Online fault diagnosis method for hydro-generator based on potential interval distance of through-core screw

Abstract

The application discloses an online fault diagnosis method for a hydro-generator based on a potential interval distance of a through screw, relates to the technical field of generators, and is suitable for detecting a stator winding ground fault. Voltage collection devices are additionally arranged on a plurality of through screws of the hydro-generator, real-time induced potential is acquired, a historical data matrix is established, and a data buffer matrix is updated at interval time; the data buffer matrix is divided into a data evaluation matrix, a fundamental wave and an odd harmonic phase are extracted through Fourier decomposition, a harmonic phase matrix is constructed, a phase difference matrix is obtained by calculating normalized phase differences of adjacent columns, a quartile distance is solved after being converted into a column vector array, a dynamic quartile distance array is formed, abnormal limit values are determined by combining a box plot principle and an error level of the voltage collection device, and faults are determined by traversing the array. The method does not need additional sensing devices and traditional threshold setting, is convenient to operate, has low implementation difficulty, and can accurately and timely identify faults.

Description

Technical Field

[0001] This invention relates to the field of generator technology, specifically to an online fault diagnosis method for hydro-generators based on the potential interval of the through-bolt, applicable to online detection of stator winding grounding faults in hydro-generators. Background Technology

[0002] Large hydro-turbine generators are the core equipment of hydropower systems, and their operational reliability plays an extremely important role in the normal operation of the power system, ensuring power quality, uninterrupted power supply to users, and the safe operation of the entire society. However, grounding faults frequently occur in the stator windings of large-capacity hydro-turbine generator units. This is not only the most probable type of fault, but it is also often a precursor to more serious internal phase-to-phase or turn-to-turn short-circuit faults. If grounding faults are not detected and addressed in time, they can cause serious damage to the generator and the power grid system. Therefore, accurately and promptly determining whether a grounding short-circuit fault has occurred in the stator windings of a hydro-turbine generator is of paramount importance.

[0003] Currently, the characteristic quantities used in generator stator winding ground fault detection methods are easily affected by factors such as the neutral point connection method, winding-to-ground capacitance, and changes in operating conditions. This is especially true for large hydro-generators, where achieving high sensitivity is difficult. For example, the fundamental / third harmonic zero-sequence voltage detection method comprehensively identifies ground faults by detecting the fundamental / third harmonic zero-sequence voltage at the generator terminals or neutral point. However, the fault threshold setting is highly correlated with unit parameters, and its sensitivity to high-resistance ground faults is low. The impressed current detection method adds an external signal power supply between the generator stator circuit and the ground, and identifies ground faults by monitoring the amplitude of the injected signal in real time. This method requires additional power supply equipment and has high requirements for the reliability and performance of the power supply, making on-site debugging complex. Furthermore, detection methods based on the transient components of the initial fault stage are mostly designed for steam turbine generators, with limited research on methods for hydro-generators.

[0004] Furthermore, regarding the fault detection problem of large hydro-generators, domestic and foreign scholars have begun to utilize the unique through-screw structure of hydro-generators. Considering that the induced electromotive force (EMF) of the through-screw is closely related to the generator's magnetic flux density, the operating status of the generator can be reflected by real-time acquisition of the induced EMF of the through-screw. For example, Chinese patent literature, publication number CN106443318A, publication date February 22, 2017, entitled "Invention of a Method for Diagnosing Inter-turn Short Circuit in Rotor Winding of Hydro-generator Based on Through-Screw," utilizes an online acquisition system to collect the induced voltage of the through-screw in real time. The collected induced voltage of the through-screw is processed in real time, and the deviation of the induced voltage on the through-screw when different magnetic poles sweep across the same position is compared. When the voltage difference exceeds a set threshold, it is determined that the hydro-generator has an inter-turn short circuit fault in the rotor winding. However, under stator winding grounding faults, the amplitude change characteristics of the induced EMF are weakened, and the sensitivity of fault identification decreases. Therefore, a new method is urgently needed to determine whether a grounding fault has occurred in the stator winding of a hydro-generator. Summary of the Invention

[0005] To address the shortcomings of the existing technology, this invention proposes an online fault diagnosis method for hydro-generators based on the potential interval of the through-bolts. This method utilizes the induced potential of multiple through-bolts in the hydro-generator to construct a dynamic interval array of the through-bolt potential, which is then used to perform box plot anomaly identification. This method is suitable for online detection of stator winding grounding faults in hydro-generators.

