A method and system for diagnosing open circuit faults in wind power parallel converter system

Abstract

The present invention belongs to the technical field of open circuit fault diagnosis. The present invention provides an open circuit fault diagnosis method and system for a wind power parallel conversion system. The present invention calculates three-phase current residuals based on a predictive control framework; designs a convolution kernel to extract convolution fault features of the three-phase current residuals to amplify the features of the three-phase current residuals at the time of a fault; sets a fault threshold according to a characteristic value and a convolution kernel to perform fault diagnosis and positioning; the present invention can realize rapid diagnosis, is not affected by working conditions, has low parameter sensitivity, and improves the reliability and availability of the system.

Description

Technical Field

[0001] The present invention belongs to the technical field of open circuit fault diagnosis, and in particular relates to an open circuit fault diagnosis method and system for a wind power parallel conversion system. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] Deep sea wind resources have huge development potential. Due to its advantages such as more stable wind speed and proximity to power load centers, it has become a research hotspot in the field of wind power generation. However, due to the complexity of the deep sea environment and its high construction cost, the development of offshore wind resources faces many challenges. In order to cope with these problems and enhance the competitiveness of offshore wind power generation systems, the key is to increase the power generation capacity of a single wind turbine, thereby reducing the cost of power generation. Therefore, the use of large-capacity offshore wind turbines has become an industry development trend. With the gradual increase in the rated power of wind turbines, a single converter can no longer meet its power requirements. To this end, operating multiple wind power converters in parallel has become an effective solution to improve the capacity and reliability of wind power generation systems.

[0004] The harsh environment of offshore wind power has seriously affected the safe operation of electrical equipment. At the same time, with the increase in the capacity of wind power generation systems, the failure rate of wind power conversion systems has also increased accordingly. Among them, open circuit faults are a common type of fault in wind power converters, which will cause the deterioration of three-phase current or voltage, and then cause secondary faults in other components, and even cause the entire converter system to shut down. Therefore, open circuit fault diagnosis and fault location of wind power parallel converter systems are important measures to reduce maintenance costs and improve system reliability.

[0005] Young-Jong Ko proposed a current vector shape method for parallel two-level converter systems, which detects converter faults by analyzing the stator current pattern within two cycles, and the diagnosis time is one current fundamental cycle. Ahmed Shahin combined the characteristics of dq axis current and zero-sequence circulating current when the power device fails to diagnose the power device fault for the parallel two-level converter system. Bo Long adopted a fault diagnosis method driven by mechanism and data for three-level converters. In the offline model analysis and learning stage, an analysis model of T-type three-level inverter was established, a mathematical model was constructed, and a fault diagnosis variable data pool was created. At the same time, experimental data collection and variable optimization were used to provide accurate input for the neural network. In the online fault diagnosis stage, the trained neural network was used for fault location and diagnosis, which improved the accuracy and efficiency of diagnosis and solved the problem that the fault characteristics of vertical and horizontal bridge arms may be similar and affect the accuracy of diagnosis.

[0006] Among the above methods, the current vector shape method proposed by Young-Jong Ko requires sampling and analysis of the current of a fundamental wave cycle, and its diagnosis time is too long. The fault diagnosis method combining dq axis current and zero-sequence circulating current proposed by Ahmed Shahin is applicable to a single working condition and is highly sensitive to parameters. The fault diagnosis method of mechanism data fusion proposed by Bo Long is only for three-level converters. Its diagnosis basis is that when the power device fails, the capacitor voltage on the upper and lower sides of the DC side neutral point will change. However, the parallel two-level converter system does not have a DC side neutral point, and its DC side capacitor voltage will not change when the power device fails. Therefore, this method is not applicable to the parallel two-level converter system. Summary of the invention

[0007] In order to solve the above problems, the present invention proposes an open circuit fault diagnosis method and system for a wind power parallel conversion system. The present invention calculates the three-phase current residual under a predictive control framework; designs a convolution kernel to extract the fault features of the three-phase current residual to amplify the features of the three-phase current residual at the time of the fault; sets the fault threshold according to the characteristic value and the convolution kernel to perform fault diagnosis and positioning; the present invention can realize rapid diagnosis, is not affected by working conditions, has low parameter sensitivity, and improves the reliability and availability of the system.

