Heterogeneous power system fault diagnosis method and device based on multilateral interaction, medium and computer equipment

Abstract

The invention discloses a heterogeneous power system fault diagnosis method and device based on multilateral interaction, a medium and computer equipment. The method comprises the following steps: establishing a state space model of a heterogeneous multi-subsystem of the distributed power system; designing a distributed state and fault estimation observer based on an unknown input method, and constructing a dynamic equation for simultaneously estimating a system state and a fault signal; defining a local state estimation error vector and a fault estimation error vector, and deducing a global error system model based on the dynamic equation and the state space model of the observer; performing stability analysis on the global error system model based on the Lyapunov stability theory, and converting a stability condition into an inequality equation set of a linear matrix to obtain a matrix parameter of the observer; and configuring an observer through matrix parameters, and outputting estimation on a system state and a fault signal in real time through the observer so as to complete fault diagnosis. According to the method, the fault diagnosis accuracy of the heterogeneous power system can be improved.

Description

Technical Field

[0001] This application relates to the field of power system condition monitoring and fault diagnosis technology, and in particular to a method, device, medium and computer equipment for fault diagnosis of heterogeneous power systems based on multilateral interaction. Background Technology

[0002] With the large-scale integration of new energy power generation and distributed power sources, the power system structure is gradually evolving from the traditional centralized and homogeneous form to a distributed, heterogeneous, and networked one. Due to differences in equipment type, control strategies, and dynamic characteristics, the nodes in the system are difficult to describe using a unified mathematical model. This structural change poses new challenges to traditional centralized state estimation, fault diagnosis, and control methods, especially when network faults occur, significantly reducing their observation capabilities and reliability.

[0003] Currently, power system state estimation and fault diagnosis mainly rely on centralized processing methods, with state reconstruction based on global measurement data and unified parameter models. However, when network problems such as line disconnections, communication anomalies, or sensor failures occur, the observability of centralized methods decreases sharply. To address this issue, distributed state estimation methods based on multilateral information interaction have emerged in recent years. However, most of these methods are still based on the ideal isomorphic assumption that "all subsystems have the same structure," making it difficult to adapt to the dynamic characteristics of heterogeneous power systems. Furthermore, existing methods often impose strong constraints on fault dynamics, accompanied by complex mathematical conditions, resulting in high computational complexity and insufficient diagnostic capability for time-varying faults. Summary of the Invention

[0004] In view of this, embodiments of this application provide a method, apparatus, medium and computer equipment for fault diagnosis of heterogeneous power systems based on multilateral interaction. The main purpose is to solve the technical problems that existing methods are difficult to adapt to the dynamic characteristics of heterogeneous power systems, have high computational complexity and insufficient diagnostic capability for time-varying faults.

[0005] According to one aspect of this application, a fault diagnosis method for heterogeneous power systems based on multilateral interaction is provided, the method comprising: Establish a state-space model of a heterogeneous multi-subsystem distributed power system, wherein the state-space model is used to describe the line dynamic characteristics, fault dynamic characteristics and coupling interconnection characteristics of each subsystem; For each of the subsystems, an observer for distributed state and fault estimation based on an unknown input method is designed. The observer constructs a dynamic equation that simultaneously estimates the system state and fault signal by introducing the derivative information of the output signal and fusing the state estimates of adjacent subsystems. Define the local state estimation error vector and fault estimation error vector for each subsystem, and derive the global error system model based on the dynamic equation of the observer and the state-space model; Stability analysis of the global error system model is performed based on Lyapunov stability theory, and matrix inequality conditions that make the global error system model asymptotically stable are obtained. The matrix inequality conditions are transformed into a system of linear matrix inequality equations, and the matrix parameters of the observer are obtained by solving the system of linear matrix inequality equations. By configuring the observer using its matrix parameters and by estimating the system state and fault signals in real time during operation, fault diagnosis of the distributed power system can be completed.

[0006] According to another aspect of this application, a fault diagnosis device for heterogeneous power systems based on multilateral interaction is provided, the device comprising: The spatial model building module is used to build a state space model of a heterogeneous multi-subsystem distributed power system, wherein the state space model is used to describe the line dynamic characteristics, fault dynamic characteristics and coupling interconnection characteristics of each subsystem. The observer design module is used to design an observer for distributed state and fault estimation based on an unknown input method for each of the subsystems. The observer constructs a dynamic equation that simultaneously estimates the system state and fault signal by introducing the derivative information of the output signal and fusing the state estimates of adjacent subsystems. The error model building module is used to define the local state estimation error vector and fault estimation error vector of each subsystem, and derive the global error system model based on the dynamic equation of the observer and the state space model. The stability analysis module is used to perform stability analysis on the global error system model based on Lyapunov stability theory, and obtain the matrix inequality conditions that make the global error system model asymptotically stable. The matrix parameter solving module is used to transform the matrix inequality conditions into a system of linear matrix inequality equations, and obtain the matrix parameters of the observer by solving the system of linear matrix inequality equations. The system fault diagnosis module is used to configure the observer through the matrix parameters of the observer, and to complete the fault diagnosis of the distributed power system through the estimation of system status and fault signals output by the observer in real time during operation.

[0007] According to another aspect of this application, a storage medium is provided that stores a computer program thereon, which, when executed by a processor, implements the above-described method for fault diagnosis of heterogeneous power systems based on multilateral interaction.

[0008] According to another aspect of this application, a computer device is provided, including a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, wherein the processor executes the program to implement the above-described method for fault diagnosis of heterogeneous power systems based on multilateral interaction.

[0009] By employing the above technical solutions, the embodiments of this application provide a method, apparatus, medium, and computer equipment for fault diagnosis of heterogeneous power systems based on multilateral interaction. By constructing a state-space model that includes line dynamic characteristics, fault dynamic characteristics, and coupled interconnection characteristics, it can realistically reflect the actual dynamic characteristics of the system, thus providing high-fidelity input for diagnostic calculations. Simultaneously, by introducing a distributed state and fault estimation observer based on an unknown input method, and by fusing the output derivative and neighborhood state information through the observer, it can effectively estimate various complex time-varying faults, including slowly varying and oscillating faults, without making any prior assumptions about fault dynamics, thereby improving the diagnostic capability for time-varying faults. Furthermore, the above method, by designing the observer based on the original state-space model, avoids the computational complexity problems caused by traditional augmented system methods, thus significantly reducing the difficulty of algorithm implementation. Moreover, by transforming the stability conditions of the error system into a system of solvable linear matrix inequality equations, it provides a rigorous mathematical guarantee for the asymptotic convergence of state and fault estimation, thereby eliminating the uncertainty of relying on empirical parameter tuning. Based on this, the above method can effectively enhance the adaptability of multi-node collaborative estimation under complex fault network conditions by constructing a state-space model that includes line dynamics and combining it with an observer structure suitable for heterogeneous environments. This can improve the overall accuracy, robustness and engineering practicality of fault diagnosis.

