A battery SOC state estimation method based on improved interval inversion filtering

Abstract

The present invention discloses a battery state-of-charge (SOC) state estimation method based on improved interval inversion filtering, belonging to the technical field of lithium battery production. This method addresses the problem that existing methods for estimating SOC states using interval inversion methods require calculating vertex values ​​in eight directions and performing multiple comparisons to obtain the maximum and minimum values ​​of the interval box when searching for the feasible domain of a set of state variable intervals. This computational complexity is excessive, consuming a large amount of computing resources, and failing to meet the requirements for rapid estimation. The present method, by setting a displacement operator, first determines the search direction, thereby requiring only a single direction for the search, significantly reducing the computational complexity.

Description

Technical Field

[0001] The present invention relates to a battery SOC state estimation method based on improved interval inversion filtering, and belongs to the technical field of lithium battery production. Background Art

[0002] Lithium-ion battery systems have the advantages of small size, large capacity, low auto-discharge power, and long cycle life. In recent years, they have been widely used in scenarios such as consumer electronics and large-scale or distributed energy storage. In the actual application of lithium-ion batteries, environmental factors and operating conditions can cause the lithium-ion battery to operate in an unstable state. If a problem with the lithium-ion battery is not discovered in time, it will cause the entire circuit state to become abnormal, posing a high safety hazard. Therefore, to ensure the safe and reliable operation of the lithium-ion battery system, it is very necessary to estimate the charge and discharge status of the lithium-ion battery in real time. The super capacitor state of charge (SOC), as an important indicator of the charge and discharge status of the lithium-ion battery, is usually the object of estimation.

[0003] Lithium battery systems operate in a complex environment and are susceptible to various environmental factors. Among existing SOC estimation methods, the open-loop ampere-hour integration method requires high sensor precision. This is an open-loop detection method in control. If the current acquisition accuracy is not high, the given initial state of charge will have a certain error, which can lead to cumulative errors as the system operates longer. The open-circuit voltage method requires the open-circuit voltage to be measured after a static state, often requiring a long wait time. This makes it unable to meet real-time monitoring requirements and has poor versatility. The Kalman filter method is a minimum variance estimation method using the time domain state space theory. It belongs to the category of statistical estimation, but it has a high dependence on the model. At the same time, considering that various interference noises in the actual environment do not meet the specific probability distribution, they can be divided into unknown but bounded noises. Therefore, someone proposed a set membership interval inversion filtering method. By collecting the current and voltage at the battery output end during battery operation, the health status of the battery is continuously updated. That is, the SOC state of the battery is estimated by adopting an inversion interval estimation method, and the corresponding relationship between the battery charging and discharging is obtained according to the corresponding relationship of the pre-established battery equivalent model. The interval inversion operation is performed based on the voltage and current of the sensor acquisition end between the two time nodes, and the upper and lower bounds of the battery SOC during charging and discharging are estimated to predict the battery SOC state, and good results have been achieved.

[0004] However, when searching for the feasible domain of a set of state variable intervals, the currently proposed interval inversion method needs to calculate the vertex values ​​in eight directions each time and perform multiple comparisons to obtain the maximum and minimum values ​​of this interval box. Therefore, the amount of calculation is too large, which occupies a lot of computing resources and cannot meet the requirements of fast estimation. Moreover, the accuracy of battery estimation in the existing interval inversion calculation method is too dependent on the accuracy of the measured voltage. Summary of the Invention

[0005] In order to reduce the amount of calculation as much as possible while ensuring the estimation accuracy of the battery state of charge (SOC), the present invention improves the existing inversion interval estimation method. By setting the displacement operator T, the directional dimension corresponding to the maximum value in the search space is found, thereby avoiding the problem of large amount of calculation caused by the need to search in eight directional dimensions each time in the existing inversion interval estimation method. In addition, the method of the present application avoids dependence on the accuracy of the measured voltage.

