Battery charging and discharging system fault diagnosis method based on polyhedral space filtering

Abstract

The application discloses a battery charging and discharging system fault diagnosis method based on polytope space filtering, belongs to the field of state estimation and fault detection, and comprises the following steps: constructing a discrete state space model considering system time lag and faults, performing state augmentation after the construction, and obtaining a discrete state space reconstruction model; introducing Lyapunov a function and a performance index to construct an observer; performing polytope space expression on the augmented state estimation value of the observer output, constructing a target observer based on a polytope space Kalman filter and real-time observation residual, separating system state estimation and system fault estimation, calculating the envelope interval of the system fault estimation, and performing fault diagnosis. Through the construction of a three-filter structure including state augmentation, ZKF optimization and dynamic residual generation, the limitation of traditional single-stage filters in time-lag systems and fault detection is broken through, and the overall improvement of state estimation accuracy, fault detection sensitivity and system robustness is realized.

Description

Technical Field

[0001] This invention relates to a fault diagnosis method for battery charging and discharging systems based on multi-cell space filtering, belonging to the field of state estimation and fault detection. Background Technology

[0002] With the rapid development of industrial automation and intelligence, fault detection technology has been widely applied in fields such as power systems, aerospace, manufacturing, automotive engineering, and chemicals, becoming an important means to ensure system safety and reliability. However, complex industrial systems are often accompanied by problems such as network transmission delays and equipment operation interference, which challenge the accuracy of system state estimation and fault detection.

[0003] Traditional state estimation and detection methods mainly include analytical model-based, signal processing-based, and knowledge-based detection methods. Among them, analytical model-based detection methods are widely used due to their high sensitivity, with classic algorithms including Kalman filters and particle filters. However, these methods typically assume that process noise and interference follow specific distribution patterns, but real-world systems often encounter non-random interference that is difficult to describe using statistical rules, leading to inaccurate or even failed state estimations. Furthermore, traditional system modeling often ignores the effect of system time delays. When system time delays are significant, this can lead to inaccurate state estimations or even estimation divergence.

[0004] Currently, state estimation and fault detection methods mainly include interval set inversion observers and multi-cell H... Observer. Multicellular H The observer can be found in "C. Li, F. Zhu, and S. Guo, "Interval estimation and fault detection for switched nonlinear systems based on zonotope method," Asian. J. Control, vol. 25, pp. 1420–1431, Sept. 2023." This method applies the multicellular approach to H... The observer's error dynamic system achieves accurate system state interval estimation in the fault-free condition by iteratively calculating the minimum enclosing polyhedron of the system state vector. However, the residuals it constructs contain disturbances, resulting in lower accuracy compared to the method in this application, making it unsuitable for direct fault detection. Furthermore, because this method does not fully consider the impact of system time delays on state estimation, it exhibits significant errors when dealing with systems containing time delays. Existing residual-based interval estimation methods can develop new fault detection strategies, but they do not consider the state estimation and fault detection problems in time-delayed systems.

[0005] In contrast, referencing the introduction in "Z. Wang, M. Zhang, Y. Wang, Y. Chen, and Z. Ji, "Guaranteed fault-estimation algorithm based on interval set inversionobserver filtering," *International Journal of Control, Automation, and Systems*, vol. 20, no. 11, pp. 3561–3572, Nov. 2022," the interval set inversion observer designs a minimal conservative interval observer by minimizing the F-norm of the state error and utilizes vector Boolean operations and dimensionality operations to shrink the interval estimation results. This method performs well in handling unknown but bounded disturbances and noise in linear discrete-time systems, but it faces several challenges. First, the computational complexity is high, especially in high-dimensional systems, resulting in long computation times. Second, the memory requirements are significant, particularly when performing multiple binary splits, posing a bottleneck for embedded systems. Furthermore, despite employing a minimal conservative observer, the impact of system time delays on state estimation is not fully considered, leading to overly conservative estimations in some cases, affecting fault detection sensitivity. Summary of the Invention

[0006] To address at least one of the aforementioned problems, this invention provides a fault diagnosis method for battery charging and discharging systems based on multi-cell space filtering, the technical solution of which is as follows:

[0007] As one aspect of the present invention, a fault diagnosis method for a battery charging and discharging system based on multi-cell space filtering is provided, comprising:

[0008] For the battery charging and discharging system, a discrete state-space model is constructed that considers the time delay and faults of the battery charging and discharging system.

[0009] State augmentation is performed on the discrete state-space model to obtain a reconstructed discrete state-space model.

[0010] For the discrete state-space reconstruction model, construct Observer, and introduce Lyapunov Functions and Performance metrics, calculation The coefficient matrix of the observer;

[0011] right The augmented state estimate of the battery charging and discharging system output by the observer is expressed in a multicellular space.