[0010] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0011] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 A flowchart illustrating a heterogeneous power system fault diagnosis method based on multilateral interaction provided in an embodiment of this application is shown. Figure 2 This paper shows a schematic diagram of the circuit topology of a wind turbine system provided in an embodiment of this application; Figure 3A schematic diagram of the circuit topology of an energy storage system provided in an embodiment of this application is shown; Figure 4 A schematic diagram of the circuit topology of a load system provided in an embodiment of this application is shown; Figure 5 This illustration shows a schematic diagram of the coupling relationship between a source system, an energy storage system, and a load system according to an embodiment of this application. Figure 6 The diagram shows a structural schematic of a heterogeneous power system fault diagnosis device based on multilateral interaction, provided in an embodiment of this application. Detailed Implementation

[0012] The present application will be described in detail below with reference to the accompanying drawings and embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in the embodiments of the present application can be combined with each other.

[0013] In one embodiment, such as Figure 1 As shown, a method for fault diagnosis of heterogeneous power systems based on multilateral interaction is provided. Taking the application of this method to computer equipment as an example, the method includes the following steps: Step 101: Establish a state-space model of the heterogeneous multi-subsystem of the distributed power system, wherein the state-space model is used to describe the line dynamic characteristics, fault dynamic characteristics and coupling interconnection characteristics of each subsystem.

[0014] State-space models, in particular, are mathematical models that describe the dynamic behavior of distributed power systems using state variables, input variables, and output variables. Specifically, they can include system matrices, control matrices, fault distribution matrices, and output matrices, and can be used to describe the line dynamic characteristics, fault dynamic characteristics, and coupling interconnection characteristics of each subsystem. Line dynamic characteristics reflect the dynamic changes in current and voltage of components such as inductors and capacitors in the system; fault dynamic characteristics model actuator faults as unknown inputs to the system; and coupling interconnection characteristics describe the power interaction relationships between subsystems through points of common coupling.

[0015] Specifically, the differential equations of each subsystem are first established using the laws of circuit physics and the equivalent physical model of the distributed power system, i.e., the state-space equations of each subsystem are established. Then, the state-space equations of each subsystem are integrated into a heterogeneous multi-subsystem state-space model. In this embodiment, the dynamic behavior of the power system under normal operation and partial fault conditions can be described based on the source system, energy storage system, load system, and the grid structure coupled through the point of common coupling (PCC). The source system can be a wind turbine generator, considering the aerodynamic conversion of wind energy, the transmission chain, the motor, and the control system, with a focus on describing the impact of pitch actuator faults on system operation. The energy storage system can be an inverter-based electrochemical energy storage device, interconnected with the grid through LC filtering and line inductance. The model includes dynamic variables such as inductor current, capacitor voltage, and line current, and allows consideration of actuator faults of equivalent current injection type. The load system is modeled using dynamic RLC branches, which can describe the characteristics of power electronic loads, inductive loads, or combined power loads, and also allows modeling of faults of equivalent power disturbance type. The network interconnection section facilitates energy exchange between sources, storage, and loads through the point of common coupling (PCC). Voltage and current are bidirectionally coupled between branches through line inductance and capacitance, demonstrating typical electrical interconnection characteristics of a new power system.

[0016] Step 102: For each subsystem, design an observer for distributed state and fault estimation based on an unknown input method. The observer constructs a dynamic equation that simultaneously estimates the system state and fault signal by introducing the derivative information of the output signal and fusing the state estimates of adjacent subsystems.

[0017] Among them, the unknown input method refers to the method of reconstructing the model by treating system faults as unknown input signals; the derivative information of the output signal contains the instantaneous dynamic characteristics of the system; and the fusion of state estimates of adjacent subsystems can reflect the technical path of multilateral knowledge interaction.

[0018] Specifically, dynamic equations for an observer capable of simultaneously estimating system state and fault signals are constructed for each subsystem. These dynamic equations can simultaneously estimate system state and fault signals by incorporating derivative terms of the output signal and neighborhood state estimates. The gain matrix of the observer's dynamic equations is a parameter that needs to be solved in subsequent steps. This observer design method does not require any prior assumptions about fault dynamics, thus avoiding the dimensionality expansion problem caused by constructing augmented systems in traditional methods.

[0019] In this embodiment, a distributed observer can be designed based on a source, storage, and load power grid system. The core idea of ​​this observer is to utilize the correlation information between subsystems, introducing output derivative information and neighborhood state estimation to construct an estimation framework capable of simultaneously reconstructing system states and faults. For example, for a source subsystem (such as a wind turbine system), an observer can be designed to estimate pitch actuator faults. The observer can accurately track the state changes of the wind turbine system and estimate the magnitude and trend of pitch actuator faults in real time by using the wind turbine's output measurements, output derivatives, and state estimation information from the energy storage and load subsystems, through a reasonable gain matrix design. For energy storage systems (such as inverters and LC filters), the observer can be designed to focus on estimating equivalent current injection faults, utilizing the current and voltage measurements and their derivatives of the energy storage system, combined with state estimation information from the source subsystems, thereby effectively decoupling system dynamics from fault effects and achieving accurate estimation of energy storage system faults. For load subsystems (such as dynamic RLC loads), the observer can be designed for equivalent power disturbance faults and accurately reconstruct the fault situation on the load side by using voltage and current measurement information of the load node and coupling state estimation from the source subsystem.

[0020] Step 103: Define the local state estimation error vector and fault estimation error vector for each subsystem, and derive the global error system model based on the observer's dynamic equations and state-space model.

[0021] Among them, the local state estimation error vector is a vector constructed based on the difference between the actual state and the estimated state of the system; the fault estimation error vector is a vector constructed based on the difference between the actual fault and the estimated fault; and the global error system model is a unified error dynamic system formed by integrating the error vectors of all subsystems.