[0006] This invention is achieved through the following technical solution: A method for online fault diagnosis of hydro-generators based on the potential distribution distance of the through-screw, characterized by the following steps: S1, at any magnetic pole of the hydro-generator A voltage acquisition device is installed on each of the through-hole screws to acquire voltage data in real time. The induced electromotive force of the through-hole screw; S2, based on the sampling rate of the voltage acquisition device Set the length of time for storing historical data. , build × dimensional historical data matrix ; for The number of voltage values ​​acquired by the voltage acquisition device within a given time period; S3, every other time At that moment, make use of that period of time × Weixin monitoring data Slide to update historical data matrix Obtain the data buffer matrix ; for The number of newly acquired voltage values ​​by the internal voltage acquisition device; S4. Buffer the data matrix Divide into equal rows Data evaluation matrix ; S5. Evaluate the matrix for each data point. of Each column is subjected to Fourier decomposition to extract the phase values ​​of the fundamental wave and its odd harmonics, forming a harmonic phase matrix. ; S6. Calculate the phase matrix for each harmonic. The phase difference between two adjacent columns is calculated and normalized using the mean of the fundamental phase difference as a reference, to obtain the phase difference matrix. ; S7, Phase difference matrix Convert to column vector array and calculate the array Upper and lower quartiles , and its interquartile range ,Will The interquartile ranges constitute a dynamic interquartile range array. ; Represents a column vector array The upper quartiles in the column vector array After sorting the data in ascending order, 75% of the data are at or below the 75th percentile. Represents a column vector array The lower quartiles in the column vector array After sorting the data in ascending order, 25% of the data are at or below the 25th percentile. S8. Calculate the dynamic percentile array. Upper and lower quartiles and Obtain the upper and lower limits of the box plot. and Then, the error level of the voltage acquisition device is introduced to obtain the upper and lower limits of abnormality. and ; Represents the dynamic quantum distance array The upper quartile is the value at the 75th percentile after sorting the array data in ascending order. 75% of the data in the array are less than or equal to this value. Represents the dynamic quantum distance array The lower quartile is the value at the 25th percentile after the data in the array is sorted from smallest to largest. 25% of the data are less than or equal to this value. S9. Traverse the dynamic percentile array If there exists an element with a value greater than the upper limit or less than the lower limit If so, it is determined that the hydro-generator unit has malfunctioned.

[0007] More preferably, in step S2, the historical data matrix The expression is:

[0008]

[0009] In the formula: For the first The first core screw was monitored and acquired. The voltage acquisition value of each; for The number of voltage values ​​acquired by the voltage acquisition device within a given time period.

[0010] More preferably, the Second.

[0011] More preferably, in step S3, the data buffer matrix... The expression is:

[0012] In the formula: This refers to the data update interval. for The number of new voltage acquisition values ​​obtained by the internal voltage acquisition device.

[0013] More preferably, the The range of values ​​is Second.

[0014] More preferably, in step S4, the data evaluation matrix The expression is:

[0015] In the formula: It is a data evaluation matrix. Number.

[0016] More preferably, in step S5, the harmonic phase matrix The expression is:

[0017] In the formula: For the first Data evaluation matrix The Middle Column data Second harmonic phase value; variable The corresponding fundamental wave and its odd harmonic orders ;parameter The value corresponding to the highest odd harmonic order is considered, which depends on the generator operating parameters.

[0018] More preferably, in step S6, the phase difference matrix The expression is:

[0019] In the formula: Phase difference matrix The Okay, number The column element values ​​represent the harmonic phase matrix. Middle elements and The normalized value of the difference; variable Values ,variable Values ; for The first of the matrix Okay, number Column element values.