[0012] Based on multicellular space expression The observer, based on a multi-cell Kalman filter, constructs the target by observing the residuals in real time. Observer, target The observer's state estimation multicell boundary is smaller than The observer's state estimates the multicellular space boundary and the target. The observer's output is the target augmented state estimate of the battery charging and discharging system;

[0013] Separate the system state estimate and the system fault estimate from the target augmented state estimate, and calculate the envelope interval of the system fault estimate to obtain the upper and lower bounds of the envelope interval of the system fault estimate;

[0014] Based on the upper and lower bounds of the envelope interval of the system fault estimation, the fault occurrence of the battery charging and discharging system is analyzed, and the fault diagnosis results of the battery charging and discharging system are obtained.

[0015] Furthermore, the expression for the discrete state-space model is:

[0016]

[0017] in, express The status of the battery charging and discharging system at all times. express The status of the battery charging and discharging system at all times. , express The measurement output at time, , Indicates a known input. This indicates a malfunction in the battery charging and discharging system. , , A , , B , C , D , E and F Each represents a coefficient matrix of a given preset dimension. h Represents a known constant time delay. , , , , , , , , This represents unknown but bounded process noise. This represents unknown but bounded measurement noise. , , Represents the set of real numbers. Represents the set of positive integers. express A column vector of n real elements express A column vector of n real elements express A column vector of n real elements express A column vector of n real elements express A column vector of n real elements express A column vector of n real elements The dimension is The matrix, The dimension is The matrix, The dimension is The matrix, The dimension is The matrix, The dimension is The matrix.

[0018] Furthermore, state augmentation is performed on the discrete state-space model to obtain a reconstructed discrete state-space model, including:

[0019] Introducing augmented state vectors Rewrite the above discrete state-space model to obtain the intermediate discrete state-space model. The expression of the intermediate discrete state-space model is as follows:

[0020]

[0021] in, , , , , , , , express A column vector of n real elements;

[0022] Introducing augmented state vectors Rewrite the discrete state space intermediate model to obtain the discrete state space reconstructed model. The expression of the discrete state space reconstructed model is:

[0023]

[0024] in, , , , , , , express A column vector of n real elements Represents the identity matrix.

[0025] Furthermore, The observer's expression is:

[0026]

[0027] in, express The augmented state estimate at time 10:00. express The measurement output at time, S , J , L Both represent coefficient matrices. , , , ,

[0028] , , P Denotes the pairwise positive definite matrix in the Lyapunov function. W , Y Each represents the coefficient matrix to be determined. λ , μ Both represent coefficient matrices. , , , , The dimension is The matrix, The dimension is The matrix.

[0029] Furthermore, regarding The state estimates of the battery charging and discharging system output by the observer are expressed in a multicellular space, including:

[0030] According to the given System status at all times sum coefficient matrix S , J , L ,get The expression for the augmented state estimate of the system state at time t is:

[0031]

[0032] in, , , Indicates package The coordinate vector of the center point of the multicell representing the system state at any given time. Indicates package The shape matrix of the multicell of the system state at any given time. Indicates package The coordinate vector of the center point of the multicell representing the system state at any given time. Indicates package The shape matrix of the multicell of the system state at any given time. The multicell generation matrix representing the measurement noise of the package system. The multi-cell generation matrix represents the process noise of the wrapping system.

[0033] Furthermore, the goal The observer's expression is:

[0034]

[0035] in, , , , This represents the covariance in Kalman filtering. , and All of these represent intermediate parameters;

[0036] In a given At time +1, the system state is... hour, Momentary Goal The expression for the target augmented state estimate output by the observer is:

[0037]

[0038] in, , Indicates package The coordinate vector of the target center point of the multicell in the system state at any given time. Indicates package The target shape matrix of the multicell in the system state at any given time. I Represents the identity matrix. express Time-amplified state vector Estimated multicell space.

[0039] Furthermore, the expression for system state estimation is:

[0040]

[0041] in, This indicates the separation based on the target augmented state estimate. System state estimation at time t. M Represents the augmented state vector The system state estimation matrix extracted from [the data]. , , express 3D identity matrix N Represents the augmented state vector The system fault estimation matrix extracted from [the data]. ,

[0042] , express 3D identity matrix R Represents the augmented state vector The augmented state vector extracted from The matrix, , express 3D identity matrix;

[0043] The expression for system fault estimation is:

[0044]

[0045] in, This indicates the separation based on the target augmented state estimate. k System fault estimation at any given time;

[0046] The expression for the envelope interval of system fault estimation is:

[0047]

[0048] in, This represents the upper bound of the envelope interval for system fault estimation. This represents the lower bound of the envelope interval for system fault estimation. , , Representing multicellular space Medium Shape Generating Matrix The All elements Frobenius The sum of norms.

[0049] Furthermore, based on the upper and lower bounds of the envelope interval of the system fault estimation, the fault occurrence of the battery charging and discharging system is analyzed to obtain the fault diagnosis results of the battery charging and discharging system, including:

[0050] if This confirms that the battery charging and discharging system is not malfunctioning.

[0051] if or This confirms that a malfunction has occurred in the battery charging and discharging system.