[0022] Specifically, based on the distributed observer design, estimation performance can be further analyzed. First, the local state estimation error of each subsystem can be defined, i.e., the difference between the actual state and the estimated state. For the source system, the state estimation error reflects the accuracy of the wind turbine system's state estimation; for the energy storage system, the error characterizes the precision of the energy storage device's state estimation; and for the load system, the error reflects the reliability of the load dynamic estimation. By combining the observer dynamic equations with the actual system model, the error dynamic equations for the state error can be derived. These equations clearly demonstrate how the estimation error changes over time and how the coupling relationships between subsystems affect error propagation. It is worth noting that through reasonable observer parameter design, fault terms can be decoupled from the error dynamic equations, thus laying the foundation for subsequent stability analysis. The analysis of the error dynamic equations also reveals the paths and mechanisms of fault propagation between subsystems, providing a theoretical basis for understanding the propagation characteristics of faults in power grid systems.

[0023] Furthermore, after obtaining the local error dynamics of each subsystem, the estimation performance of the entire interconnected system can be analyzed from a global perspective. Here, a unified global error system model can be constructed by selecting appropriate observer gain matrices and fault estimation parameters. The construction of the global error system model can consider all coupling relationships between the source, storage, and load subsystems. Specifically, the source system can influence the energy storage and load subsystems through the point of common coupling (PCC); the energy storage system can couple with the system through line inductance current; and the load system can interact with other subsystems through voltage dynamics at the PCC point. In this step, fault estimation error variables can also be defined to quantify the difference between actual faults and estimated faults. By analyzing the characteristics of the global error system, the overall performance of the distributed observers in the entire power grid system can be evaluated.

[0024] Step 104: Based on Lyapunov stability theory, perform stability analysis on the global error system model to obtain the matrix inequality conditions that make the global error system model asymptotically stable.

[0025] Lyapunov stability theory is a mathematical method for analyzing system stability by constructing energy functions; asymptotic stability requires that the system state error eventually converges to zero; matrix inequality conditions, also known as stability conditions, refer to sufficient conditions for system stability expressed in matrix form.

[0026] In this embodiment, to ensure the reliable operation of the designed distributed observer, a rigorous stability analysis of the global error system model can be performed using Lyapunov stability theory. Here, by constructing a suitable Lyapunov function, the convergence of the estimation error and the stability of the system can be analyzed. The selection of the Lyapunov function considers the weighted combination of errors from each subsystem to ensure that the estimation error of the entire system asymptotically converges to zero. During the analysis, it can be assumed that fault changes are relatively slow, an assumption that is reasonable in practical engineering because the rate of change of most system faults is much slower than the system's dynamic response. Through stability analysis, not only is the convergence of state estimation guaranteed, but also the accuracy of fault estimation, thus providing theoretical support for the practical application of the observer.

[0027] Step 105: Transform the matrix inequality conditions into a system of linear matrix inequality equations, and obtain the matrix parameters of the observer by solving the system of linear matrix inequality equations.

[0028] Here, a linear matrix inequality refers to a set of inequality constraints that are linear with respect to matrix variables; a system of inequalities refers to a constraint system composed of multiple linear matrix inequalities; and the matrix parameters of the observer refer to the parameters of all the gain matrices of the observer.

[0029] In this embodiment, the stability conditions derived from Lyapunov stability theory can be transformed into linear matrix inequality constraints, i.e., into a system of linear matrix inequality equations. This transformation converts the complex stability analysis problem into a numerically solvable optimization problem. Then, using MATLAB's LMI toolbox, the gain matrices of the observers for each subsystem are obtained. These gain matrices ensure that the observers possess good dynamic response characteristics while meeting stability requirements. Specifically, regional pole placement constraints can be introduced to restrict the observer poles to specific regions within the complex plane, thereby ensuring good transient performance in the estimation process. The obtained observer parameters can be directly applied to practical power grid monitoring systems to achieve real-time and accurate estimation of the state and faults of each subsystem (source, storage, and load). This complete design process can provide effective technical support for fault diagnosis and health management of smart grids.

[0030] Step 106: Configure the observer through the matrix parameters of the observer, and complete the fault diagnosis of the distributed power system by estimating the system state and fault signals in real time by the observer during operation.

[0031] Among them, configuring the observer refers to embedding the obtained gain matrix into the observer algorithm; real-time output refers to the observer updating the state and fault estimate in each sampling period; fault diagnosis is to determine whether the system has a fault and the type of fault based on the amplitude, trend and other characteristics of the fault estimate signal.

[0032] Specifically, the obtained observer gain matrix and other parameters can be embedded into the observers of each subsystem. During system operation, the measurement outputs and control inputs of each node are continuously collected, and state estimates and fault estimates are generated in real time by solving the observer's dynamic equations. When the magnitude of the fault estimate exceeds a preset threshold or exhibits a specific change pattern, a corresponding fault can be determined in the system, triggering an alarm or protection action. This achieves real-time monitoring of the power system state and fault diagnosis, ensuring stable system operation. It should be noted that the observer design method proposed in this embodiment is applicable to complex power system network environments and provides stable and accurate estimates for system state and fault estimation.

[0033] By applying the technical solution of this embodiment, a state-space model incorporating line dynamics, fault dynamics, and coupled interconnection characteristics can be constructed, realistically reflecting the actual dynamic characteristics of the system and providing high-fidelity input for diagnostic calculations. Simultaneously, by introducing a distributed state and fault estimation observer based on an unknown input method, and fusing the output derivative and neighborhood state information through the observer, various complex time-varying faults, including slowly varying and oscillating faults, can be effectively estimated without making any prior assumptions about fault dynamics, thereby improving the diagnostic capability for time-varying faults. Furthermore, the above method, by designing the observer based on the original state-space model, avoids the computational complexity problems associated with traditional augmented system methods, significantly reducing the difficulty of algorithm implementation. Moreover, by transforming the stability conditions of the error system into a solvable system of linear matrix inequalities, a rigorous mathematical guarantee can be provided for the asymptotic convergence of state and fault estimation, thereby eliminating the uncertainty of relying on empirical parameter tuning. Based on this, the above method can effectively enhance the adaptability of multi-node collaborative estimation under complex fault network conditions by constructing a state-space model that includes line dynamics and combining it with an observer structure suitable for heterogeneous environments. This can improve the overall accuracy, robustness and engineering practicality of fault diagnosis.