[0006] A battery SOC state estimation method based on improved interval inversion filtering, the method comprising:

[0007] Step S1, considering the sensor accuracy constraints in the battery system, constructing a state equation and a measurement equation for battery SOC estimation;

[0008] Step S2, based on the constructed state equation and measurement equation, design a set-membership interval observer with unknown but bounded noise;

[0009] In step S3, the interval maximum value corresponding to the space mapping is inverted based on the state equation, the measurement equation, and the set membership interval observer. By adding a displacement operator T, the direction of the interval maximum value is obtained, reducing the calculation time. Condition judgment and classification are performed based on the interval maximum value, and the next search space is determined at the same time. In step S4, the battery state of charge (SOC) value is estimated by interval inversion using the voltage and current values ​​of the battery under the charge and discharge state measured under the specified environment.

[0010] In step S5, the battery SOC at the next moment is estimated based on the voltage and current values ​​measured next time, until the estimation process is completed.

[0011] Optionally, the state equation and measurement equation constructed in step S1 are:

[0012] x k =Ax k-1 +Bu k-1 +w k-1

[0013] y k =Cx k +Du k +v k

[0014] Among them, x k =[SOC k ,U pa,k ,U pc,k ] T is the state matrix of the system, SOC k 、U pa,k 、Upc,k are system state variables, representing the state of charge, activation polarization voltage, and concentration polarization voltage at time k; u k =U k is the input matrix of the system, I k represents the discharge current at time k, y k =U k is the output matrix of the system, U k represents the output voltage at time k;

[0015] C=[f(z k ) -1 -1] is the parameter matrix of the system, w(k) and v(k) are process noise and measurement noise respectively, and both are bounded; Δt is the sampling time of the system; R pa and C pa are the equivalent electrochemical polarization internal resistance and capacitance of the power battery respectively; R pc and C pc are the equivalent concentration polarization resistance and capacitance of the power battery, τ pa =R pa C pa The time constant of the equivalent activation polarization of the battery, τ pc =R pc C pc Represents the time constant of the equivalent concentration polarization of the battery; η i,t represents the comprehensive influencing factor of temperature and discharge rate; Q0 represents the rated capacity of the battery; f(z k ) represents the fitting relationship between battery SOC and output; z k Indicates the state of charge at time k, i.e. SOC k .

[0016] Optionally, the set membership interval observer designed in step S2 is:

[0017]

[0018]

[0019] Among them, R n Represents n-dimensional real number space; Represents the three-dimensional space box corresponding to the system state variables at time k; O (k:k+n) represents the observability matrix; and Y k Respectively represent the upper and lower bounds of the state matrix;

[0020]

[0021]

[0022] in:

[0023]

[0024]

[0025]

[0026]

[0027]

[0028] y (k:k+s) 、u (k:k+s) 、 w (k:k+ss) 、 v (k:k+s) They are the output, input, process noise upper bound, process noise lower bound, measurement noise upper bound and measurement noise lower bound of the system from time k to k+s respectively.

[0029] Optionally, step S3 includes:

[0030] set up The a priori search domain S is determined according to the maximum and minimum values ​​of the SOC in the system state variable; at this time, As a three-dimensional box space, there are eight directional dimensions at its center point. Therefore, to calculate the maximum and minimum values, it is necessary to calculate the vertex values ​​in the eight directional dimensions and then compare them with each other to determine the maximum and minimum values ​​of the current search space.

[0031] To avoid calculating the vertex values ​​in eight directions every time, a displacement operator T is set. The search direction determined by the displacement operator is used to calculate only the vertex values ​​in the search direction. The position operator is:

[0032]

[0033]

[0034] where h sm is the element at the corresponding position in the state matrix x of the system at the sth moment after the kth moment, s=1,2,3…; m=1,2,3;

[0035] Each row of the shift operator T Representative Space The directional dimension of the maximum value in the displacement operator T, each element of which, if 1, corresponds to the space The right edge of the corresponding element in the , if it is -1, it should be space The left border of Representative Space The direction dimension of the minimum value in ;

[0036] The displacement operator is updated during initialization. Each operation determines the search direction through the displacement operator and obtains a new When, find Perform an operation on the corresponding boundary value in the search direction, which is The interval corresponds to Maximum and minimum values, and find the maximum and minimum values ​​in 2s as boundaries and sets Make comparisons and judgments to determine the next search space.