[0052] As another aspect of the present invention, a battery charging and discharging system fault diagnosis device based on multi-cell space filtering is provided, for implementing the above-mentioned battery charging and discharging system fault diagnosis method based on multi-cell space filtering, comprising:

[0053] The model building module is used to construct a discrete state-space model of the battery charging and discharging system that considers the time delay and faults of the battery charging and discharging system.

[0054] The model reconstruction module is used to augment the state of the discrete state-space model to obtain a reconstructed discrete state-space model.

[0055] The observer building module is used to build an observer for a discrete state-space reconstruction model. Observer, and introduce Lyapunov Functions and Performance metrics, calculation The coefficient matrix of the observer;

[0056] Multicellular space expression module, used for... The augmented state estimate of the battery charging and discharging system output by the observer is expressed in a multicellular space.

[0057] Target The observer building module is used to build upon the multicellular representation. The observer, based on a multi-cell Kalman filter, constructs the target by observing the residuals in real time. Observer, target The observer's state estimation multicell boundary is smaller than The observer's state estimates the multicellular space boundary and the target. The observer's output is the target augmented state estimate of the battery charging and discharging system;

[0058] The envelope interval calculation module is used to separate the system state estimate and the system fault estimate from the augmented state estimate, and to calculate the envelope interval of the system fault estimate, thereby obtaining the upper and lower bounds of the envelope interval of the system fault estimate.

[0059] The judgment module is used to analyze the fault occurrence of the battery charging and discharging system based on the upper and lower bounds of the envelope interval of the system fault estimation, and to obtain the fault diagnosis results of the battery charging and discharging system.

[0060] As another aspect of the present invention, a battery charging and discharging system is provided, comprising:

[0061] The battery charging and discharging system body is used to control the charging and discharging process of the battery, manage and regulate the battery power, and enable the battery to work within a predetermined charging and discharging range.

[0062] The aforementioned battery charging and discharging system fault diagnosis device based on multi-cell space filtering is used to diagnose faults in the battery charging and discharging system.

[0063] A DC-DC circuit system is used to convert the input voltage to a predetermined battery charge / discharge voltage range.

[0064] The beneficial effects of this invention are:

[0065] The fault diagnosis method for battery charging and discharging systems based on multi-cell space filtering provided by this invention overcomes the limitations of traditional single-stage filters in time-delay systems and fault detection by constructing a triple collaborative filtering structure including state augmentation, multi-cell space Kalman filter (ZKF) optimization, and dynamic residual generation. It innovatively achieves a comprehensive balance optimization of state estimation accuracy, fault detection sensitivity, and system robustness. The core multi-cell space Kalman filter uses a set of Zonotope cells to describe the state space and achieves a balance between estimation interval contraction and computational complexity through the Frobenius norm radius minimization criterion. Further, it introduces… Performance metrics were assessed, and a robust observer was designed to suppress uncertainty interference. State augmentation techniques were combined to effectively decouple time delays and fault signals, while simplifying the observer design process. To enhance fault detection capabilities, a detection module based on a dynamic residual generator was constructed, leveraging the geometric properties of the Zonotope set to achieve high sensitivity and low false negative rate. This further optimized the balance between robustness and accuracy in traditional filtering methods, providing an efficient solution for state estimation and fault detection in complex time-delay systems. Attached Figure Description

[0066] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0067] Figure 1 This is a simplified flowchart of the battery charging and discharging system fault diagnosis method based on multi-cell space filtering provided in Embodiment 1 of the present invention.

[0068] Figure 2 This is a schematic diagram of the circuit structure of the Buck-Boost converter system provided in Embodiment 1 of the present invention.

[0069] Figure 3A The state variables obtained in the absence of system faults using Embodiment 1 of the present invention, obtained by combining Embodiment 1 with existing methods. Estimated curves and state variables Simulation comparison chart of the estimated curves;

[0070] Figure 3B The state variables obtained by using Embodiment 1 of the present invention and existing methods when a system failure occurs are provided in Embodiment 1 of the present invention. Estimated curves and state variables Simulation comparison chart of the estimated curves;

[0071] Figure 3C The state variables obtained by using Embodiment 1 of the present invention and existing methods when a system failure occurs are provided in Embodiment 1 of the present invention. and state variables Simulation comparison chart of the estimated curves;

[0072] Figure 3D The method provided in Embodiment 1 of this invention, obtained by combining Embodiment 1 with existing methods, is as follows: k =50, k =100, k =150, k Simulation comparison chart of the estimated range obtained when =200. Detailed Implementation

[0073] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments of the present invention will be described in further detail below with reference to the accompanying drawings.