[0034] In one embodiment, step 101 can be implemented by: establishing state-space equations for each subsystem to describe the dynamics of the filter inductor current and the voltage at the common connection point, the dynamics of the filter inductor current and the line current, and the coupling relationship between each subsystem; modeling system faults as unknown input terms of the state-space equations; and then coupling the state-space equations of each subsystem to establish a state-space model that includes line dynamic characteristics, fault dynamic characteristics, and coupling interconnection characteristics.

[0035] This embodiment describes the dynamic behavior of a distributed power system under normal operation and partial fault conditions, based on the power grid structure of power sources, storage, loads, and their mutual coupling through the point of common coupling (PCC). It should be noted that the state-space equations of the following systems are merely examples and are not intended to limit the implementation of this embodiment. Due to space limitations, the construction of the state-space equations for each system is only briefly introduced below.

[0036] The source system (S) is exemplified by a wind turbine system, and its system model structure is as follows: Figure 2 As shown, the system consists of the following components: an impeller system for converting wind energy into mechanical energy; a pitch servo system for adjusting the pitch angle to control the capture of wind energy; a transmission system for connecting the low-speed shaft and the high-speed shaft via a gearbox to transmit mechanical energy; and a generator and converter section for converting mechanical energy into electrical energy.

[0037] The modeling process for each circuit module is omitted here. By integrating the physical models of each circuit module, the state-space equations of the wind turbine system can be obtained as follows: State vector: (1) Input vector: (2) The state equation can be expressed as follows: (3) in, Represents the impeller speed. Represents the generator speed. Represents the torsion angle of the drive train. Represents the pitch angle. Represents the electromagnetic torque of the generator. For inductor current, The voltage across the capacitor is . For pitch reference, For control input, i.e. torque reference, This is the equivalent voltage output by the source-side inverter. and These represent the system's state vector, input vector, and measurable output vector, respectively. It's a pitch actuator malfunction, matrix switcher. It is a known constant matrix of the corresponding dimension, where This represents the coupling matrix of other nodes to the source system. The specific expression of this model matrix is ​​as follows:

[0038]

[0039]

[0040]

[0041] in, For the rotational inertia of the low-speed shaft ( ), For the rotational inertia of the high-speed shaft ( ), The damping coefficient is... For torsional stiffness ( ), The gear torque damping coefficient ( ), This is the gear ratio. For the damping ratio, For natural frequency ( ), The time constant of the torque control loop ( ), These are the filter inductor and the parallel capacitor, respectively. This represents the equivalent resistance of capacitor loss.

[0042] Furthermore, the circuit topology of the energy storage system (E) is as follows: Figure 3 As shown, the system includes a grid-connected inverter, an LC filter circuit, and a line inductor. Its physical quantities are defined as follows: This indicates the output voltage of the storage-side inverter. Indicates the storage-side filter inductor. This represents the filter inductor current. Indicates the line inductance current. This indicates the parallel capacitor on the storage side. Indicates the node voltage of the storage-side capacitor. This represents the equivalent resistance of capacitor loss. This indicates the line inductance.

[0043] The modeling process for each circuit module is omitted here. By integrating the physical models of each circuit module, the state-space equations for energy storage can be obtained as follows: Selection of state variables: (4) Input vector and fault vector: (5) Its state-space equations are expressed as follows: (6) in, , , This represents the state vector of the three energy storage modules in the energy storage system. , , Let these represent the state vector, input vector, and measurable output vector of the energy storage system, respectively. The equivalent voltage indicating an inverter output fault or abnormality. All of them are known constant matrices.

[0044] The matrix is ​​specifically represented as follows:

[0045]

[0046] Furthermore, the circuit topology of the load system (L) is as follows: Figure 4 As shown, the system includes a parallel RLC branch connected to the PCC point. Its physical quantities are defined as follows: Represents the series inductance on the load side. Represents the load inductor current. Represents the parallel capacitor of the load. Represents the load node voltage. This represents the equivalent resistance of capacitors in parallel. This represents the voltage at the load capacitor node.

[0047] The modeling process for each circuit module is omitted here. By integrating the physical models of each circuit module, the state-space equations for energy storage can be obtained as follows: State variable selection: (7) The local differential equations are summarized as follows: (8) Its state-space equations are expressed as follows: (9) in, , , These represent the state vector, input vector, and measurable output vector of the load system, respectively. This indicates an equivalent current injection fault. These are all known constant matrices, and the specific representations of each matrix are as follows:

[0048]

[0049] Furthermore, a simplified diagram of the coupling relationship among the three is shown below. Figure 5 As shown. Figure 5 As shown, in a distributed power system comprising a source system, an energy storage system, and a load system, these three systems are tightly coupled and interact with each other through the voltage and current at the point of common coupling (PCC). The voltage output by the source system directly constitutes the PCC voltage of the distributed power system; the energy storage subsystem is connected to the PCC through line inductance current, and changes in its current directly affect the dynamics of the PCC voltage; simultaneously, the load system draws current from the PCC, and changes in its current also disturb the PCC voltage. Therefore, the PCC voltage becomes a critical coupling hub. Any change in the state or power of a single subsystem, such as energy storage charging and discharging, load switching, or source-side output adjustment, will be rapidly transmitted to the other two subsystems through this common variable of the PCC voltage, thus triggering a cascading dynamic response. This strong coupling relationship involving electrical quantities makes the dynamic behavior of the entire system highly interactive and holistic, and it also becomes the core physical foundation that must be considered when designing observers capable of real-time sensing of system status and fault location.

[0050] In one embodiment, in step 102, the dynamic equations of the observer can be constructed as follows: In this embodiment, a distributed state and fault estimation observer based on an unknown input method can be designed, and a dynamic equation for fault estimation can be constructed by introducing output derivative information and neighborhood state estimation.

[0051] First, based on the construction process of the state-space equations of each subsystem in the previous embodiment, it can be seen that the state-space equations of similar interconnected subsystems can be described in the following way: (10) in, , , , Representing subsystems The system status, control inputs, system faults, and measurable outputs. , , , , It is a subsystem The known matrices in the state-space equations express System and Coupling relationships between systems It is an adjacent subsystem The system state, here we assume It has full rank. It has full order.