[0037] Optional, during the search,

[0038] (1) If and There is an intersection but it does not completely belong to and If the width of the represented interval box is greater than the set interval precision, the interval box is bisected along the dimension of the maximum interval width to obtain two new interval boxes;

[0039] (2) If and The intersection of is empty, then The interval box represented is an infeasible subset;

[0040] (3) If Completely belongs to the set but The represented interval boxes are feasible subsets and are placed into the pre-assigned feasible set;

[0041] (4) If and If there is a partial intersection and the width of the corresponding interval box is less than the interval precision parameter ε, then the interval box is an uncertain subset and is placed in the uncertain set.

[0042] Optionally, step S1 includes:

[0043] Step S1.1: Determine the equivalent circuit of the battery, establish an equivalent circuit model of the battery based on the equivalent circuit, discretize the equivalent circuit model, and obtain a second-order Thevenin discrete linear model of the battery;

[0044] Step S1.2: Perform charge and discharge experiments on the battery and determine the parameters of the second-order Thevenin discrete linear model of the battery based on the experimental data;

[0045] Step S1.3: Based on the second-order Thevenin discrete linear model of the battery and the sensor accuracy constraints, a state equation and a measurement equation for battery SOC estimation are constructed to characterize the constrained battery system; the sensors include sensors for measuring current and voltage respectively.

[0046] Optionally, the equivalent circuit model of the battery is:

[0047] U=U oc -IR0-U pa -U pc

[0048]

[0049]

[0050] Among them, R0 represents the DC internal resistance of the battery.

[0051] Optionally, the second-order Thevenin discrete linear model of the battery is:

[0052] U k =U oc,k -I k R0-U pa,k -U pc,k

[0053]

[0054]

[0055] Among them, I k-1 Represents the discharge current at time k-1.

[0056] Optionally, the prior search domain Contains all system state variable interval sets that meet the requirements. The beneficial effects of the present invention are:

[0057] By setting the displacement operator T, the directional dimension corresponding to the maximum value in the search space is found, thereby avoiding the problem of large computational complexity caused by the need to search in eight directional dimensions each time in the existing inverse interval estimation method, and the method of this application avoids dependence on the measurement voltage accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0059] Figure 1 This is a flowchart of a battery SOC state estimation method based on improved interval inversion filtering according to an embodiment of the present invention.

[0060] Figure 2 The following is a logic flow chart of an improved interval inversion filtering method according to an embodiment of the present invention.

[0061] Figure 3 This is an equivalent circuit diagram of a lithium battery dual-polarization battery model in a battery SOC state estimation method based on improved interval inversion filtering according to an embodiment of the present invention.

[0062] Figure 4 It is a schematic diagram of the estimation result of the battery SOC when the battery is discharging according to the battery SOC state estimation method based on improved interval inversion filtering described in one embodiment of the present invention.

[0063] Figure 5 This is a diagram showing simulation results of SOC estimation error obtained by a battery SOC state estimation method based on improved interval inversion filtering according to an embodiment of the present invention.

[0064] Figure 6 This is a simulation comparison diagram of the time used by the battery SOC state estimation method based on the improved interval inversion filtering according to an embodiment of the present invention and the battery SOC state estimation method based on the traditional interval inversion filtering. DETAILED DESCRIPTION

[0065] To make the objectives, technical solutions and advantages of the present invention more clear, the embodiments of the present invention will be described in further detail below with reference to the accompanying drawings.

[0066] Example 1:

[0067] This embodiment provides a battery SOC state estimation method based on improved interval inversion filtering, see Figure 1 , the method comprising:

[0068] The battery is charged and discharged under specified conditions, and a second-order Thevenin discrete linear model is constructed based on the experimental data and lithium battery performance parameters.