[0074] Example 1:

[0075] This embodiment uses a Buck-Boost converter system as an example, combined with... Figure 1 and Figure 2 This paper describes a fault diagnosis method for battery charging and discharging systems based on multi-cell space filtering. A Buck-Boost converter system is a power electronic device used for voltage boosting and bucking, widely used in DC power supply systems. It mainly includes an input power supply, switching elements, diodes, an energy storage inductor, a capacitor, and a load. During operation, the input power supply controls energy transfer through switching elements, accumulates energy in the energy storage inductor, is rectified by diodes, and filtered by capacitors to ultimately provide a stable output voltage. This embodiment is based on the capacitor voltage during the operation of the Buck-Boost converter system. and inductor current The current value is used to estimate the value at the next time step, and to determine whether a fault has occurred in the current system. Specifically, this includes:

[0076] S100: For the battery charging and discharging system, construct a discrete state-space model that considers the time delay and faults of the battery charging and discharging system.

[0077] Specifically, step S100 includes:

[0078] S110. Establish a mathematical model for the Buck-Boost converter system, as shown in formula (1):

[0079] (1)

[0080] in, Indicates inductor current. Indicates capacitor voltage. Indicates voltage. Indicates the load of the circuit. Indicates capacitance. Indicates the inductance value. This indicates the duty cycle of the transistor PWM signal;

[0081] S120. Discretize the mathematical model of the Buck-Boost converter system represented by formula (1), as shown in formula (2):

[0082]

[0083] in, , Represents the natural constant. Indicates the sampling period of a discrete system;

[0084] S130. Based on the mathematical model of the Buck-Boost converter system, construct the corresponding discrete state-space model considering the time delay and faults of the battery charging and discharging system, as shown in formula (3):

[0085] (3)

[0086] in, express The status of the battery charging and discharging system at all times. express The status of the battery charging and discharging system at all times. , express The measurement output at time, , Indicates a known input. This indicates a malfunction in the battery charging and discharging system. , , A , , B ,C , D , E and F Each represents a coefficient matrix of a given preset dimension. h Represents a known constant time delay. , , , , , , , , This represents unknown but bounded process noise. This represents unknown but bounded measurement noise. , , Represents the set of real numbers. Represents the set of positive integers. express A column vector of n real elements express A column vector of n real elements express A column vector of n real elements express A column vector of n real elements express A column vector of n real elements express A column vector of n real elements The dimension is The matrix, The dimension is The matrix, The dimension is The matrix, The dimension is The matrix, The dimension is The matrix.

[0087] It should be noted that, The measured output refers to the system result that can be directly observed or measured; specifically, it refers to the current in the inductor of the Buck-Boost converter system that can be directly measured during operation. and the voltage of the capacitor .

[0088] Process noise and measuring noise The following requirements of formula (4) must be met:

[0089] (4)

[0090] in, and Both represent unit boxes with preset dimensions. The process noise multicell generation matrix is ​​represented by the matrix. This represents the measurement noise multicell generation matrix.

[0091] S200. Perform state augmentation on the discrete state-space model to obtain a discrete state-space reconstructed model.

[0092] Specifically, embodiments of the present invention will address system failures. The discrete state-space model of the battery charging and discharging system is incorporated as part of the state estimation, and then augmented state vectors are used. , Simplify the discrete state-space reconstruction model, including:

[0093] S210, Introducing Augmented State Vector The above discrete state space model is rewritten to obtain the intermediate discrete state space model, as shown in the following formula (5):

[0094] (5)

[0095] in, , , , , , , , express A column vector of n real elements;

[0096] S220. Introduce augmented state vectors based on the discrete state-space intermediate model. The intermediate model in the discrete state space is rewritten to obtain the reconstructed model in the discrete state space, as shown in formula (6):

[0097] (6)

[0098] in, , , , , , , express A column vector of n real elements Represents the identity matrix.

[0099] By introducing Performance metrics include designing robust observers to suppress uncertainty disturbances, effectively decoupling time delays and fault signals by combining state augmentation techniques, and simplifying the observer design process.

[0100] S300. For the discrete state-space reconstruction model, construct... Observer, and introduce Lyapunov (Lyapunov) functions and Performance metrics, calculation The coefficient matrix of the observer;

[0101] Specifically, step S300 includes:

[0102] S310, Construction Observer, The observer is shown in the following formula (7):

[0103] (7)

[0104] in, express The augmented state estimate at time 10:00. express The measurement output at time, S , J , L Both represent coefficient matrices. , , , ,

[0105] , , P Denotes the pairwise positive definite matrix in the Lyapunov function. W , Y Each represents the coefficient matrix to be determined. λ , μ Both represent coefficient matrices. , , , , The dimension is The matrix, The dimension is The matrix.

[0106] S320, Introduction Lyapunov functions and Observer performance indicators J Find them separately S , J , L Optimal coefficient matrix:

[0107] First, introduce the residual parameter. We can obtain the following formula (8):

[0108] (8)

[0109] Next, introduce Lyapunov function The following formula (9) can be obtained:

[0110] (9)

[0111] Then select the following formula (10) as Observer performance metrics:

[0112] (10)

[0113] in, γ A scalar coefficient representing the degree of contraction of a multicellular body;

[0114] according to Since it is always less than 0, formula (11) can be derived:

[0115] (11)

[0116] For the formula (7) shown The observer substitutes formula (9) into formula (11) to construct the LMI inequality, making the solution to the current problem more efficient. k Time-space residuals The problem of minimizing the gain of the state estimation error is transformed into the problem of finding the optimal solution of LMI. The LMI inequality is shown in equation (12):

[0117] (12)

[0118] in, , , , , M Represents the augmented state vector The system state estimation matrix extracted from [the data]. express 3D identity matrix express 3D identity matrix express 3D identity matrix.