[0020] More preferably, in step S7, the dynamic quantile array The calculation expression is:

[0021] In the formula: For matrix The List all element values.

[0022] More preferably, in step S8, the abnormal upper and lower limits are... and The calculation expression is:

[0023] In the formula: This represents the upper limit of the box plot; This represents the lower limit of the box plot. for The upper quartiles; for The lower quartile; parameter This is an adjustment factor, which depends on the error level of the voltage acquisition device; This is the upper limit of the abnormal value; This is the lower limit of the abnormal value.

[0024] The beneficial effects of this invention are as follows: 1. Under stator winding grounding faults, the amplitude variation characteristics of the induced electromotive force in the through-core screw may weaken, leading to a decrease in the sensitivity of fault identification. This invention extracts the fundamental and odd harmonic phases through Fourier decomposition, calculates the adjacent phase differences, and normalizes them to form a phase difference matrix. The phase difference matrix is ​​then converted into a column vector array. By calculating the upper quartile, lower quartile, and interquartile distance, a dynamic interquartile distance array is constructed using the interquartile distance, which reflects the phase distribution changes in real time. Even if the amplitude change of the induced electromotive force is not obvious, abnormal phase distribution can be accurately captured, thus solving the problem of insufficient sensitivity in fault identification.

[0025] 2. Traditional methods rely on manually preset fixed thresholds. Thresholds that are too wide are prone to missed detections, while those that are too strict are prone to misjudgments. Furthermore, they cannot adapt to different unit parameters and the errors of the acquisition devices. This invention constructs a dynamic quartile array by considering the distribution characteristics of the phase of the induced potential of the through-hole screw. Through the principle of box plot outlier detection, it dynamically generates a basic threshold using the quartiles of historical data. Then, it introduces the error level of the voltage acquisition device to correct and obtain the final abnormal threshold. No manual intervention is required. The grounding short circuit fault of the hydro-generator can be accurately identified through data-driven methods, avoiding the defects of manually set subjective thresholds and reducing the probability of misjudgment and missed detection.

[0026] 3. This invention constructs a historical data matrix, then obtains a data buffer matrix through sliding updates, sets the historical data storage time length and dynamic update interval, and realizes the dynamic fusion of historical data and real-time new data, so that the diagnostic data always fits the current operating conditions.

[0027] 4. The operation of hydro-generators generates a large amount of monitoring data. Direct processing of this data results in a large amount of computation, affecting the timeliness of online diagnosis. By dividing the data buffer matrix into multiple data evaluation matrices by row, a large amount of data can be processed in blocks. Block processing reduces the amount of computation per operation and ensures real-time performance. Each evaluation matrix independently extracts phase features and contributes them to the dynamic quantile array, ensuring the comprehensiveness of feature extraction and balancing real-time performance and accuracy.

[0028] 5. This invention does not require additional sensing equipment. It only uses a voltage acquisition device to obtain the voltage acquisition values ​​of multiple through-bolts of the hydro-generator. It does not rely on other parameters of the unit. The operation process is simple and quick, the implementation difficulty is low, it can be adapted to hydro-generators of different specifications, and it has high engineering versatility. Attached Figure Description

[0029] Fig. 1 This is a flowchart of the method of the present invention; Fig. 2 Diagram showing the connection and acquisition of the induced electromotive force of the through-hole screw; Fig. 3 This is a diagram showing the induced electromotive force distribution of the through-hole screw before and after a unidirectional grounding fault. Detailed Implementation