[0069] The equivalent circuit of the lithium battery bipolarization battery model is as follows Figure 3 As shown, R0 represents the DC internal resistance of the battery, R pa and C pa are the equivalent electrochemical polarization internal resistance and capacitance of the power battery, U pa is the voltage across the two terminals after they are connected in parallel, i.e. the equivalent activation polarity voltage of the battery; R pc and C pcare the equivalent concentration polarization resistance and capacitance of the power battery, U pc is the voltage across the two terminals after they are connected in parallel, i.e., the equivalent concentration polarization voltage; I is the discharge current, U oc is the open circuit voltage of the battery's internal power supply; U is the battery terminal voltage, that is, the output voltage.

[0070] according to Figure 3 The equivalent circuit of the lithium battery bipolarization battery model is shown, and the bipolarization second-order continuous-time model of the lithium battery is established:

[0071] U t =U oc -IR0-U pa -U pc

[0072]

[0073]

[0074] After discretization, the battery discrete linear circuit model is obtained:

[0075] U k =U oc,k -I k R0-U pa,k -U pc,k

[0076]

[0077]

[0078] Among them, I k 、U k 、U pa,k 、U pc,k 、U oc,k represents the discharge current, terminal voltage, activation polarization voltage, concentration polarization voltage, and open circuit voltage at time k; Δt is the sampling time of the system. Where R0 represents the DC internal resistance of the battery, τ pa =R pa C pa Represents the time constant of battery activation polarization, τ pc =R pc C pc Represents the time constant of battery concentration polarization;

[0079] The above parameters are determined by charge and discharge experiments.

[0080] According to the performance parameters of the lithium battery, the circuit characteristics of charge and discharge, and the sensor accuracy constraints, the state equation and measurement equation for battery SOC estimation are constructed.

[0081] The determined state equation and measurement equation are shown below:

[0082] x k =Ax k-1 +Bu k-1 +w k-1

[0083] y k =Cx k +Du k +v k

[0084] Among them, x k =[SOC k ,U pa,k ,U pc,k ] T Defined as the state matrix of the system, SOC k 、U pa,k 、U pc,k is the state matrix x k The elements in u k =I k Define the input matrix of the system, y k =U k Defined as the output matrix of the system;

[0085] C=[f(z k ) -1 -1] is the parameter matrix of the system, k is the parameter matrix of the system in discrete time, w(k) and v(k) are the process noise and measurement noise respectively, and both are bounded.

[0086] Substitute the data determined by the charge and discharge experiments into the established state equation and measurement equation:

[0087]

[0088]

[0089] The battery equivalent circuit model uses the least square method to perform offline parameter identification on the experimental data to establish a linear function equation f(SOC) between the open circuit voltage OCV and the battery state SOC. k ).

[0090] Perform a complete discharge experiment on the battery and use sensors to measure the current and voltage of the battery during normal operation.

[0091] For constrained battery systems, design a set-membership interval filter with unknown but bounded noise;

[0092] Design a set membership interval observer to estimate the upper and lower bounds of the state interval corresponding to the battery system state matrix, including:

[0093]

[0094]

[0095] in

[0096]

[0097]

[0098]

[0099]

[0100]

[0101]

[0102]

[0103] y (k:k+s) 、u (k:k+s) 、 w (k:k+s) 、 v (k:k+s) are the output, input, upper bound of process noise, lower bound of process noise, upper bound of measurement noise, and lower bound of measurement noise of the system from time k to k+s, respectively. As the time length s increases, the accuracy of the interval filtering method increases, but the corresponding computational effort also increases.

[0104] set up Prior search domain The a priori search domain S is determined based on the maximum and minimum values ​​of the SOC in the system state variables. In order to ensure that it contains all feasible sets, S is set to be larger than the actual Possible values. In the prior search domain Inner search feasible set and look for The maximum and minimum points in the , and then calculate the corresponding The maximum value of The maximum value of is the same as that obtained by the set membership interval observer designed above. Make a comparison, and determine to further narrow the search domain based on the comparison results, and finally obtain the battery SOC state estimation value.