[0119] Seek S , J , L The optimal coefficient matrix is ​​shown in formula (13):

[0120] (13)

[0121] in, , .

[0122] S400, to The state estimates of the battery charging and discharging system output by the observer are expressed in a multicellular space.

[0123] That is to Augmented state estimate of the battery charge-discharge system output by the observer Perform multicellular space expression, that is, give the current k Time to the k The iterative form of the multi-cell space at time +1;

[0124] Specifically, given Time system augmented state Then there exists a coefficient matrix. , , ,make The augmented state estimate of the system at time t satisfies the following formula (14):

[0125] (14)

[0126] in, , , Indicates package The coordinate vector of the center point of the multicell representing the system state at any given time. Indicates package The shape matrix of the multicell of the system state at any given time. Indicates package The coordinate vector of the center point of the multicell representing the system state at any given time. Indicates package The shape matrix of the multicell of the system state at any given time. The multicell generation matrix representing the measurement noise of the package system. The multi-cell generation matrix represents the process noise of the wrapping system.

[0127] It should be understood that in formula (7) This is the recursive formula for a discrete system. If the multicell space is substituted as the state variable of the previous time step into formula (7), the state estimate of the multicell space range for the next time step can be obtained, as shown in formula (14) here, and specifically as shown in formula (15) below:

[0128]

[0129] S500, based on multicellular space expression The observer, based on a multi-cell Kalman filter, constructs the target by observing the residuals in real time. Observer, target The observer's state estimation multicell boundary is smaller than The observer's state estimates the multicellular space boundary and the target. The observer's output is the target augmented state estimate of the battery charging and discharging system;

[0130] Specifically, the goal The observer is designed as an optimized estimate of the augmented state vector, and its structure is shown in Equation (16):

[0131]

[0132] in, , , , This represents the covariance in Kalman filtering. , and All of these represent intermediate parameters;

[0133] Given Constantly expanding state At that time, Momentary Goal The target augmented state estimate output by the observer is given by the following formula (17):

[0134] (17)

[0135] in, , Indicates package The coordinate vector of the target center point of the multicell in the system state at any given time. Indicates package The target shape matrix of the multicell in the system state at any given time. I Represents the identity matrix. express Time-amplified state vector Estimated multicell space.

[0136] This invention describes the state space using a multi-cell Kalman filter, and through... Frobenius The (Flobenius) radius minimization criterion constructs a more accurate model from real-time observation residuals. The observer achieves a balance between shrinking the estimation interval and computational complexity, further narrowing the multi-cell boundary of the state estimation, and can greatly improve the accuracy of system state estimation.

[0137] S600. Separate the system state estimate and the system fault estimate from the target augmented state estimate, calculate the envelope interval of the system fault estimate, and obtain the upper and lower bounds of the envelope interval of the system fault estimate.

[0138] Specifically, k The estimated value of the target augmented state at time +1 is: ,but k State estimation at time +1 As in formula (18):

[0139] (18)

[0140] in, This indicates the separation based on the target augmented state estimate. System state estimation at time t. M Represents the augmented state vector The system state estimation matrix extracted from [the data]. , , express 3D identity matrix N Represents the augmented state vector The system fault estimation matrix extracted from [the data]. ,

[0141] , express 3D identity matrix R Represents the augmented state vector The augmented state vector extracted from The matrix, , express 3D identity matrix;

[0142] System fault estimation is shown in formula (19):

[0143] (19)

[0144] in, This indicates the separation based on the target augmented state estimate. k System fault estimation at any given time;

[0145] For any multicellular body There exists a minimum interval envelope. ,make Therefore, the expression for the envelope interval of system fault estimation is derived as shown in formula (20):

[0146] (20)

[0147] in, This represents the upper bound of the envelope interval for system fault estimation. This represents the lower bound of the envelope interval for system fault estimation. , , Representing multicellular space Medium Shape Generating Matrix The All elements Frobenius The sum of norms.

[0148] S700. Based on the upper and lower bounds of the envelope interval of the system fault estimation, analyze the fault occurrence of the battery charging and discharging system and obtain the fault diagnosis results of the battery charging and discharging system.

[0149] The fault diagnosis criteria for time-delayed battery charging and discharging systems are as follows:

[0150] If it exists , then it means ;otherwise, .when "Time" indicates the moment. No system failure occurs at this time On the contrary, when When this occurs, it indicates that a system failure has occurred. .

