable, the current is used as the input variable, and the terminal voltage is used as the output variable. In combination with the actual capacity of the lithium battery, the lithium battery state space equation based on SOE is determined, and its expression is: ； ； in, The second state variable at time ; express The estimated SOE at the moment; express The estimated SOE at the moment; express The first polarization voltage value at the moment; express The first polarization voltage value at the moment; express The second polarization voltage value at the moment; express The second polarization voltage value at the moment; express The current value at the moment, and as The input amount at the time; express The terminal voltage value at the moment, and as Output at a given moment; represents the second measurement function; express The hysteresis voltage value at the moment; represents the parameter value of the first polarization capacitance; Indicates the parameter value of the second polarization capacitance; A parameter value representing the first polarization resistance; Indicates the parameter value of the second polarization resistance; Parameter value indicating the internal resistance of lithium battery; Indicates the sampling period; express The process noise at each moment; express The measurement noise at the moment.

7. The model-based battery management system sensor fault detection method according to claim 1, characterized in that: Determining the estimated capacity at the current moment based on the actual capacity of the lithium battery, the actual discharge time, the current value at the current moment, and the SOC estimated value at the current moment; Based on the difference between the actual capacity of the lithium battery and the estimated capacity at the current moment, the capacity residual at the current moment is obtained, including: According to the actual capacity of the lithium battery, the actual discharge time, The current at the moment and The estimated SOC value at the moment is determined Estimated capacity at the moment , whose expression is: ； in, Indicates the actual capacity of the lithium battery; Indicates the actual discharge time of the lithium battery; express Current at the moment; express SOC estimation value at the moment; According to the actual capacity of lithium battery Estimated capacity at the moment Determine the difference between Capacity residual at time , whose expression is: .

8. The model-based battery management system sensor fault detection method according to claim 1, characterized in that: Determining the reference energy of the lithium battery according to the rated voltage and actual capacity of the lithium battery; Determine the estimated energy at the current moment based on the reference energy of the lithium battery, the actual discharge time, the current current value, the terminal voltage value at the current moment, and the SOE estimated value at the current moment; According to the difference between the reference energy of the lithium battery and the estimated energy at the current moment, the energy residual at the current moment is obtained, which includes: Determine the reference energy of the lithium battery based on the rated voltage and actual capacity of the lithium battery , whose expression is: ; in, Indicates the actual capacity of the lithium battery; Indicates the rated voltage of the lithium battery; According to the reference energy of lithium battery, actual discharge time, Current at the moment, The terminal voltage at the moment and The estimated SOE value at time Estimated energy at time , whose expression is: ； in, Indicates the actual discharge time of the lithium battery; express Current at the moment; express Terminal voltage at the moment; express The estimated SOE at the moment; According to the reference energy of lithium battery Estimated energy at time Determine the difference between Energy residual at time , whose expression is: .

9. The model-based battery management system sensor fault detection method according to claim 1, characterized in that: The parameter identification of the lithium battery equivalent circuit model is performed based on the forgetting factor recursive least squares method to obtain the parameter values of each component in the lithium battery equivalent circuit model, including: S21: Based on the circuit equation of the lithium battery equivalent circuit model, determine the differential equation after discretization of the lithium battery equivalent circuit model, which is expressed as: ； ； ； in, Indicates the The output of the sub-sampled lithium battery equivalent circuit model, , Indicates the Sub-sampling terminal voltage value of lithium battery equivalent circuit model; Indicates the Input quantity of sub-sampled lithium battery equivalent circuit model; 、 、 Respectively represent sequence sequence The open circuit voltage value of the sub-sampled lithium battery equivalent circuit model; 、 、 Respectively represent sequence sequence The voltage difference of the sub-sampled lithium battery equivalent circuit model, = ; Indicates the Sub-sampling the first polarization voltage value of the lithium battery equivalent circuit model; Indicates the Sub-sampling the second polarization voltage value of the lithium battery equivalent circuit model; Indicates the The hysteresis voltage value of the sub-sampling lithium battery equivalent circuit model; Respectively represent sequence sequence Sub-sampling the current value of the lithium battery equivalent circuit model; The parameter vector representing the lithium battery equivalent circuit model; 、 、 、 、 represent the optimal estimated value of the first direct identification parameter, the optimal estimated value of the second direct identification parameter, the optimal estimated value of the third direct identification parameter, the optimal estimated value of the fourth direct identification parameter, and the optimal estimated value of the fifth direct identification parameter, respectively; S22: Based on the forgetting factor recursive least squares method, perform parameter identification on the lithium battery equivalent circuit model. The specific steps include: S221: Based on the differential equation after discretization of the lithium battery equivalent circuit model, determine the least squares equation of the lithium battery equivalent circuit model, which is expressed as: ； ； in, Indicates the The output of the sub-sampled lithium battery equivalent circuit model; Indicates the Estimated error of the subsampled lithium battery equivalent circuit model; Indicates the Input quantity of sub-sampled lithium battery equivalent circuit model; Indicates the Parameter vector of the subsampled lithium battery equivalent circuit model; Indicates the Estimated values of the parameter vector of the subsampled lithium battery equivalent circuit model; S222: Determine the initial value of the parameter vector of the lithium battery equivalent circuit model and the initial value of the covariance matrix of the lithium battery equivalent circuit model; S223: Calculate the Estimated error of the subsampled lithium battery equivalent circuit model; S224: Calculate the The algorithm gain of the sub-sampling, The estimated value of the parameter vector of the sub-sampled lithium battery equivalent circuit model, The covariance matrix of the subsampled lithium battery equivalent circuit model is expressed as follows: ； ； ； in, Indicates the Algorithmic gain of subsampling; Indicates the Estimated values of the parameter vector of the subsampled lithium battery equivalent circuit model; Indicates the Estimated values of the parameter vector of the subsampled lithium battery equivalent circuit model; represents the forgetting factor; Indicates the The covariance matrix of the subsampled lithium battery equivalent circuit model; Indicates the The covariance matrix of the subsampled lithium battery equivalent circuit model; represents the identity matrix; S225: The number of cycles is increased by 1, and steps 3 to 5 are executed repeatedly until the preset number of cycles is reached, thereby obtaining the optimal estimated value of each directly identified parameter in the parameter vector of the lithium battery equivalent circuit model; S226: Based on the optimal estimated values of each directly identified parameter in the parameter vector of the lithium battery equivalent circuit model, an intermediate parameter set is obtained The intermediate parameters in are expressed as follows: ； ； ； ； ； in, 、 、 、 、 Respectively represent the first intermediate parameter value, the second intermediate parameter value, the third intermediate parameter value, the fourth intermediate parameter value, and the fifth intermediate parameter value; Indicates the sampling period; S227: Optimal estimated values and intermediate parameter sets for each parameter in the parameter vector based on the lithium battery equivalent circuit model The intermediate parameters in the , and the parameter values of each component in the lithium battery equivalent circuit model are obtained, which are: ； in, represents the parameter value of the first polarization capacitance; Indicates the parameter value of the second polarization capacitance; A parameter value representing the first polarization resistance; Indicates the parameter value of the second polarization resistance; Parameter value indicating the internal resistance of lithium battery; 、 Represent the first intermediate variable and the second intermediate variable respectively.

10. The model-based battery management system sensor fault detection method according to claim 1, characterized in that: The preset capacity threshold is the actual capacity of the lithium battery. The preset energy threshold is the reference energy of the lithium battery .

Images