[0008] According to some embodiments, the present invention adopts the following technical solutions:

[0009] A method for diagnosing an open circuit fault in a wind power parallel converter system comprises the following steps:

[0010] According to the topological structure and electrical parameters of the wind power parallel conversion system, a continuous model of the wind power parallel conversion system is constructed;

[0011] Predict the grid-side output current of parallel converter systems;

[0012] Calculating three-phase current residuals based on the predicted grid-side output current and the continuous model;

[0013] The convolution kernel and sliding window coefficient are set, and the three-phase current residuals are normalized according to the maximum value of the root mean square of the three-phase current residuals. Then, the normalized three-phase current residuals are convolved to extract fault features;

[0014] Compare the extracted feature values ​​with the preset fault thresholds, and perform fault diagnosis and location based on the comparison results.

[0015] As an optional implementation, the wind power parallel conversion system is a two-level wind power parallel conversion system.

[0016] As an optional implementation method, the process of constructing a continuous model of a wind power parallel conversion system includes: using a gate signal to define the switch state of a two-level parallel wind power conversion system, combining the topological result of the two-level parallel wind power conversion system, and applying Kirchhoff's voltage law to obtain a continuous model of the two-level parallel wind power conversion system:

[0017] ;

[0018] in, and Respectively represent the output phase current and voltage, x is a, b or c; m is 1 or 2, indicating the converter number in the converter system; , , is the voltage of the three-phase voltage source, Output phase voltage, and are the filter inductor and filter resistor.

[0019] As an optional implementation, a model predictive current control method is used to predict the grid-side output current of the parallel converter system.

[0020] As an optional implementation, the process of calculating the three-phase current residual includes: establishing a cost function, discretizing the converter phase current at the sampling time, replacing the row phase current derivative with a forward Euler approximation, and using the modified formula to replace the continuous model of the parallel wind power conversion system to calculate the three-phase current residual.

[0021] As a further implementation, the cost function is:

[0022] ;

[0023] In the formula, It refers to predicting the corresponding phase current vector of the converter; is the reference current vector of the corresponding phase, x is a, b or c, m is 1 or 2, and represents the converter number in the converter system.

[0024] As an optional implementation, the process of setting the convolution kernel and the sliding window coefficient and extracting the convolution fault features of the three-phase current residual includes:

[0025] Set the step convolution kernel and set the window size of the sliding window to the length of the convolution kernel;

[0026] The three-phase current residuals on the grid side are normalized according to the maximum value of the root mean square of the three-phase current residuals;

[0027] Convolution calculation is performed on the normalized data of the three-phase current residuals within the sliding window.

[0028] As an alternative implementation, the fault threshold is set to ,in, is the step convolution kernel, and is calculated If it is greater than Thrl, the lower tube of phase x is faulty. If it is less than -Thrl, the upper tube of phase x is faulty. k The calculation results of sliding window convolution at the moment , Kl is the window size of the sliding window, k The data in the time sliding window is , the normalized value of the three-phase current residual is , , rms( ) and rms( ) is the RMS of the three-phase current residuals in the sliding window, and is the maximum value of the three-phase current residual of different converters, and is the three-phase current residual of different converters.

[0029] An open circuit fault diagnosis system for a wind power parallel converter system, comprising:

[0030] A model building module is configured to build a continuous model of the wind power parallel conversion system according to the topological structure and electrical parameters of the wind power parallel conversion system;

[0031] A current prediction module is configured to predict a grid-side output current of a parallel converter system;

[0032] A current residual calculation module, configured to calculate the three-phase current residual based on the predicted grid-side output current and the continuous model;

[0033] A fault feature extraction module is configured to set a convolution kernel and a sliding window coefficient, normalize the three-phase current residuals, and extract fault features through convolution calculation;

[0034] The fault diagnosis module is configured to compare the extracted characteristic value with a preset fault threshold, and perform fault diagnosis and location based on the comparison result.

