A fault detection isolation and localization method for general battery systems

Abstract

The application provides a fault detection, isolation and positioning method for a general battery system. The method can realize detection, isolation and positioning of various types of faults including voltage sensor faults, current sensor faults, poor contact, cell faults and the like by using currently widely used voltage and current data. The modeling and calculation process is relatively simple, and additional disassembly, modification and sensor installation of the battery system are not required, thereby providing a simple, efficient and reliable comprehensive fault diagnosis solution for general battery systems represented by lithium ion batteries.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of battery system safety state monitoring and fault diagnosis, and particularly relates to a variety of fault comprehensive detection, isolation and positioning technology of large-scale series-parallel group battery system. BACKGROUND

[0002] At present, for a complex battery system composed of a large number of battery monomers and branches, internal faults cannot be completely avoided, the causes of various faults and the electrical coupling phenomenon between monomers have not been completely mastered in the field, and therefore the position of fault diagnosis technology in the reliability management of the battery system is crucial. Accurate fault diagnosis can not only ensure the safe operation of the battery system, but also effectively reduce the risk of thermal runaway fire accidents induced by faults. In the existing fault diagnosis for general battery systems, threshold detection is mostly used, which is often difficult to accurately and quickly detect and isolate faults at the system level using limited sensing information, thereby greatly increasing the safety risk of the system. Moreover, the current fault diagnosis method is generally designed for a single fault type, and lacks systematic fault modeling and isolation design, so it is easy to misreport in actual use. SUMMARY

[0003] Therefore, in view of the technical problems existing in the field, the present application provides a fault detection, isolation and positioning method for general battery systems, which specifically includes the following steps:

[0004] Step one, obtain the battery system structure parameters of the measured battery system, including the number of battery monomers and the series-parallel composition form, and the model parameters for battery monomer and system modeling, including the open circuit voltage of the battery monomer;

[0005] Step two, select a suitable model to model the voltage and current characteristics of the battery monomer; and according to the established battery monomer model and battery system composition structure, construct a residual-based battery system fault diagnosis model for calculating the residual corresponding to the battery monomer, branch and entire battery system, and establish a mapping relationship corresponding to different fault types by using the combination of three levels of residuals;

[0006] Step three, initialize the fault diagnosis process, including setting the initial state of the battery monomer and system including OCV, SOC, etc., the algorithm parameters for battery monomer state estimation, and setting time t = 0;

[0007] Step four, let t = t + 1, execute the state estimation algorithm to update the state variables of the battery monomer, and calculate the OCV, SOC, polarization voltage and other parameters;

[0008] Step five, using the parameters calculated in step four, input the fault diagnosis model established in step two to calculate the third level residual; by comparing each residual with the given threshold, the residuals are converted into logical quantities containing only 0 and 1, defined as residual flag quantities;

[0009] Step six, using the combination of residual flag quantities to determine whether a fault has occurred, if all elements of the residual flag quantity are 0, no fault occurs, return to step four; if there are elements of the residual flag quantity that are not 0, go to step seven;

[0010] Step seven, using the mapping relationship obtained in step two to convert into a fault mapping table corresponding to different combinations of residual flag quantities, obtain the fault type and positioning result by looking up the table, output the fault diagnosis conclusion and return to step four.

[0011] Further, in step one, the effects of correction temperature, aging, and SOC change on other parameters are also obtained to improve the accuracy of subsequent modeling.

[0012] Further, in step two, when modeling the battery monomer, an equivalent circuit model is selected that can at least describe the current power, polarization state, such as SOC, OCV, polarization voltage, etc.

[0013] The specific form of the third level residual r is defined as follows:

[0014] r = [r1, r2, r3]

[0015] Where the first element monomer residual r1 is a p x s matrix, p is the number of parallel branches, and s is the number of series monomers on each branch, so r1 corresponds to each monomer in the battery system; the second element branch residual r2 is a vector of length p, corresponding to each branch; the third element system residual r3 is a scalar; the calculation formula of each element in the residual is as follows:

[0016]

[0017] Where k is the time, subscript i represents the branch it belongs to, i = 1, 2, 3, …, p, and subscript j represents the jth monomer on a branch, j = 1, 2, 3, …, s; OCV, R, U D , y I , y U , y Ii , y Uij are the open circuit voltage, ohmic resistance, and polarization voltage of the battery monomer, respectively; y

[0018] Further, in step four, real-time voltage and current data are specifically taken as the input of the state estimation algorithm, and state estimation is performed at a sampling interval of Δt, and the estimation result is corrected by using an estimation error feedback link.