[0052] In this embodiment, As the state variables of the system, they can be divided into electrical sub-blocks and mechanical sub-blocks in the power system. The electrical sub-blocks mainly consist of dynamic electrical quantities related to power electronic interfaces and networks, while the mechanical sub-blocks can consist of things like speed, torsion angle, and pitch angle. As the input to the system, it can be divided into controllable electrical port excitation and controllable mechanical excitation in the power system. Among them, electrical excitation mainly consists of the settings of equivalent inverter port / current and power / voltage, while mechanical excitation is related to the specific model. System faults are classified into electrical faults (device deviations on the electrical side, line faults, external disturbances) and component faults (mechanical component faults) in power systems. For the system's measurable output, it typically consists of measurable voltage / current / power and necessary mechanical sensing quantities in a power system. Furthermore, The dynamics of electrical components (linearized equations for inductor / capacitor circuits, etc.) can also be expressed as the dynamics of mechanical components; Indicates a controllable input channel for an electrical / mechanical port; Indicates the equivalent fault / disturbance injection channel; These represent the measurement gating matrix; all of these are known matrices. express System and The coupling relationships between systems, and assuming It has full rank. It has full row rank. Unlike centralized or decoupled approaches, this structure provides interpretable network effects and independent fault inputs while maintaining sparsity and scalability; compared to static estimation, it can cover the fast dynamics of power system scenarios; and it can seamlessly integrate with subsequent observer designs in the same coordinate system.

[0053] Furthermore, in order to simultaneously reconstruct the fault signals of the interconnected subsystem and system status This embodiment is the first one. The subsystem constructs a novel distributed state and fault estimation observer based on an unknown input method, whose dynamic equations are shown below: (11) in, For subsystem The state of the observer, For subsystem System status The estimated value, For subsystem System failure The estimated value. For subsystem Measurable output, For subsystem The control input, express The system and its neighbors Coupling relationships between systems It is an adjacent subsystem System status The online estimate. Matrix , , , , , , It is a series of observer gain matrices that need to be solved subsequently. , It is a subsystem The known matrices in the state-space equations. Note that here we use... To indicate the first Subsystem The generalized inverse is defined as follows: .

[0054] In this embodiment, by introducing the output derivative into the observer, it is equivalent to expanding the observation information without adding sensors, significantly improving the observability near the linearized operating point and the separability of unknown inputs. Compared with the unknown input method of augmented state, this method can linearly reconstruct the equivalent injection without complex augmentation, the parameter meaning is clear, and the implementation is simple, making it suitable for engineering implementation in power electronics-dominated systems.

[0055] As can be seen from equation (11), this observer does not only utilize coupling, but also the interaction of multilateral knowledge. Firstly, the estimated values ​​of the state variables are not the directly measured raw data, but rather the optimal or suboptimal estimates obtained by the observer after integrating information from multiple sources. The interaction between the subsystem observers is the first The estimates of the observers of each subsystem are based on information exchange with neighboring nodes, and are not simply a matter of coupling.

[0056] In one embodiment, step 103 can be implemented by the following method: defining the local state estimation error vector of each subsystem, and deriving the local state error dynamic equation of each subsystem based on the state-space model and the dynamic equation of the observer; selecting a set of matrix parameters for the gain matrix of the dynamic equation of the observer, and simplifying the local state error dynamic equation based on the selected matrix parameters; defining the fault estimation error vector of each subsystem, and expressing the fault estimation error vector as a linear combination of the local state estimation error vectors based on the dynamic equation of the observer; integrating the simplified local state error dynamic equations and the expressed fault estimation error vectors of each subsystem to construct a global error system model.

[0057] In this embodiment, the local state estimation error vector of each subsystem is first defined, and the local state error dynamic equation of each subsystem is derived based on the state-space model and the dynamic equation of the observer. The goal is to illustrate how the differences between measurements such as bus voltage, port current, and branch current and model predictions in a power grid propagate under the influence of adjacency coupling, input changes, and faults, and how they decay at a set rate, thereby ensuring that the estimation and fault reconstruction results are consistent with the physical state of the power grid.

[0058] Specifically, for the first For each subsystem, its local state estimation error vector is defined as: (12) Then, the subsystem shown by formula (10) The state-space equations and formula (11) of the subsystem By deriving the dynamic equations of the observer, the subsystem can be obtained. The local state error vector is represented as: (13) Here, record Then the local state error vector can be further expressed as: (14) Then, from formula (11), we can obtain: (15) Substituting formula (15) into formula (14) yields the subsystem. The local state error dynamic equation is: (16) For formulas (12) to (16), the parameters are described as follows: For subsystem The local state estimation error vector, For subsystem The system status, For subsystem System status The estimated value, For subsystem Measurable output, For subsystem The state of the observer, For subsystem The system failure For subsystem The control input, It is a subsystem The known matrices in the state-space equations It is a subsystem The gain matrix of the observer, For adjacent subsystems The local state estimation error vector, for 3D identity matrix.

[0059] Unlike the traditional approach where local observers treat the network as noise, this step explicitly incorporates neighborhood propagation into the error equation and compensates for it within the algorithm, resulting in lower false alarms and more accurate localization. Furthermore, compared to traditional methods using centralized large models, this step employs a sparse structure combining local errors and adjacency interfaces, exchanging only necessary information with neighbors, leading to lower scalability and better real-time performance. For observers requiring augmentation of unknown inputs, this step retains an equivalent injection channel in the error domain, separating unknown inputs without augmentation, resulting in wider conditions and easier parameter interpretation.

[0060] Furthermore, a suitable parameter matrix can be selected for the dynamic equations of the observer, and a fault estimation error vector can be defined to further derive the global error system. In this embodiment, the operating environment is designed for multi-node grid connection under fault networking conditions, primarily using power electronic interfaces, characterized by high dynamics and strong mutual interference. The coupling strength of adjacent nodes and potential communication delays / missing measurements are also considered. Based on the above, the estimation error vector is defined as the deviation of the measurable electrical quantities of this node from the model predictions. This is then used to construct a global error system to reflect the network propagation effects, providing an electrically consistent convergence and robustness objective for the subsequent calculation of the observer gain matrix.