[0105] The search domain is a three-dimensional box space with eight directional dimensions at its center point. It is necessary to find the corresponding maximum and minimum points in this space. In the traditional calculation method, each time the space is divided into two parts, The maximum and minimum boundary values ​​of will change, and the corresponding The points in the space also need to be re-determined. The model is a linear model. First, Find the vertex values ​​at the center point of the space and its eight directional dimensions, and perform one operation on each point to obtain the possible After obtaining all possible maximum values, a search operation is performed to find the maximum and minimum values. This process will take up a lot of computing resources. Here, the method of this application sets a displacement operator T, including:

[0106]

[0107]

[0108] where h sm is the element in the state matrix x of the system at the sth moment after k moments, s=1,2,3…; m=1,2,3; where each row of the displacement operator T Representative Space The directional dimension of the maximum value in the displacement operator T, each element of which, if 1, corresponds to the space The right edge of the corresponding element in the , if it is -1, it should be space The left border of . Representative Space The direction dimension of the minimum value in .

[0109] The displacement operator is updated during initialization, and the search direction is determined by the displacement operator in each operation.

[0110] Based on this, we can get a new When, find The corresponding boundary value of The interval corresponds to The maximum and minimum values, and find the maximum and minimum values ​​among the smaller number of s values ​​as the boundary and set Compare and judge, use the voltage and current value data of a series of complete discharge experiments to estimate the battery SOC value through interval inversion;

[0111] Then update the voltage and current values ​​measured next time and estimate the battery SOC at the next moment until the estimation process is completed.

[0112] The final feasible set X and the uncertain set X∪ε are obtained. Since the state estimation interval of X is small and the wrapping is lower, a tighter interval estimation can be obtained by using the row set X.

[0113] The center of X, that is, the Chebyshev center of X, is taken as the estimated value of the parameter to be estimated.

[0114] Example 2:

[0115] This embodiment provides a battery SOC state estimation method based on improved interval inversion filtering, such as Figure 2 As shown, the method includes:

[0116] Set the observation time length s, initialize the algorithm, set the system parameters and the interval accuracy ε; in the subsequent simulation experiments, set the observation time length s to 4 seconds and the interval accuracy ε to 0.0001;

[0117] Set the a priori search domain The interval is required to be large enough to include all state variable interval sets that meet the requirements; since the maximum value of the ratio of the remaining battery power to the capacity before use in the actual scenario is 1, considering the requirement that the interval is large enough, in the subsequent simulation experiments, the prior search domain for SOC is set to [0, 1.5]; for U pa,k ,U pc,k , all set to [0, 0.2].

[0118] Calculate y at this moment k Corresponding And use the dichotomy method to calculate the corresponding The interval set of the state variable x;

[0119] The proposed method uses a bisection method to recursively and systematically search the feasible domain. When searching for the feasible region, it is necessary to calculate The maximum and minimum values ​​of the interval can be determined and relationship. At this time There are eight directional dimensions. Since the model is a linear model, the traditional method needs to calculate the vertex values ​​in eight directions and compare them to get the maximum and minimum values ​​of the interval box. Multiple comparisons are required each time the search is performed, which wastes a lot of time. The method of this application improves this by adding a displacement operator T. By first determining its search direction through the displacement operator, it can be easily obtained. The corresponding maximum value.

[0120] During the search process, the following judgments need to be made about the interval:

[0121] (1) If and There is an intersection but it does not completely belong to and The width of the represented interval box is greater than the set interval precision. The interval box needs to be divided into two parts along the dimension of the maximum interval width to obtain two new interval boxes.

[0122] (2) If and The intersection of is empty, The interval box represented is an infeasible subset.