[0035] An electronic device comprises a memory and a processor, and computer instructions stored in the memory and executed on the processor. When the computer instructions are executed by the processor, the steps in the above method are completed.

[0036] Compared with the prior art, the present invention has the following beneficial effects:

[0037] Compared with the traditional method, the present invention normalizes the three-phase current residual and then performs sliding window convolution feature extraction, so that the fault characteristics will not change with the change of system parameters, and has parameter desensitization;

[0038] The present invention utilizes the existing current sensor of the system and does not require additional hardware, thereby reducing economic costs;

[0039] The method proposed in the present invention is easier to expand to various topologies because there is no need to modify the control strategy.

[0040] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0042] Figure 1 is a schematic diagram of a two-level wind power parallel conversion system according to an embodiment;

[0043] Figure 2 The present invention is a schematic diagram of a fault diagnosis process of an embodiment. DETAILED DESCRIPTION

[0044] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0045] It should be noted that the following detailed descriptions are all illustrative and intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0046] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0047] In the absence of conflict, the embodiments in this application and the features in the embodiments may be combined with each other.

[0048] Embodiment 1

[0049] A method for open-circuit fault diagnosis of parallel two-level wind power conversion system is proposed. Based on the predictive control framework, the three-phase current residual is calculated. A convolution kernel is designed to extract the convolution fault feature of the three-phase current residual to amplify the characteristics of the three-phase current residual at the time of fault. Finally, the fault threshold is set according to the characteristic value and the convolution kernel to perform fault diagnosis and location. This method has rapid diagnosis, is not affected by working conditions, has low parameter sensitivity, and improves the reliability and availability of the system.

[0050] The specific process is described below.

[0051] This embodiment uses a two-level parallel converter system topology to illustrate the specific implementation steps of the method.

[0052] like Figure 2 As shown, the following steps are included:

[0053] Step 1: Calculate the three-phase current residual on the grid side:

[0054] (1) First, establish a system model

[0055] The topological structure of a two-level parallel wind power conversion system (taking the grid-side converter as an example) is shown in Figure 1.

[0056] The two-level parallel wind power conversion system includes a three-phase voltage source ( , , ), three with equivalent resistance Grid-side filter inductor , 12 IGBTs ( x { a, b, c}, y {1, 2, 3, 4}, the same below) and 12 anti-parallel freewheeling diodes , and the DC link capacitor , and are the filter inductor and filter resistor. Figure 1 middle, represents the capacitor voltage, and (m {1, 2}, the same below) represents the output phase current and voltage.

[0057] The switch state of the two-level parallel wind power conversion system can be defined by the gate signal:

[0058] (1)

[0059] The output voltage of phase x can be described as:

[0060] (2)

[0061] like Figure 1 As shown, applying Kirchhoff's voltage law, the continuous model of the two-level parallel wind power conversion system is obtained as follows:

[0062] (3)

[0063] (2) Calculation of three-phase current residuals based on the predictive control framework

[0064] In this embodiment, a model predictive current control (MPCC) algorithm is used.

[0065] First, establish the cost function g:

[0066] (4)

[0067] In the formula, It refers to the predicted converter phase current vector at time k+1; is the reference current vector at time k+1.