[0019] Further, in step six, after three-level residual calculation and comparison with a given threshold, a residual flag combination r' containing only 0 and 1 elements is obtained; in step seven, a fault feature quantity f with the same combination form is searched for in a mapping table obtained by converting the combination of three-level residuals and the mapping relationship of different fault types, that is, the diagnosis of the fault location of different fault types is realized.

[0020] The fault detection, isolation and positioning method for general battery systems provided by the present application only needs to use the currently widely used voltage and current data, and can realize the detection, isolation and positioning of various types of faults including voltage sensor faults, current sensor faults, poor contact, cell faults and the like, the modeling and calculation process is relatively simple, and additional disassembly, modification and sensor installation of the battery system are not required, thereby providing a simple, efficient and reliable comprehensive fault diagnosis solution for general battery systems represented by lithium ion batteries. BRIEF DESCRIPTION OF DRAWINGS

[0021] Figure 1 The flowchart of the method provided by the present application is shown in the figure;

[0022] Figure 2 The diagnosis results of the present application for specific branch and single cell voltage sensor faults are shown in the figure. DETAILED DESCRIPTION

[0023] The technical solutions of the present application will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0024] The fault detection, isolation and positioning method for general battery systems provided by the present application, as shown in the figure, specifically includes the following steps: Figure 1

[0025] Step one, obtaining the battery system structure parameters of the measured battery system, including the number of battery monomers and the series-parallel connection form, and the model parameters for battery monomer and system modeling, etc.

[0026] ​Step two, select appropriate model to model the voltage and current characteristics of the battery monomer; and according to the established battery monomer model and battery system composition structure, the battery system fault diagnosis model based on residual error is constructed, which is used to calculate the residual error corresponding to the battery monomer, branch and whole battery system respectively, and the mapping relationship corresponding to different fault types is established by using the combination of three levels of residual error;

[0027] Step three, initialize the fault diagnosis process, including setting the initial state of the battery monomer and system including OCV, SOC and the like, the algorithm parameters for battery monomer state estimation, such as Kalman filter related process noise covariance matrix, measurement noise covariance matrix and the like, and let time t = 0;

[0028] Step four, let t = t + 1, execute state estimation algorithm to update the state variable of the battery monomer, and calculate OCV, SOC, polarization voltage and the like;

[0029] Step five, using the parameters calculated in step four, input the fault diagnosis model established in step two to calculate three levels of residual error; by comparing each residual error with the given threshold, each residual error is converted into a logic quantity containing only 0 and 1, which is defined as residual error flag quantity;

[0030] Step six, judge whether a fault occurs by using the combination of residual error flag quantity, if all elements of residual error flag quantity are 0, no fault occurs, return to step four; if there is an element of residual error flag quantity which is not 0, enter step seven;

[0031] Step seven, convert the mapping relationship obtained in step two into a fault mapping table corresponding to different residual error flag quantity combinations, obtain the fault type and positioning result by looking up the table, output the fault diagnosis conclusion and return to step four.

[0032] In a preferred embodiment of the present application, the above method is used for fault diagnosis of a certain 2 string 2 parallel lithium ion battery system. The two branches of the 2 string 2 parallel battery system are denoted as s and p, and the four battery monomers are denoted as s1, s2, p1 and p2 respectively.

[0033] In step one, the influence of temperature correction, aging and SOC change on other parameters is also obtained to improve the accuracy of subsequent modeling.

[0034] A second-order RC equivalent circuit model is established for the battery monomer;

[0035] The specific form of three levels of residual error r is defined as follows:

[0036] r = [r1, r2, r3]

[0037] Wherein, the first element monomer residual error r1 is a p x s matrix, p is the number of parallel branches, s is the number of series monomers on each branch, so r1 corresponds to each monomer in the battery system one by one; the second element branch residual error r2 is a p-length vector, which corresponds to each branch one by one; the third element system residual error r3 is a scalar; the calculation formula of each element in the residual error is as follows:

[0038]

[0039] Wherein, k is the time, subscript i represents the branch to which it belongs, i = 1, 2, 3, …, p, subscript j represents the jth monomer on a branch, j = 1, 2, 3, …, s; OCV, R, U D , y I , y U , y Ii , y Uij are the total current, total voltage, and branch current and monomer voltage sensor readings, respectively.

[0040] In step four, real-time voltage and current data are specifically used as the input of the state estimation algorithm, Kalman filter algorithm is selected, and state estimation is performed with Δt as the sampling interval. The estimation result is corrected by using the estimation error feedback link.