[0123] (3) If Completely belongs to the set The represented interval boxes are feasible subsets and are placed into the pre-assigned feasible set;

[0124] (4) If and If there is a partial intersection and the width of the corresponding interval box is less than the interval precision parameter ε, then the interval box is an uncertain subset and is placed in the uncertain set.

[0125] If the four judgments are completed, the next interval can be judged. This process ends when all interval boxes have been traversed.

[0126] like Figure 4 It can be seen that this embodiment can well estimate the battery SOC state in the continuous discharge state.

[0127] Figure 5 This is the SOC estimation error simulation result diagram obtained in this embodiment. It can be seen that the SOC estimation error obtained by the method of the present application is within 0.015. At present, if the SOC estimation error is within 0.03, it can meet the general system requirements; for example, CN 112858928 A, a lithium battery SOC estimation method based on online parameter identification, with an application date of 2021.03.08, and CN 113138344 A, a SOC estimation method based on fractional-order extended Kalman filter algorithm, with an application date of 2021.04.23, have SOC estimation errors of 0.03 and 0.025 respectively. It can be seen that the SOC estimation error obtained by the method of the present application can meet the system requirements, and relatively speaking, the error is smaller.

[0128] like Figure 6 As shown, the total computation time of this embodiment is faster under different sampling times. Compared with the current interval inversion method, it can save about 40% of the computation time, which greatly improves the speed of the algorithm.

[0129] In summary, the method of the present application reduces the number of operations and significantly reduces the operation time while meeting the requirements for SOC estimation error, and can meet the requirements for fast estimation.

[0130] It should be noted that the various parameters set in the above simulation experiment can be set by technical personnel according to the actual situation of the battery, and this application does not limit the specific values.

[0131] Some steps in the embodiments of the present invention may be implemented using software, and the corresponding software program may be stored in a readable storage medium, such as a CD or a hard disk.

[0132] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A battery SOC state estimation method based on improved interval inversion filtering, characterized in that: The method comprises: Step S1, considering the sensor accuracy constraints in the battery system, constructing a state equation and a measurement equation for battery SOC estimation; Step S2, based on the constructed state equation and measurement equation, design a set-membership interval observer with unknown but bounded noise; Step S3: Based on the state equation, measurement equation, and set membership interval observer, the interval maximum value corresponding to the inverse calculation space mapping is obtained. By adding a displacement operator T, the direction of the interval maximum value is obtained, which reduces the calculation time. Condition judgment and classification are performed based on the interval maximum value, and the next search space is determined at the same time. Step S4, using the voltage and current values ​​of the battery under the charge and discharge state measured under the specified environment, the battery SOC value is estimated by interval inversion; Step S5, estimating the battery SOC at the next moment based on the voltage and current values ​​measured next time, until the estimation process is completed; The step S3 comprises: set up The a priori search domain S is determined according to the maximum and minimum values ​​of the SOC in the system state variable; at this time, As a three-dimensional box space, there are eight directional dimensions at its center point. Therefore, to calculate the maximum and minimum values, it is necessary to calculate the vertex values ​​in the eight directional dimensions and then compare them with each other to determine the maximum and minimum values ​​of the current search space. To avoid calculating the vertex values ​​in eight directions every time, a displacement operator T is set. The search direction determined by the displacement operator is used to calculate only the vertex values ​​in the search direction. The displacement operator is: where h sm is the element at the corresponding position in the state matrix x of the system at the sth moment after the kth moment, s=1,2,3…; m=1,2,3; Each row of the shift operator T Representative Space The directional dimension of the maximum value in the displacement operator T, each element of which, if 1, corresponds to the space The right edge of the corresponding element in the , if it is -1, it should be space The left border of Representative Space The direction dimension of the minimum value in ; The displacement operator is updated during initialization, and the search direction is determined by the displacement operator in each operation; In each binary division, a new When, find Perform an operation on the corresponding boundary value in the search direction, which is The interval corresponds to Maximum and minimum values, and find the maximum and minimum values ​​in 2s as boundaries and sets Compare and judge to determine the next search space, O (k:k+s) represents the observability matrix; and Y k Represent the upper and lower bounds of the state matrix respectively.