[0061] Specifically, in formula (16), if the gain matrix of the observer can be found to satisfy the following condition: (17) (18) (19) Therefore, the dynamic equation for the local state error can be simplified to the following form: (20) On the other hand, for the first For each subsystem, its fault estimation error vector can be defined as: (twenty one) By substituting the third equation of formula (11) into formula (21), that is, substituting the equation of the system fault estimate in the observer's dynamic equation into formula (21), we can obtain: (twenty two) in, Therefore, for the algebraic equation (22), we can define: (twenty three) Then, the fault estimation error vector can be expressed as: (twenty four) For formulas (17) to (24), the parameters are described as follows: For subsystem The local state estimation error vector, For subsystem The system status, For subsystem System status The estimated value, For subsystem Measurable output, For subsystem The fault estimation error vector, For subsystem The system failure For subsystem The estimated value of system failure. For subsystem The control input, It is a subsystem The known matrices in the state-space equations Through The derived matrix, It is a subsystem The gain matrix of the observer, For adjacent subsystems The local state estimation error vector, It is an identity matrix.

[0062] Furthermore, to represent the local state estimation error (20) and fault estimation error (24) from a global perspective, some global variables and matrices are defined as follows: (25) Then systems (20) and (24) can be integrated into a global error system model, which is expressed as follows: (26) (27) in, For global state estimation error, This is the fault estimation error. It's important to note here that in equation (27), if we want to... Stability means making Approaching 0, here we only need to... Approaching 0. Based on the above analysis, it can be concluded that as long as certain conditions can be found to stabilize the global error dynamic equation (26), the system state and fault signal can be asymptotically estimated through the dynamic equation (11) of the distributed observer.

[0063] Compared to previous approaches that relied on experience or centralized parameter tuning, this step first clearly defines the estimation error of the fault / state under a unified coordinate system and derives the global error system, allowing the tuning to revolve around the error propagation mechanism rather than trial and error. Subsequently, several sets of algebraic relationships are used to eliminate the original system terms from the error dynamics, resulting in a simpler structure and more robust analysis. This step also provides the gain satisfaction condition, transforming the problem from "whether it can converge" to "solving according to the conditions," thus providing a solid foundation for subsequent linear transformations.

[0064] In one embodiment, step 104 can be implemented by the following method: First, the global error system model is configured with regional poles so that the eigenvalues ​​of the global error system model fall within a preset stable region. Then, the derivative of the global error system model along the system trajectory is calculated using a quadratic Lyapunov function. Finally, the derivative of the global error system model along the system trajectory is set to be negative in order to derive the matrix inequality conditions that guarantee the asymptotic stability of the global error system model.

[0065] In this embodiment, the stability of the global error system model can be analyzed based on Lyapunov stability theory, and the stability conditions that make the global error system asymptotically stable can be obtained. In this embodiment, under the conditions of multi-node grid connection, coupling and fault coexistence, it can be provably guaranteed that the estimation error and fault reconstruction error converge as required, and the deviations of electrical quantities such as bus voltage / port and branch current decay within a limited time and remain within the operating boundary. This ensures that the observer does not generate false alarms due to normal command or interconnection propagation in a grid environment with a high proportion of power electronics, has robustness and feasibility, and can maintain the above performance as the scale expands.

[0066] First, regional pole configuration can be performed, defining... If equation (28) can be satisfied, then The eigenvalues ​​will fall in the region The formula is as follows: (28) Next, we will perform a stability analysis, first selecting the Lyapunov function as follows: (29) From equation (20), we can obtain The derivative is as follows: (30) This makes The following equation holds for values ​​less than 0, where equation (31) is the matrix inequality condition that makes the global error system asymptotically stable.

[0067] (31) in

[0068] If equation (31) holds, then the state error dynamic (26) is asymptotically stable. Therefore, the system state and faults can be asymptotically reconstructed by the dynamic equation (11) of the distributed observer.

[0069] In one embodiment, step 105 can be implemented by the following method: First, define intermediate variables and perform variable substitution on the nonlinear product terms in the dynamic equations of the observer to transform the matrix inequality conditions into a system of inequality equations of linear matrices about the intermediate variables. Then, solve the system of inequality equations of linear matrices to obtain the values ​​of the intermediate variables. Finally, based on the values ​​of the intermediate variables, solve in reverse for the matrix parameters of all gain matrices of the observer.

[0070] In this embodiment, the stability condition can be transformed into a linear matrix inequality constraint, thereby solving for the matrix parameters of the gain matrix in the observer's dynamic equations. In this embodiment, the "theoretical stability condition" based on Lyapunov can be reduced to a set of computable engineering constraints. This step involves incorporating requirements such as "response time, steady-state deviation, input decoupling, neighborhood compensation, and fault reconfiguration" into the same set of linear matrix inequality equations. The sparse structure of the power grid blocks supports distributed or parallel solving by node / adjacency, thus obtaining directly deployable gain matrix parameters instead of relying on trial and error for parameter tuning.

[0071] First, define And assume: (32) Then we can obtain from the definition. Using equation (32), It can be expressed as follows: (33) Returning to equation (17), substituting equations (32) and (33) yields: (34) because We can get Equation (32) is equivalent to (17).

[0072] It is important to note that before this transformation, the observer gain was very difficult to solve because the gain matrices were all coupled together. The above transformation proposes a decoupling transformation, which makes the subsequent observer gain easier to solve.

[0073] Furthermore, let's combine equation (18) with... Combining these, we can obtain: (35) Here, we can use the Moore-Penrose pseudoinverse to find the answer. The general solution is expressed as follows: (36) in, This is a relaxation matrix used to increase design freedom, and we define it here: (37) From the above formula, we can obtain Based on this, it can be It is expressed as follows: (38) According to (32) and (38), we can The following is an expression: (39) Returning to the stability analysis, the nonlinear terms in the resulting matrix inequality conditions are: (40) (41) To represent the observer synthesis conditions within the framework of LMI (Linear Matrix Inequality), we define here... Equations (40) and (41) can be written as: (42) (43) If the expression obtained by substituting (42) and (43) into (28) holds true, then it can be explained that... The eigenvalues ​​will fall in the region Inside.

[0074] Next, substituting (42) and (43) into equation (31) yields: (44) in

[0075]

[0076]

[0077]

[0078] If equation (44) holds, it can be shown that the state and fault can be progressively reconstructed through the dynamic equation (11) of the distributed observer. Furthermore, the matrix parameters of the observer's gain matrix can be solved as follows: (45) With the help of equation (45), It can be solved by equation (36), and then It can also be by Please solve. All of these can be solved by equations (19), (32) and (33) respectively, thereby obtaining the matrix parameters of all gain matrices in the dynamic equation of the observer.