[0068] For sampling time The converter phase current formula is discretized and the phase current derivative is replaced by the forward Euler approximation. , that is, approximating the derivative by the following formula:

[0069] (5)

[0070] Substituting equation (3) with equation (5) yields the following expression:

[0071] (6)

[0072] This expression predicts the current value at time k+1 for each of the seven voltage vectors generated by the converter, where is the predicted current vector at time k+1, is the current vector at time k, is the voltage vector at time k, is the voltage vector of the three-phase voltage source at time k, and x is a, b or c. The actual current value at time k+1 can be measured by the current sensor. In this way, the three-phase current residual at time k+1 can be calculated. :

[0073] (7)

[0074] Step 2: Use sliding window convolution to extract fault features

[0075] (1) Set the convolution kernel and sliding window coefficient

[0076] When an open-circuit fault occurs in a power device, the phase current cannot flow through the faulty power device, and the expected current path is blocked, resulting in the actual switching state of the converter being different from the expected switching state. At this time, the phase voltage calculated based on formula (2) will also deviate from the expected phase voltage. This deviation is transmitted to the grid-side phase current through formula (3), causing the three-phase current residual to change stepwise when a fault occurs. Therefore, a step convolution kernel is set. In this embodiment, it is set to , where b>a>0, the window size of the sliding window is the length of the convolution kernel Kl, and the sliding window step size is set to 1.

[0077] (2) Normalize the three-phase current residual on the grid side according to the maximum value of the three-phase current residual root mean square

[0078] First, calculate the root mean square rms of the three-phase current residuals in the sliding window ( ) and rms( ), find the maximum value of the three-phase current residual and . Normalize the three-phase current residuals: , , , is the normalized three-phase current residual; and is the three-phase current residual of different converters.

[0079] (3) Perform convolution calculation on the three-phase current residual normalized data within the sliding window

[0080] The data in the sliding window at time k is ; The calculation result of sliding window convolution at time k is .

[0081] Step 3: Fault diagnosis and fault location

[0082] The fault threshold is set to , real-time computing If it is greater than Thrl, the lower tube of phase x is faulty; if it is less than -Thrl, the upper tube of phase x is faulty.

[0083] Embodiment 2

[0084] An open circuit fault diagnosis system for a wind power parallel converter system, comprising:

[0085] A model building module is configured to build a continuous model of the wind power parallel conversion system according to the topological structure and electrical parameters of the wind power parallel conversion system;

[0086] A current prediction module is configured to predict a grid-side output current of a parallel converter system;

[0087] A current residual calculation module, configured to calculate the three-phase current residual based on the predicted grid-side output current and the continuous model;

[0088] A fault feature extraction module is configured to set a convolution kernel and a sliding window coefficient to perform convolution fault feature extraction on the three-phase current residual;

[0089] The fault diagnosis module is configured to compare the extracted characteristic value with a preset fault threshold, and perform fault diagnosis and location based on the comparison result.

[0090] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made by those skilled in the art within the spirit and principle of the present invention without creative labor shall be included in the protection scope of the present invention.

Claims

1. A method for open circuit fault diagnosis of a wind power parallel converter system, characterized in that: The following steps are involved: According to the topological structure and electrical parameters of the wind power parallel conversion system, a continuous model of the wind power parallel conversion system is constructed; Predict the grid-side output current of parallel converter systems; Calculating three-phase current residuals based on the predicted grid-side output current and the continuous model; Set the convolution kernel and sliding window coefficient to extract fault features of the three-phase current residuals; Compare the extracted feature values ​​with the preset fault thresholds, and perform fault diagnosis and location based on the comparison results; The process of calculating the three-phase current residual includes: establishing a cost function, for the sampling time The converter phase current formula is discretized and the phase current derivative is replaced by the forward Euler approximation. , forming an expression for approximating the derivative, using the expression to replace the continuous model of the parallel wind power conversion system, and calculating the three-phase current residual; The cost function is: ； In the formula, It refers to predicting the corresponding phase current vector of the converter; is the reference current vector of the corresponding phase, x is a, b or c, m is 1 or 2, indicating the converter number in the converter system; The process of setting the convolution kernel and sliding window coefficient and extracting the fault features of the three-phase current residual by convolution includes: Set the step convolution kernel and set the window size of the sliding window to the length of the convolution kernel; The three-phase current residuals on the grid side are normalized according to the maximum value of the root mean square of the three-phase current residuals; Perform convolution calculation on the three-phase current residual normalized data within the sliding window; The fault threshold is set to ,in, is the step convolution kernel, and is calculated If it is greater than Thrl, the lower tube of phase x is faulty. If it is less than -Thrl, the upper tube of phase x is faulty. k The calculation results of sliding window convolution at the moment , Kl is the window size of the sliding window, k The data in the time sliding window is , the normalized value of the three-phase current residual is , , rms( ) and rms( ) is the RMS of the three-phase current residuals in the sliding window, and is the maximum value of the three-phase current residual of different converters, and is the three-phase current residual of different converters.

