Fault detection method for photovoltaic micro-grid based on outlier memory type event triggering resistance

Abstract

The invention discloses a photovoltaic microgrid fault detection method based on anti-outlier memory event triggering. The method comprises the following steps: step 1, establishing a state space model of a photovoltaic micro-grid single-phase inverter with external disturbance and fault signals; 2, designing an anti-outlier memory type event trigger communication mechanism based on a historical data mean value for outliers caused by noise pollution and environmental sudden change existing in measurement signals, reducing excessive triggering caused by measurement noise, and avoiding false triggering and false alarm caused by the outliers; and step 3, establishing a photovoltaic micro-grid closed-loop fault detection system model based on a residual filter, solving parameters of the residual filter and an event triggering mechanism by means of a Lyapunov method, a Simpson law and a linear matrix inequality technology, and carrying out fault detection on the photovoltaic micro-grid by adopting the designed anti-outlier memory type event triggering mechanism. Compared with an existing event triggering mechanism, the anti-outlier memory type event triggering mechanism is provided, the triggering and updating frequency of detection signals is effectively reduced, the fault detection precision is improved, and certain engineering application value is achieved.

Description

Technical Field

[0001] This invention relates to a fault detection method for photovoltaic microgrids based on anti-outlier memory-type event triggering, and more particularly to an event-triggered fault detection method for networked photovoltaic microgrid systems where outliers exist in the measurement signal. Background Technology

[0002] With the continuous advancement of industrialization, human society's demand for energy is increasing daily. Due to the dwindling reserves of non-renewable resources such as oil and natural gas, and the international consensus on carbon neutrality, the development and utilization of renewable and clean energy sources such as wind and solar power can alleviate the energy crisis. However, because photovoltaic panels and other equipment are exposed to the outdoors for extended periods, they are susceptible to aging and damage to components, leading to system instability and potential power outages. Therefore, research on fault detection in photovoltaic microgrid systems based on communication networks is of significant value. Simultaneously, event-triggered communication mechanisms with "on-demand transmission" characteristics are increasingly being applied to the control and estimation design of networked microgrid systems. These mechanisms can effectively reduce the transmission and computation of redundant signals and improve the resource utilization rate of communication networks.

[0003] However, when measurement data is affected by random fluctuations, existing event-triggered mechanisms based on instantaneous data are prone to over-triggering of measurement signals, wasting limited communication resources. Furthermore, the presence of outliers in the measurement data can lead to false triggering of measurement signals, resulting in erroneous fault alarms. To address these issues, it is essential to redesign a new event-triggered transmission mechanism that effectively avoids over-triggering and false triggering of measurement signals, thereby improving the accuracy of fault detection, while ensuring adequate system performance. Summary of the Invention

[0004] Purpose of the invention: Based on the above analysis, in order to improve the utilization rate of network resources and reduce the false alarm rate of fault detection, this invention constructs a fault detection method for photovoltaic microgrids based on anti-outlier memory event triggering.

[0005] The specific technical solution of the present invention includes the following steps:

[0006] Establish a linear state-space model for the photovoltaic microgrid system;

[0007] Design a memory-based event triggering mechanism to resist outliers;

[0008] Solve for the parameters of the event trigger and filter that make the residual filtering system stable, and perform fault detection on the photovoltaic microgrid system.

[0009] The aforementioned fault detection method for photovoltaic microgrids based on anti-outlier memory events is characterized by the following state-space model: in y(t) represents the state vector, control input, and measurement output, respectively; i f (t), R f L f Represents the current, resistance, and inductance of an inductor-capacitor filter; R o L o Indicates the resistance and inductance of the transmission line; i o (t), v o (t) represents the output current and voltage; v c (t) represents the capacitor voltage; D represents the duty cycle of the Boost circuit; V sun This indicates the storage voltage of the photovoltaic microgrid; This is the gain of the state controller.

[0010] Selecting the gain of the state feedback controller Make It is the Herwitz matrix.

[0011] Considering external disturbances w(t) and faults g(t) caused by component aging, the following stable closed-loop photovoltaic microgrid state-space model can be obtained: in E and G are system parameter matrices.

[0012] The aforementioned fault detection method for photovoltaic microgrids based on anti-outlier memory-type event triggering is characterized by the following designed anti-outlier memory-type event triggering mechanism: In the formula: t k h, t k+1 h represents the current and next trigger times, h represents the sampling period, and θ represents the interval length of historical data. These represent the historical data mean at the current sampling time and the historical data mean at the trigger time, respectively, with the symbol... express Let σ1 and σ2 be the event triggering error terms, representing the triggering threshold parameters. When the event triggering condition (3) is met, the mean signal at the current time is triggered, and the input of the fault detection filter is updated to... Otherwise, the mean signal at the current moment is not triggered, and the input to the fault detection filter retains the value from the previous trigger moment.