[0151] The battery charging and discharging system fault diagnosis method based on multi-cell space filtering provided in this invention overcomes the limitations of traditional single-stage filters in time-delay systems and fault detection by constructing a triple collaborative filtering structure including state augmentation, ZKF optimization, and a dynamic residual generator. It innovatively achieves comprehensive optimization of state estimation accuracy, fault detection sensitivity, and system robustness. The core ZKF utilizes a Zonotope set to describe the state space and achieves a balance between shrinking the estimation interval and computational complexity through the Frobenius radius minimization criterion. Furthermore, H... Performance metrics were assessed, and a robust observer was designed to suppress uncertainty interference. State augmentation techniques were combined to effectively decouple time delays and fault signals, while simplifying the observer design process. To enhance fault detection capabilities, a detection module based on a dynamic residual generator was constructed, leveraging the geometric properties of the Zonotope set to achieve high sensitivity and low false negative rate. This further optimized the balance between robustness and accuracy in traditional filtering methods, providing an efficient solution for state estimation and fault detection in complex time-delay systems.

[0152] To verify the effectiveness of the battery charging and discharging system fault diagnosis method based on multi-cell space filtering provided in this embodiment, the method is compared with existing methods. The experimental simulation results are as follows: Figure 3A , Figure 3B and Figure 3C As shown, where, Figure 3A State variables of the Buck-Boost converter system when the system is fault-free. Estimated curves and state variables The estimated curve, Figure 3B State variables of the Buck-Boost converter system when a system failure occurs. Estimated curves and state variables The estimated curve, Figure 3C State variables of the Buck-Boost converter system when a system failure occurs. and Estimated curve. Figure 3A , Figure 3B and Figure 3C In the diagram, solid lines represent the true values, thick dashed lines represent the estimated upper and lower bounds obtained by the method proposed in this invention, and dotted and thin dashed lines represent the estimated upper and lower bounds of other existing methods. It can be seen that, regardless of whether a system fault has occurred, the upper and lower bounds of each method can encompass the true state of the Buck-Boost converter system. However, the TS-ZKF H method for fault diagnosis of battery charging and discharging systems based on multi-cell space filtering provided in this embodiment... The state estimation interval of the observer is compared with the interval set inversion observer (MCIO H) observer) and multicellular observer (zonotope H The narrower boundary of the observer indicates that the battery charging and discharging system fault diagnosis method based on multi-cell space filtering in this embodiment can achieve more accurate fault estimation, with a smooth and compact effect that closely approximates the true value. At the same time, the boundary of the fault estimation changes as the fault occurs, further making it more accurate and enabling earlier and more precise detection of faults.

[0153] Figure 3D The estimation range curves of the method provided in this embodiment and the existing method are given at different times (k = 50, k = 100, k = 150, k = 200). Figure 3D middle, The dots represent the true values ​​at the current moment, the thick dashed lines represent the estimation range obtained by the method provided in this embodiment, and the dotted lines and thin dashed lines represent the existing method zonotope H. The estimated ranges obtained by observer and MICO are shown. It can be seen that the estimated ranges of the method proposed in this invention are significantly close to the true values ​​at all time steps, and the state estimation ranges at different time points are more compact; when k=100, only TS-ZKF H... Observer and zonotope H The observer's estimation range can detect the occurrence of faults, while the MCIO algorithm has a certain false negative rate. This is because TS-ZKF H The observer's estimated range is greater than that of TS-ZKF H The observer and MICO estimates have a small range, making them more sensitive to fault detection.

[0154] Example 2:

[0155] This invention provides a battery charging and discharging system fault diagnosis device based on multi-cell space filtering, used to implement the above-mentioned battery charging and discharging system fault diagnosis method based on multi-cell space filtering, including:

[0156] The model building module is used to construct a discrete state-space model of the battery charging and discharging system that considers the time delay and faults of the battery charging and discharging system.

[0157] The model reconstruction module is used to augment the state of the discrete state-space model to obtain a reconstructed discrete state-space model.

[0158] The observer building module is used to build an observer for a discrete state-space reconstruction model. Observer, and introduce Lyapunov Functions and Performance metrics, calculation The coefficient matrix of the observer;

[0159] Multicellular space expression module, used for... The augmented state estimate of the battery charging and discharging system output by the observer is expressed in a multicellular space.

[0160] Target The observer building module is used to build upon the multicellular representation. The observer, based on a multi-cell Kalman filter, constructs the target by observing the residuals in real time. Observer, target The observer's state estimation multicell boundary is smaller than The observer's state estimates the multicellular space boundary and the target. The observer's output is the target augmented state estimate of the battery charging and discharging system;

[0161] The envelope interval calculation module is used to separate the system state estimate and the system fault estimate from the augmented state estimate, calculate the envelope interval of the system fault estimate, and obtain the upper and lower bounds of the envelope interval of the system fault estimate.

[0162] The judgment module is used to analyze the fault occurrence of the battery charging and discharging system based on the upper and lower bounds of the envelope interval of the system fault estimation, and to obtain the fault diagnosis results of the battery charging and discharging system.