2. The open circuit fault diagnosis method for a wind power parallel converter system according to claim 1, characterized in that: The wind power parallel conversion system is a two-level wind power parallel conversion system.

3. The open circuit fault diagnosis method for a wind power parallel converter system according to claim 1, characterized in that: The process of constructing the continuous model of the wind power parallel conversion system includes: using the gate signal to define the switch state of the two-level parallel wind power conversion system, combining the topological results of the two-level parallel wind power conversion system, and applying Kirchhoff's voltage law to obtain the continuous model of the two-level parallel wind power conversion system: ； in, , , is the voltage of the three-phase voltage source, and are the filter inductor and filter resistor.

4. The open circuit fault diagnosis method for a wind power parallel converter system according to claim 1, characterized in that: The grid-side output current of the parallel converter system is predicted using the model predictive current control method.

5. An open circuit fault diagnosis system for a wind power parallel converter system, characterized in that: include: A model building module is configured to build a continuous model of the wind power parallel conversion system according to the topological structure and electrical parameters of the wind power parallel conversion system; A current prediction module is configured to predict a grid-side output current of a parallel converter system; A current residual calculation module, configured to calculate the three-phase current residual based on the predicted grid-side output current and the continuous model; A fault feature extraction module is configured to set a convolution kernel and a sliding window coefficient to perform convolution fault feature extraction on the three-phase current residual; A fault diagnosis module is configured to compare the extracted characteristic value with a preset fault threshold, and perform fault diagnosis and location based on the comparison result; The process of calculating the three-phase current residual includes: establishing a cost function, for the sampling time The converter phase current formula is discretized and the phase current derivative is replaced by the forward Euler approximation. , forming an expression for approximating the derivative, using the expression to replace the continuous model of the parallel wind power conversion system, and calculating the three-phase current residual; The cost function is: ； In the formula, It refers to predicting the corresponding phase current vector of the converter; is the reference current vector of the corresponding phase, x is a, b or c, m is 1 or 2, indicating the converter number in the converter system; The process of setting the convolution kernel and sliding window coefficient and extracting the fault features of the three-phase current residual by convolution includes: Set the step convolution kernel and set the window size of the sliding window to the length of the convolution kernel; The three-phase current residuals on the grid side are normalized according to the maximum value of the root mean square of the three-phase current residuals; Perform convolution calculation on the three-phase current residual normalized data within the sliding window; The fault threshold is set to ,in, is the step convolution kernel, and is calculated If it is greater than Thrl, the lower tube of phase x is faulty. If it is less than -Thrl, the upper tube of phase x is faulty. k The calculation results of sliding window convolution at the moment , Kl is the window size of the sliding window, k The data in the time sliding window is , the normalized value of the three-phase current residual is , , rms( ) and rms( ) is the RMS of the three-phase current residuals in the sliding window, and is the maximum value of the three-phase current residual of different converters, and is the three-phase current residual of different converters.

6. An electronic device, characterized in that: The method comprises a memory and a processor, and computer instructions stored in the memory and executed on the processor, wherein when the computer instructions are executed by the processor, the steps in the method according to any one of claims 1 to 4 are completed.

Images