[0013] When the interval length of historical data approaches 0, i.e., θ→0, and without considering the upper bound constraint of the event triggering error term, the designed triggering mechanism will be simplified to the following conventional event triggering mechanism based on instantaneous measurement signals: in

[0014] The aforementioned fault detection method for photovoltaic microgrids based on anti-outlier memory-type event triggering is characterized by deriving a closed-loop residual filter system model, solving for the parameters of the residual filter and the event triggering mechanism, and then designing a fault detection algorithm to detect and warn of faults in the photovoltaic microgrid. Specifically, the state-space model of the residual filter is designed as follows: in τ k For network transmission latency, A d B d C d D d These are the parameters for the filter that needs to be designed.

[0015] A reference model for the fault signal is introduced to generate a desired residual evaluation result. The fault signal g(t) is then used to... The transfer function is Its state-space model can be represented as:

[0016] definition It can be represented as:

[0017] Simpson's rule is used to approximate the integral term in formula (7). This can be further expressed as: in

[0018] definition The following fault detection closed-loop system can be obtained: in

[0019] To solve for the parameters of the filter and the event trigger, the following Lyapunov complex is selected: in:

[0020] Define the following new vector:

[0021] Based on the above definition, the closed-loop system (9) can be expressed as:

[0022] The derivative of the Lyapunov complex is:

[0023] Using Jensen's inequality and the technique of mutual convexity, the integral term in formula (13) can be scaled as follows:

[0024] To make the system satisfy H ∞ For stable performance, the derivative of the Lyapunov function must satisfy:

[0025] Based on the event triggering condition (3), we can obtain: in It is a positive number.

[0026] Based on the above conditions, the following matrix inequality form guarantees that (15) holds:

[0027] By utilizing And construct a matrix Formula (17) is equivalent to: in:

[0028] By selecting the matrix We can obtain:

[0029] definition Substituting equation (19) into equation (18), we get: in:

[0030] The filter parameters can be obtained as follows:

[0031] To design a fault detection algorithm, a residual evaluation function is introduced.

[0032] In order to detect fault signals, a fault alarm threshold needs to be predetermined:

[0033] Based on the residual evaluation function and fault alarm threshold mentioned above, the following fault alarm conditions are given:

[0034] Substitute the event trigger matrix Ω and the filter parameters obtained in equation (21) into the closed-loop fault detection system to monitor the fault alarm conditions designed above. When the residual evaluation function Greater than the alarm threshold An alarm signal is generated to indicate that a fault has occurred; otherwise, it indicates that the system is in normal operating condition.

[0035] The above method for obtaining linear matrix inequality conditions is based on Lyapunov stability theory. It uses Lyapunov method, Simpson's rule and matrix transformation techniques to give a convex optimization method for the joint design of residual filter and event triggering mechanism.

[0036] The beneficial effects of this invention are as follows: The method of this invention proposes a memory-type event triggering mechanism that resists outliers. Compared with the traditional event triggering mechanism, it can effectively reduce the over-triggering caused by random disturbances and the false triggering caused by outliers, save more network communication resources, improve the accuracy of fault detection and alarm in photovoltaic microgrid systems, and has certain engineering application value. Attached Figure Description

[0037] Figure 1 This is a flowchart of a method according to an embodiment of the present invention;

[0038] Figure 2 This is a distribution map of outlier points in the measurement signal used in the example.

[0039] Figure 3 The evaluation function curve is shown in the example using the memory-based event triggering mechanism for resisting outliers proposed in this invention.

[0040] Figure 4 The example uses the memory-based event triggering mechanism for anti-outlier points proposed in this invention to determine the triggering time and interval;

[0041] Figure 5The evaluation function curves of the memory-type event triggering mechanism with anti-outlier points proposed in this invention and the traditional event triggering mechanism are compared in the case of no faults. Detailed Implementation

[0042] The present invention will be further illustrated below with reference to specific embodiments. It should be understood that these embodiments are for illustrative purposes only and are not intended to limit the scope of the invention. After reading the present invention, any modifications of the present invention in various equivalent forms by those skilled in the art will fall within the scope defined by the appended claims.

[0043] like Figure 1 As shown, a memory-type event-triggered fault detection method for a photovoltaic microgrid with outliers includes the following steps:

[0044] Step 1: Establish the following stable closed-loop photovoltaic microgrid state-space model: in E and G are system parameter matrices.

[0045] Step 2: Design the memory-based event triggering mechanism for resisting outliers as follows: In the formula: t k h, t k+1 h represents the current and next trigger times, h represents the sampling period, and θ represents the interval length of historical data. These represent the historical data mean at the current sampling time and the historical data mean at the trigger time, respectively, with the symbol... express Let σ1 and σ2 be the event triggering error terms, representing the triggering threshold parameters. When the event triggering condition (3) is met, the mean signal at the current time is triggered, and the input of the fault detection filter is updated to... Otherwise, the mean signal at the current moment is not triggered, and the input to the fault detection filter retains the value from the previous trigger moment.