[0163] The battery charging and discharging system fault diagnosis device based on multi-cell space filtering provided in this invention overcomes the limitations of traditional single-stage filters in time-delay systems and fault detection by constructing a triple collaborative filtering structure including state augmentation, ZKF optimization, and dynamic residual generation. It innovatively achieves comprehensive optimization of state estimation accuracy, fault detection sensitivity, and system robustness. The core ZKF uses a Zonotope set to describe the state space and achieves a balance between shrinking the estimation interval and computational complexity through the Frobenius radius minimization criterion. Furthermore, H... Performance metrics were assessed, and a robust observer was designed to suppress uncertainty interference. State augmentation techniques were combined to effectively decouple time delays and fault signals, while simplifying the observer design process. To enhance fault detection capabilities, a detection module based on a dynamic residual generator was constructed, leveraging the geometric properties of the Zonotope set to achieve high sensitivity and low false negative rate. This further optimized the balance between robustness and accuracy in traditional filtering methods, providing an efficient solution for state estimation and fault detection in complex time-delay systems.

[0164] The specific principle of the battery charging and discharging system fault diagnosis device based on multi-cell space filtering provided in this embodiment of the invention can be found in the principle description of the battery charging and discharging system fault diagnosis method based on multi-cell space filtering, and will not be repeated here.

[0165] Example 3:

[0166] This invention provides a battery charging and discharging system, comprising:

[0167] The battery charging and discharging system body is used to control the charging and discharging process of the battery, manage and regulate the battery power, and enable the battery to work within a predetermined charging and discharging range.

[0168] The aforementioned battery charging and discharging system fault diagnosis device based on multi-cell space filtering is used to diagnose faults in the battery charging and discharging system.

[0169] A DC-DC circuit system is used to convert the input voltage to a predetermined battery charge / discharge voltage range.

[0170] The embodiments of the present invention, through a battery charging and discharging system fault diagnosis device based on multi-cell space filtering, overcome the limitations of traditional single-stage filters in time-delay systems and fault detection. When performing state estimation and fault detection on the battery charging and discharging system itself, it can achieve comprehensive optimization of state estimation accuracy, fault detection sensitivity and system robustness.

[0171] The specific principle of the battery charging and discharging system provided in this embodiment of the invention can be found in the principle description of the battery charging and discharging system fault diagnosis method based on multi-cell space filtering, and will not be repeated here.

[0172] Some steps in the embodiments of the present invention can be implemented using software, and the corresponding software program can be stored in a readable storage medium, such as an optical disc or a hard disk.

[0173] The above are merely preferred embodiments of the present invention and are 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 within the protection scope of the present invention.

Claims

1. A battery charging and discharging system fault diagnosis method based on polytope space filtering, characterized in that, Comprising: For the battery charging and discharging system, a discrete state space model considering the time delay of the battery charging and discharging system and the fault of the battery charging and discharging system is constructed; The expression of the discrete state space model is: wherein denotes the state of the battery charging and discharging system at time denotes the state of the battery charging and discharging system at time , denotes the measured output at time , denotes the known input denotes a fault of the battery charging and discharging system , , A , A h , B , C , D , E and F denote a coefficient matrix of given preset dimension h denotes a known constant time delay , , , , , , , , denotes an unknown but bounded process noise denotes an unknown but bounded measurement noise , , denotes a set of real numbers denotes a set of positive integers denotes a column vector of real elements denotes a column vector of real elements denotes a column vector of real elements denotes a column vector of real elements denotes a column vector of real elements denotes a column vector of real elements denotes a matrix of dimension denotes a matrix of dimension denotes a matrix of dimension denotes a matrix of dimension denotes a matrix of dimension​​​​ matrix of the equation The state augmentation is performed on the discrete state space model to obtain a discrete state space reconstruction model; For the discrete state space reconstruction model, construct Observer, and introduce Lyapunov Functions and Performance metrics, calculation The coefficient matrix of the observer; The The augmented state estimate of the battery charging and discharging system is expressed in a polytope space. According to the polytope space expression, the target The observer constructs the target The observer, the target The state estimation polytope space boundary of the observer is smaller than the target The state estimation polytope space boundary of the observer, the target The output of the observer is the target augmented state estimation value of the battery charging and discharging system; The system state estimate and the system fault estimate are separated from the target augmented state estimate, and the envelope interval of the system fault estimate is calculated to obtain the upper and lower bounds of the envelope interval of the system fault estimate; According to the upper and lower bounds of the envelope interval of the system fault estimate, the fault occurrence of the battery charging and discharging system is analyzed to obtain the fault diagnosis result of the battery charging and discharging system.

2. The method of claim 1, wherein, The state augmentation is performed on the discrete state space model to obtain a discrete state space reconstruction model, comprising: Introducing an augmented state vector Rewriting the above discrete state space model, a discrete state space intermediate model is obtained, and the expression of the discrete state space intermediate model is: wherein , , , , , , , denotes a column vector of real elements; Introducing an augmented state vector Rewriting the discrete state space intermediate model, a discrete state space reconstructed model is obtained, and the expression of the discrete state space reconstructed model is as follows: wherein , , , , , , denotes a column vector of real elements, denotes the identity matrix.