[0079] In one embodiment, step 106 can be implemented by the following method: First, based on the matrix parameters of the observer, the observer of each subsystem is configured with parameters. Then, during the operation of the distributed power system, the observer of each subsystem performs real-time calculations based on the locally measured output signal, the derivative of the output signal, the control input, and the state estimates from adjacent subsystems, and synchronously outputs the state estimates and fault estimates of the subsystem. Finally, the fault estimates are used as the quantitative diagnostic results of the actuator faults of the subsystem to achieve fault diagnosis of the distributed power system.

[0080] For example, in a photovoltaic microgrid system, the photovoltaic inverter subsystem can be configured and monitored. The observer for this subsystem is assigned a corresponding system matrix, control matrix, and observer gain. During operation, it continuously measures the inverter's output current, calculates the rate of change of current, and receives pulse width modulation commands from the controller. Simultaneously, it obtains state estimation information from the observer of the connected energy storage converter subsystem. Using this information, the observer calculates estimated values ​​of the inverter's internal inductor current and capacitor voltage in real time and generates a fault estimation signal reflecting the actuator's health status. If this signal value continuously exceeds a set threshold, it indicates that the inverter's switching transistor drive circuit may have a specific degree of fault, thereby achieving quantitative location and assessment of the fault in this power unit's actuator.

[0081] This embodiment utilizes an observer to collect information from local measurements and information from adjacent subsystems, and performs interactive distributed collaborative computing to achieve real-time, quantitative, and parallel diagnosis of actuator faults in each subsystem, thereby effectively improving the timeliness and accuracy of system fault diagnosis.

[0082] Furthermore, as Figure 1 In terms of specific implementation, this application provides a fault diagnosis device for heterogeneous power systems based on multilateral interaction, such as... Figure 6 As shown, the device includes: The spatial model establishment module 21 can be used to establish a state space model of a heterogeneous multi-subsystem of a distributed power system, wherein the state space model is used to describe the line dynamic characteristics, fault dynamic characteristics and coupling interconnection characteristics of each subsystem. The observer design module 22 can be used to design an observer for distributed state and fault estimation based on an unknown input method for each of the subsystems, wherein the observer constructs a dynamic equation that simultaneously estimates the system state and fault signal by introducing the derivative information of the output signal and fusing the state estimates of adjacent subsystems. The error model building module 23 can be used to define the local state estimation error vector and fault estimation error vector of each subsystem, and derive the global error system model based on the dynamic equation of the observer and the state space model. Stability analysis module 24 can be used to perform stability analysis on the global error system model based on Lyapunov stability theory, and obtain the matrix inequality conditions that make the global error system model asymptotically stable. The matrix parameter solving module 25 can be used to transform the matrix inequality conditions into a system of linear matrix inequality equations, and obtain the matrix parameters of the observer by solving the system of linear matrix inequality equations. The system fault diagnosis module 26 can be used to configure the observer through the matrix parameters of the observer, and to complete the fault diagnosis of the distributed power system through the estimation of system status and fault signals output by the observer in real time during operation.

[0083] In specific application scenarios, the space model establishment module 21 can be used to establish the state space equations of each subsystem to describe the dynamics of the filter inductor current and the voltage at the common connection point, the dynamics of the filter inductor current and the line current, and the coupling relationship between each subsystem. System faults are modeled as unknown input terms of the state space equations. The state space equations of each subsystem are coupled to establish a state space model that includes line dynamic characteristics, fault dynamic characteristics, and coupling interconnection characteristics.

[0084] In specific application scenarios, for any subsystem, the dynamic equations of the observer are as follows:

[0085] in, For subsystem The state of the observer, For subsystem System status The estimated value, For subsystem System failure The estimated value, For subsystem Measurable output, For subsystem The control input, express The system and its neighbors Coupling relationships between systems It is an adjacent subsystem System status The estimated value, For subsystem The known matrices in the state-space equations For the first Subsystem The generalized inverse is defined as , Let be the gain matrix of the observer to be solved.

[0086] In specific application scenarios, the error model establishment module 23 can be used to define the local state estimation error vector of each subsystem, and derive the local state error dynamic equation of each subsystem based on the state space model and the dynamic equation of the observer; select a set of matrix parameters for the gain matrix of the dynamic equation of the observer, and simplify the local state error dynamic equation based on the selected matrix parameters; define the fault estimation error vector of each subsystem, and express the fault estimation error vector as a linear combination of the local state estimation error vectors based on the dynamic equation of the observer; integrate the simplified local state error dynamic equations and the expressed fault estimation error vectors of each subsystem to construct the global error system model.

[0087] In specific application scenarios, the stability analysis module 24 can be used to configure regional poles of the global error system model so that the eigenvalues ​​of the global error system model fall within a preset stable region; calculate the derivative of the global error system model along the system trajectory using a quadratic Lyapunov function; set the derivative of the global error system model along the system trajectory to be negative to derive the matrix inequality conditions that guarantee the asymptotic stability of the global error system model.

[0088] In specific application scenarios, the matrix parameter solving module 25 can be used to define intermediate variables, perform variable substitution on the nonlinear product terms in the dynamic equations of the observer, so as to transform the matrix inequality conditions into a system of inequality equations about the linear matrix of the intermediate variables; solve the system of inequality equations of the linear matrix to obtain the value of the intermediate variables; and, based on the value of the intermediate variables, solve in reverse for the matrix parameters of all gain matrices of the observer.

[0089] In specific application scenarios, the system fault diagnosis module 26 can be used to configure the parameters of the observers of each subsystem based on the matrix parameters of the observers. During the operation of the distributed power system, the observers of each subsystem perform real-time calculations based on the locally measured output signal, the derivative of the output signal, the control input, and the state estimates from adjacent subsystems, and synchronously output the state estimates and fault estimates of the subsystem. The fault estimates are used as quantitative diagnostic results for the actuator faults of the subsystem to achieve fault diagnosis of the distributed power system.

[0090] It should be noted that other corresponding descriptions of the functional units involved in the heterogeneous power system fault diagnosis device based on multilateral interaction provided in this application embodiment can be found in the following references. Figure 1 The corresponding descriptions in the methods shown will not be repeated here.

[0091] This application also provides a computer device, specifically a personal computer, server, network device, etc. The computer device includes a bus, processor, memory, and communication interface, and may also include input / output interfaces and a display device. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database of the computer device stores location information. The network interface of the computer device is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements the steps in the various method embodiments.