[0046] Step 3: Design the state-space model of the residual filter as follows: in τ k For network transmission latency, A d B d C d D d These are the parameters for the filter that needs to be designed.

[0047] A reference model for the fault signal is introduced to generate a desired residual evaluation result. The fault signal g(t) is then used to... The transfer function is Its state-space model can be represented as:

[0048] Using Simpson's rule, Lyapunov stability theory, Jensen's inequality, and the mutual convexity technique, the matrix inequalities that guarantee the stability of the system are derived: in:

[0049] By selecting the matrix We can obtain:

[0050] definition Substituting equation (19) into equation (18), we get: in:

[0051] The filter parameters can be obtained as follows:

[0052] To design a fault detection algorithm, a residual evaluation function is introduced.

[0053] In order to detect fault signals, a fault alarm threshold needs to be predetermined:

[0054] Based on the residual evaluation function and fault alarm threshold mentioned above, the following fault alarm conditions are given:

[0055] Substitute the event trigger matrix Ω and the filter parameters obtained in equation (21) into the closed-loop fault detection system to monitor the fault alarm conditions designed above. When the residual evaluation function Greater than the alarm threshold An alarm signal is generated to indicate that a fault has occurred; otherwise, it indicates that the system is in normal operating condition.

[0056] Step two of this invention proposes a memory-based event triggering mechanism to resist outliers. Compared with traditional event triggering mechanisms, this mechanism effectively reduces excessive triggering of measurement signals, lowers the transmission frequency of measurement signals, and saves limited network communication resources. In addition, it avoids false triggering caused by measurement outliers, improves the accuracy of fault detection, and provides a guarantee for the stable operation of the system.

[0057] An embodiment of the present invention is described below: Consider the photovoltaic microgrid system model in system (1), where the parameters are selected as follows: Table 1. Parameter values ​​for photovoltaic microgrid system

[0058] Select the following parameters: σ1 = 0.1, σ2 = 100. β=1, T M =0.06, θ=0.2, γ=1, solving the linear matrix inequality in formula (20), we can obtain the fault filter gain and trigger matrix as follows: C d = [-0.2627 -0.8621 0.7988], D d = -0.0195, Ω = 22.8654.

[0059] When performing system simulation, a simulation step size of 0.01s is selected, and the fault signal is: System disturbance w(t)=ψ(t)e -0.5t Let ψ(t) be a random variable that satisfies a uniform distribution, and |ψ(t)| < 5. Substitute these parameters into the photovoltaic microgrid fault detection system for simulation. Figure 1 The flowchart of the algorithm used in the example is shown below. Figure 2 This is a distribution map of outlier points in the measurement signal. Figure 3 The graph shows the evaluation function curves under the memory-based event triggering mechanism for resisting outliers. Curve a represents the evaluation function curve when a fault occurs, curve b represents the evaluation function curve when a fault occurs, and curve c represents the fault detection alarm threshold. Figure 4 The above figure shows the triggering time and interval generated by the memory-type event triggering mechanism for anti-outlier points proposed in this invention. The figure illustrates that when a fault occurs, the designed detection scheme can issue a fault alarm response after 0.58 seconds, effectively detecting the fault and providing support for subsequent system maintenance and repair.

[0060] To further illustrate the advantages of the proposed memory-based event triggering mechanism for preventing outliers, it is compared with traditional event triggering mechanisms. First, Table 2 presents the comparison results of the number of triggers generated by the proposed mechanism with existing mechanisms. Table 2 shows that, under the same simulation conditions, the proposed event triggering mechanism can significantly reduce the number of event triggers, effectively saving network communication resources. Table 2 Comparison of trigger counts generated by different triggering mechanisms

[0061] Secondly, a comparison of fault detection results under the two event-triggered mechanisms is presented when no fault occurs and only outliers in the measurement signal exist. Figure 5 The comparison results of the evaluation function curves under the two event triggering mechanisms are shown in the case of no faults. Curve a represents the evaluation function curve under the anti-outlier memory type event triggering mechanism proposed in this invention, curve b represents the evaluation function curve under the traditional event triggering mechanism, and curve c represents the fault detection alarm threshold. Figure 5 It should be noted that when no fault occurs, the event triggering mechanism proposed in this invention will not be affected by outliers and will not generate false alarms. However, the traditional event triggering mechanism will issue false alarm signals after outliers appear, which will interfere with the normal operation of the detection system and reduce the accuracy of fault detection.