3. The method of claim 2, wherein, The The expression of the observer is: wherein denotes augmented state estimate at time instant denotes measurement output at time instant S , J , L each denote a coefficient matrix, , , , , , , P denotes a positive definite matrix in the Lyapunov function, W , Y denote the coefficient matrix to be solved, Lambda , Mu denote the coefficient matrix, , , , , denotes a matrix of dimension , denotes a matrix of dimension .

4. The method of claim 3, wherein, The pair of the The augmented state estimate of the battery charging and discharging system output by the observer is expressed in a polytope space, comprising: According to the given System state at time And the coefficient matrix S , J , L , obtain The expression of the augmented state estimation value of the system state at time wherein, , , represents a center point coordinate vector of the polytope wrapping the system state at time instant, represents a shape matrix of the polytope wrapping the system state at time instant, represents a center point coordinate vector of the polytope wrapping the system state at time instant, represents a shape matrix of the polytope wrapping the system state at time instant, represents a polytope generation matrix wrapping system measurement noise, represents a polytope generation matrix wrapping system process noise.

5. The method of claim 4, wherein, The target The expression of the observer is: wherein , , , denotes the covariance in the Kalman filter, , and all denote intermediate variables; At a given The system state at time , The target The expression for the target augmented state estimate from the observer output is wherein, , represents a polytope that encloses the target center point coordinate vector of the polytope at the system state, represents a polytope that encloses the target shape matrix of the polytope at the system state, I represents an identity matrix, represents the augmented state vector at the time the estimated polytopic space.

6. The method of claim 5, wherein, The expression of the system state estimate is: wherein denotes the system state estimate separated from the target augmented state estimate value at the time instant t, M denotes the matrix of the system state estimate extracted from the augmented state vector at the time instant t, , , denotes the dimensional identity matrix, N denotes the matrix of the system fault estimate extracted from the augmented state vector at the time instant t, , , denotes a dxd identity matrix, R denotes the augmented state vector extracted from the augmented state vector , , denotes a dxd identity matrix; The expression of the system fault estimate is: wherein, represents the system fault estimate at the time instant separated out from the augmented state estimate k at the time instant. The expression of the envelope interval of the system fault estimate is: in, This represents the upper bound of the envelope interval for system fault estimation. This represents the lower bound of the envelope interval for system fault estimation. , , Representing multicellular space Medium Shape Generating Matrix The All elements Frobenius The sum of norms.

7. The method of claim 6, wherein, According to the upper and lower bounds of the envelope interval of the system fault estimate, the fault occurrence of the battery charging and discharging system is analyzed to obtain the fault diagnosis result of the battery charging and discharging system, comprising: If , it is determined that the battery charging and discharging system has not failed; If or a fault in the battery charging and discharging system is determined.

8. A device for diagnosing a fault of a battery charging / discharging system based on a polytope space filter, for implementing the method for diagnosing a fault of a battery charging / discharging system based on a polytope space filter according to any one of claims 1 to 7, characterized by, Comprising: A model construction module is configured to, for the battery charging and discharging system, construct a discrete state space model considering the time delay of the battery charging and discharging system and the fault of the battery charging and discharging system; A model reconstruction module is configured to perform state augmentation on the discrete state space model to obtain a discrete state space reconstruction model; an observer construction module for constructing, for the discrete state space model, an observer and introducing Lyapunov a function with a performance index, computing the coefficient matrix of the observer; a polytope space representation module for representing the a polytope space representation of an augmented state estimate of the battery charging and discharging system; Objectives an observer construction module for constructing the target an observer based on a multi-hypercube Kalman filter, which constructs the target an observer, the target the state estimation multi-hypercube boundary of the observer is less than the target the state estimation multi-hypercube boundary of the observer, the target the output of the observer is the target augmented state estimation value of the battery charging and discharging system; An envelope interval calculation module is configured to separate the system state estimate and the system fault estimate from the augmented state estimate, and calculate the envelope interval of the system fault estimate to obtain the upper and lower bounds of the envelope interval of the system fault estimate; A judgment module is configured to, according to the upper and lower bounds of the envelope interval of the system fault estimate, analyze the fault occurrence of the battery charging and discharging system to obtain the fault diagnosis result of the battery charging and discharging system.

9. A battery charging and discharging system, characterized by, Comprising: A battery charging and discharging system body is configured to control the charging and discharging process of the battery, manage and regulate the battery power, and make the battery work within a predetermined charging and discharging range; The battery charging and discharging system fault diagnosis device based on the multi-cell space filtering in claim 8 is configured to diagnose the fault of the battery charging and discharging system; A DC-DC circuit system is configured to convert the input voltage to a predetermined battery charging and discharging voltage range.

Images