[0092] Those skilled in the art will understand that the structure of the computer device described above is only a partial structure related to the solution of this application, and does not constitute a limitation on the computer device to which the solution of this application is applied. A specific computer device may include more or fewer components, or combine certain components, or have different component arrangements.

[0093] In one embodiment, a computer-readable storage medium is provided, which may be non-volatile or volatile, having stored thereon a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0094] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0095] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, graphics processors, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0096] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0097] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A fault diagnosis method for heterogeneous power systems based on multilateral interaction, characterized in that, The method includes: Establish a state-space model of a heterogeneous multi-subsystem distributed power system, wherein the state-space model is used to describe the line dynamic characteristics, fault dynamic characteristics and coupling interconnection characteristics of each subsystem; For each of the subsystems, an observer for distributed state and fault estimation based on an unknown input method is designed. The observer constructs a dynamic equation that simultaneously estimates the system state and fault signal by introducing the derivative information of the output signal and fusing the state estimates of adjacent subsystems. Define the local state estimation error vector and fault estimation error vector for each subsystem, and derive the global error system model based on the dynamic equation of the observer and the state-space model; Stability analysis of the global error system model is performed based on Lyapunov stability theory, and matrix inequality conditions that make the global error system model asymptotically stable are obtained. The matrix inequality conditions are transformed into a system of linear matrix inequality equations, and the matrix parameters of the observer are obtained by solving the system of linear matrix inequality equations. By configuring the observer using its matrix parameters and by estimating the system state and fault signals in real time during operation, fault diagnosis of the distributed power system can be completed.

2. The method according to claim 1, characterized in that, The distributed power system comprises multiple heterogeneously interconnected subsystems with different physical structures and dynamic characteristics; therefore, establishing a state-space model of the heterogeneous multi-subsystem of the distributed power system includes: State-space equations for each subsystem are established to describe the dynamics of the filter inductor current and the voltage at the common connection point, the dynamics of the filter inductor current and the line current, and the coupling relationships between each subsystem. System faults are modeled as unknown input terms of the state-space equations. The state-space equations of each subsystem are coupled to establish a state-space model that includes line dynamic characteristics, fault dynamic characteristics, and coupled interconnection characteristics.

3. The method according to claim 1, characterized in that, For any subsystem, the dynamic equations of the observer are as follows: in, For subsystem The state of the observer, For subsystem System status The estimated value, For subsystem System failure The estimated value, For subsystem Measurable output, For subsystem The control input, express The system and its neighbors Coupling relationships between systems It is an adjacent subsystem System status The estimated value, For subsystem The known matrices in the state-space equations For the first Subsystem The generalized inverse is defined as , Let be the gain matrix of the observer to be solved.

4. The method according to claim 1, characterized in that, The local state estimation error vector and fault estimation error vector of each subsystem are defined, and a global error system model is derived based on the dynamic equations of the observer and the state-space model, including: Define the local state estimation error vector for each subsystem, and derive the local state error dynamic equation for each subsystem based on the state space model and the dynamic equation of the observer; A set of matrix parameters is selected for the gain matrix of the dynamic equation of the observer, and the local state error dynamic equation is simplified based on the selected matrix parameters; Define a fault estimation error vector for each of the subsystems, and express the fault estimation error vector as a linear combination of the local state estimation error vectors based on the dynamic equations of the observer; The simplified local state error dynamic equations and the expressed fault estimation error vectors of each subsystem are integrated to construct the global error system model.

5. The method according to claim 1, characterized in that, The stability analysis of the global error system model based on Lyapunov stability theory yields matrix inequality conditions that make the global error system model asymptotically stable, including: The global error system model is configured with regional poles so that the eigenvalues ​​of the global error system model fall within a preset stable region. The derivative of the global error system model along the system trajectory is calculated using a quadratic Lyapunov function; The derivative of the global error system model along the system trajectory is set to be negative, so as to derive the matrix inequality conditions that guarantee the asymptotic stability of the global error system model.

6. The method according to claim 1, characterized in that, The process of transforming the matrix inequality conditions into a system of linear matrix inequalities, and obtaining the observer's matrix parameters by solving the system of linear matrix inequalities, includes: Define intermediate variables and substitute them for the nonlinear product terms in the dynamic equations of the observer to transform the matrix inequality conditions into a system of inequality equations with respect to the linear matrix of the intermediate variables. Solve the system of inequality equations for the linear matrix to obtain the values ​​of the intermediate variables; Based on the values ​​of the intermediate variables, the matrix parameters of all gain matrices of the observer are solved in reverse.

7. The method according to claim 1, characterized in that, The process of configuring the observer through its matrix parameters and performing fault diagnosis on the distributed power system by estimating the system state and fault signals in real time during operation, includes: Based on the matrix parameters of the observer, the observer parameters of each subsystem are configured. During the operation of a distributed power system, the observer of each subsystem performs real-time calculations based on the locally measured output signal, the derivative of the output signal, the control input, and the state estimates from adjacent subsystems, and synchronously outputs the state estimates and fault estimates of the subsystem. The fault estimate is used as a quantitative diagnostic result for the actuator fault of the subsystem, so as to realize the fault diagnosis of the distributed power system.

8. A fault diagnosis device for heterogeneous power systems based on multilateral interaction, characterized in that, The device includes: The spatial model building module is used to build a state space model of a heterogeneous multi-subsystem distributed power system, wherein the state space model is used to describe the line dynamic characteristics, fault dynamic characteristics and coupling interconnection characteristics of each subsystem. The observer design module is used to design an observer for distributed state and fault estimation based on an unknown input method for each of the subsystems. The observer constructs a dynamic equation that simultaneously estimates the system state and fault signal by introducing the derivative information of the output signal and fusing the state estimates of adjacent subsystems. The error model building module is used to define the local state estimation error vector and fault estimation error vector of each subsystem, and derive the global error system model based on the dynamic equation of the observer and the state space model. The stability analysis module is used to perform stability analysis on the global error system model based on Lyapunov stability theory, and obtain the matrix inequality conditions that make the global error system model asymptotically stable. The matrix parameter solving module is used to transform the matrix inequality conditions into a system of linear matrix inequality equations, and obtain the matrix parameters of the observer by solving the system of linear matrix inequality equations. The system fault diagnosis module is used to configure the observer through the matrix parameters of the observer, and to complete the fault diagnosis of the distributed power system through the estimation of system status and fault signals output by the observer in real time during operation.

9. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 7.

10. A computer device, comprising a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 7.

Images