Claims

1. A fault detection method for photovoltaic microgrids based on anti-outlier memory-type event triggering, characterized in that, It includes the following steps: S1. Establish the state-space model of a single-phase inverter in a photovoltaic microgrid; S2. Design a memory-based event-triggered communication mechanism based on the mean of historical data to resist outliers; S3. Derive the closed-loop residual filter system model, calculate the parameters of the residual filter and the event triggering mechanism, and perform fault detection for the photovoltaic microgrid. The fault detection method for photovoltaic microgrids based on anti-outlier memory events is characterized in that the state variables of the single-phase inverter can be used as the state variables of the photovoltaic microgrid. According to Kirchhoff's voltage and current laws, the state-space model of the photovoltaic microgrid is as follows: in C = [0 1 0]; y(t) represents the state vector, control input, and measurement output, respectively; i f (t), R f L f Represents the current, resistance, and inductance of an inductor-capacitor filter; R o L o Indicates the resistance and inductance of the transmission line; i o (t), v o (t) represents the output current and voltage; v c (t) represents the capacitor voltage; D represents the duty cycle of the Boost circuit; V sun This indicates the storage voltage of the photovoltaic microgrid; This is the gain of the state controller. Give the gain of the state feedback controller Make It is the Herwitz matrix. Considering external disturbances w(t) and faults g(t) caused by component aging, the following stable closed-loop photovoltaic microgrid state-space model can be obtained: in E and G are system parameter matrices. Using the average of historical data as the input signal for the event triggering mechanism can effectively reduce high-frequency triggering caused by measurement noise. Simultaneously, introducing an upper bound for the event triggering error term helps eliminate false triggers caused by outliers, improving the accuracy of fault detection. Specifically, this includes: Construct the following memory-based event triggering mechanism to resist outliers: In the formula: t k h, t k+1 h represents the current and next trigger times, h represents the sampling period, and θ represents the interval length of historical data. These represent the historical data mean at the current sampling time and the historical data mean at the trigger time, respectively, with the symbol... express Let σ1 and σ2 be the event triggering error terms, representing the triggering threshold parameters. When the event triggering condition (3) is met, the mean signal at the current time is triggered, and the input of the fault detection filter is updated to... Otherwise, the mean signal at the current moment is not triggered, and the input to the fault detection filter retains the value from the previous trigger moment. When the interval length of historical data approaches 0, i.e., θ→0, and without considering the upper bound constraint of the event triggering error term, the designed triggering mechanism will be simplified to the following conventional event triggering mechanism based on instantaneous measurement signals: in 2. The fault detection method for photovoltaic microgrids based on anti-outlier memory event triggering according to claim 1, characterized in that, A closed-loop residual filter system model was derived. Using Lyapunov's method, Simpson's rule, matrix transformation techniques, and linear matrix inequalities, the parameters of the residual filter and event triggering mechanism were solved. Based on this, a fault detection algorithm was designed to detect and provide early warnings for faults in the photovoltaic microgrid. Details are as follows: 1) Give the state-space model of the residual filter: in τ k For network transmission latency, A d B d C d D d These are the parameters for the filter that needs to be designed. 2) Introduce a reference model for the fault signal to generate a desired residual evaluation result. The fault signal g(t) is then used to... The transfer function is Its state-space model can be represented as: 3) Definition It can be represented as: 4) Simpson's rule is used to approximate the integral term in formula (7). This can be further expressed as: in 5) Definition The following fault detection closed-loop system can be obtained: in 6) To solve for the parameters of the filter and the event trigger, the following Lyapunov function is first selected: in: 7) Define the following new vector: 8) According to the above definition, the closed-loop system (9) can be expressed as: 9) The derivative of the Lyapunov function is: 10) Using Jensen's inequality and the technique of mutual convexity, the integral term in formula (13) can be scaled as follows: 11) To make the system satisfy H ∞ For stable performance, the derivative of the Lyapunov function must satisfy: 12) Based on the event triggering condition (3), we can obtain: in It is a positive number. 13) Combining the above conditions, the following matrix inequality form guarantees that (15) holds: 14) By utilizing And construct a matrix Formula (17) is equivalent to: in: 15) By selecting a matrix We can obtain: 16) Definition Substituting equation (19) into equation (18), we get: in: 17) The filter parameters can be obtained as follows: 18) To design a fault detection algorithm, a residual evaluation function is introduced. 19) In order to detect fault signals, a fault alarm threshold needs to be given in advance: 20) Based on the above residual evaluation function and fault alarm threshold, the following fault alarm conditions are given: 21) Substitute the event trigger matrix Ω and the filter parameters obtained in equation (21) into the closed-loop fault detection system to monitor the fault alarm conditions designed above. When the residual evaluation function Greater than the alarm threshold An alarm signal is generated to indicate that a fault has occurred; otherwise, it indicates that the system is in normal operating condition.

Images