2. The method according to claim 1, characterized in that The state equation and measurement equation constructed in step S1 are: x k =Ax k-1 +Bu k-1 +w k-1 y k =Cx k +You k +v k Among them, x k =[SOC k ,U pa,k ,U pc,k ] T is the state matrix of the system, SOC k 、U pa,k 、U pc,k are system state variables, representing the state of charge, activation polarization voltage, and concentration polarization voltage at time k; u k =I k is the input matrix of the system, I k represents the discharge current at time k, y k =U k is the output matrix of the system, U k represents the output voltage at time k; C=[f(z k ) -1 -1] is the parameter matrix of the system, w(k) and v(k) are process noise and measurement noise respectively, and both are bounded; Δt is the sampling time of the system; R pa and C pa are the equivalent electrochemical polarization internal resistance and capacitance of the power battery respectively; R pc and C pc are the equivalent concentration polarization resistance and capacitance of the power battery, τ pa =R pa C pa The time constant of the equivalent activation polarization of the battery, τ pc =R pc C pc Represents the time constant of the equivalent concentration polarization of the battery; η i,t represents the comprehensive influencing factor of temperature and discharge rate; Q0 represents the rated capacity of the battery; f(z k ) represents the fitting relationship between battery SOC and output; z k Indicates the state of charge at time k, i.e. SOC k .

3. The method according to claim 2, characterized in that The set membership interval observer designed in step S2 is: Among them, R n Represents n-dimensional real number space; The three-dimensional space box representing the system state variables at time k; Y k =y (k:k+s) -O ux(k:k+s) u (k:k+s) -O w(k:k+s) w (k:k+s) -THE v(k:k+s) v (k:k+s) -THE uy(k:k+s) u (k:k+s) , in: y (k:k+s) 、u (k:k+s) 、 w (k:k+s) 、 v (k:k+s) They are the output, input, process noise upper bound, process noise lower bound, measurement noise upper bound and measurement noise lower bound of the system from time k to k+s respectively.

4. The method according to claim 3, characterized in that During the search process, (1) If and There is an intersection but it does not completely belong to and If the width of the represented interval box is greater than the set interval precision, the interval box is bisected along the dimension of the maximum interval width to obtain two new interval boxes; (2) If and The intersection of is empty, then The interval box represented is an infeasible subset; (3) If Completely belongs to the set but The represented interval boxes are feasible subsets and are placed into the pre-assigned feasible set; (4) If and If there is a partial intersection and the width of the corresponding interval box is less than the interval precision parameter ε, then the interval box is an uncertain subset and is placed in the uncertain set.

5. The method according to claim 4, characterized in that The step S1 comprises: Step S1.1: Determine the equivalent circuit of the battery, establish an equivalent circuit model of the battery based on the equivalent circuit, discretize the equivalent circuit model, and obtain a second-order Thevenin discrete linear model of the battery; Step S1.2: Perform charge and discharge experiments on the battery and determine the parameters of the second-order Thevenin discrete linear model of the battery based on the experimental data; Step S1.3: Based on the second-order Thevenin discrete linear model of the battery and the sensor accuracy constraints, a state equation and a measurement equation for battery SOC estimation are constructed to characterize the constrained battery system; the sensors include sensors for measuring current and voltage respectively.

6. The method according to claim 5, characterized in that The equivalent circuit model of the battery is: U=U oc -IR0-U pa -IN pc Among them, R0 represents the DC internal resistance of the battery, U oc is the open circuit voltage of the battery's internal power supply.

7. The method according to claim 6, characterized in that The second-order Thevenin discrete linear model of the battery is: IN k =U oc,k -AND k R0-U pa,k -IN pc,k Among them, I k-1 Represents the discharge current at time k-1.

8. The method according to claim 7, characterized in that The prior search domain Contains a set of all system state variable intervals that meet the requirements.

Images