[0041] In step six, after three-level residual error calculation and comparison with the given threshold, the residual flag quantity combination r' containing only 0 and 1 elements is obtained; in step seven, in the mapping table obtained by converting the combination of three-level residual errors and the mapping relationship of different fault types, the fault feature quantity f with the same combination form is searched, that is, the fault location diagnosis of different fault types is realized. In this embodiment, the mapping table shown in Table 1 is specifically used:

[0042] Table 1 Fault type, location and residual flag quantity mapping table

[0043]

[0044]

[0045] Through table lookup, it is found in this embodiment that the first monomer of branch s has a voltage sensor fault, as shown in Table 1. Figure 2

[0046] It should be understood that the size of the serial number of each step in the embodiment of the present application does not mean the order of execution, and the execution order of each process should be determined according to its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present application.

[0047] ​While embodiments of the application have been shown and described, it is to be understood that the embodiments described are merely exemplary of the principles and application of the present application. Numerous modifications and adaptions can be effected without departing from the spirit and scope of the present application, which is not limited to the exact construction and arrangement described. It is intended, therefore, to cover all modifications and adaptions that fall within the scope of the claims and their equivalents.

Claims

1. A method for fault detection, isolation and localization for a general battery system, the method comprising: Specifically comprising the following steps: ​ Step one, obtaining the battery system structure parameters of the battery system to be tested, including the number of battery monomers and the parallel-serial connection form, and the battery monomer open circuit voltage and model parameters for battery monomer and system modeling; Step two, selecting a suitable model to model the voltage and current characteristics of the battery monomer; and according to the established battery monomer model and battery system composition structure, a residual-based battery system fault diagnosis model is constructed for calculating the residual corresponding to the battery monomer, branch and entire battery system respectively, and a mapping relationship corresponding to different fault types is established by using the combination of three levels of residuals; Step three, initializing the fault diagnosis process, including setting the initial state of the battery monomer and system including OCV, SOC, algorithm parameters for battery monomer state estimation, and setting time t = 0; Step four, setting t = t + 1, executing the state estimation algorithm to update the state variables of the battery monomer, and calculating the OCV, SOC, and polarization voltage parameters; Step five, using the parameters calculated in step four, inputting the fault diagnosis model established in step two to calculate the three levels of residuals; by comparing each residual with the given threshold, each residual is converted into a logic quantity containing only 0 and 1, defined as a residual flag quantity; Step six, using the combination of residual flag quantities to determine whether a fault has occurred, if all elements of the residual flag quantity are 0, no fault has occurred, and return to step four; if there is an element in the residual flag quantity that is not 0, go to step seven; Step seven, using the mapping relationship obtained in step two to convert into a fault mapping table corresponding to different residual flag quantity combinations, and obtaining the fault type and positioning result by looking up the table, outputting the fault diagnosis conclusion and returning to step four.

2. The method of claim 1, wherein: In step one, the effects of temperature correction, aging, and SOC change on other parameters are also obtained to improve the accuracy of subsequent modeling.

3. The method of claim 1, wherein: In step two, at least an equivalent circuit model capable of describing the current capacity, polarization state, such as SOC, OCV, and polarization voltage of the battery monomer is selected for battery monomer modeling; The specific form of the three levels of residuals r is defined as follows: r＝[r1,r2,r3] Wherein, the first element monomer residual r1 is a p x s matrix, p is the number of parallel branches, and s is the number of series monomers on each branch, so r1 corresponds to each monomer in the battery system; the second element branch residual r2 is a vector with length p, corresponding to each branch; the third element system residual r3 is a scalar; the calculation formula of each element in the residual is as follows: wherein k is the time, subscript i represents the branch to which it belongs, i = 1, 2, 3, …, p, subscript j represents the jth monomer on a branch, j = 1, 2, 3, …, s; OCV, R, U D are respectively the open-circuit voltage of the battery monomer, the ohmic internal resistance, the polarization voltage, y I , y U , y Ii , y Uij are respectively the total current, the total voltage, and the sensor readings of the branch current and the monomer voltage.

4. The method of claim 1, wherein: In step four, real-time voltage and current data are used as the input of the state estimation algorithm, and state estimation is performed at a sampling interval of Δt, and the estimation result is corrected using an estimation error feedback link.

5. The method of claim 1, wherein: In step six, after calculating the three levels of residuals and comparing them with the given threshold, the residual flag quantity combination r' containing only 0 and 1 elements is obtained; in step seven, the fault characteristic quantity f with the same combination form is found in the mapping table converted from the mapping relationship between the combination of three levels of residuals and different fault types, i.e. the diagnosis of the fault location of different fault types is realized.

Images