[0030] Example 1 like Figs. 1-3 As shown, the online fault diagnosis method for hydro-generators based on the potential gradation distance of the through-bolt includes the following steps: S1, at any magnetic pole of the hydro-generator A voltage acquisition device is installed on each of the through-hole screws to acquire voltage data in real time. The induced electromotive force of the through-hole screw; S2, based on the sampling rate of the voltage acquisition device Set the length of time for storing historical data. , build × dimensional historical data matrix ; for The number of voltage values ​​acquired by the sampling device within a given time period; S3, every other time At that moment, make use of that period of time × Weixin monitoring data Slide to update historical data matrix Obtain the data buffer matrix ; for The number of newly acquired voltage values ​​by the internal voltage acquisition device; S4. Buffer the data matrix Divide into equal rows Data evaluation matrix ; S5. Evaluate the matrix for each data point. of Each column is subjected to Fourier decomposition to extract the phase values ​​of the fundamental wave and its odd harmonics, forming a harmonic phase matrix. ; S6. Calculate the phase matrix for each harmonic. The phase difference between two adjacent columns is calculated and normalized using the mean of the fundamental phase difference as a reference, to obtain the phase difference matrix. ; S7, Phase difference matrix Convert to column vector array and calculate the array Upper and lower quartiles , and its interquartile range ,Will The interquartile ranges constitute a dynamic interquartile range array. ; S8. Calculate the dynamic percentile array. upper and lower quartiles and Obtain the upper and lower limits of the box plot. and Then, the error level of the voltage acquisition device is introduced to obtain the upper and lower limits of abnormality. and ; S9. Traverse the dynamic percentile array If there exists an element with a value greater than the upper limit or less than the lower limit If so, it is determined that the hydro-generator unit has malfunctioned.

[0031] To address the issue of insignificant changes in the amplitude of induced potential during stator winding grounding faults, a phase difference matrix is ​​formed by extracting the fundamental and odd harmonic phases through Fourier decomposition of each data buffer matrix, calculating the adjacent phase differences, and normalizing the results. This phase difference matrix is ​​then transformed into a column vector array. By calculating the upper and lower quartiles and the interquartile range, a dynamic interquartile range array is constructed, which reflects changes in phase distribution in real time. Even if the amplitude of the induced potential does not change significantly, abnormal phase distribution can be accurately detected, thus solving the problem of insufficient sensitivity caused by weakened fault characteristics. Traditional methods rely on manually preset fixed thresholds. Thresholds that are too wide are prone to missed detections, while those that are too strict are prone to misjudgments. Furthermore, they cannot adapt to different unit parameters and the errors of the data acquisition devices. By calculating the upper and lower limits of the box plot using the upper and lower quartiles of a dynamic quantile group, and by introducing error level correction from the voltage acquisition device to obtain the upper and lower limits of abnormalities, misjudgments or missed detections caused by voltage acquisition device errors are avoided. This makes the abnormal limits more in line with actual operating conditions. By traversing the dynamic quantile array, it is determined whether there are elements greater than the upper limit of the abnormality or less than the lower limit of the abnormality, thus completing the fault determination. No manual intervention is required. Data-driven methods can accurately identify grounding short-circuit faults in hydro-generators, avoiding the defects of subjective thresholds set manually and reducing the probability of misjudgments and missed detections.

[0032] Example 2 This embodiment is a further detailed description and supplement to the implementation of the present invention based on Embodiment 1.

[0033] In step S2, the historical data matrix The expression is:

[0034]

[0035] In the formula: For the first The first core screw was monitored and acquired. The voltage acquisition value of each; for The number of voltage values ​​acquired by the voltage acquisition device within a given time period.

[0036] More preferably, the Second.

[0037] Specifically, in the hydro-generator A voltage acquisition device is installed on each of the through-hole screws, according to the sampling rate. Real-time acquisition of voltage values, setting the historical data storage time length. Calculate the number of sampling points within this time period. , build × dimensional historical data matrix .

[0038] Example 3 This embodiment further elaborates and supplements the implementation of the present invention based on Embodiment 1 or Embodiment 2.

[0039] Set data update interval ,calculate Number of new sampling points within a time period Every At any moment, with newly collected × Dimensional new monitoring data matrix Replace row 1-row 1 in the historical data matrix Rows of data are used to form an updated data buffer matrix. The data buffer matrix is ​​a fixed-length, sliding-up real-time data window that always retains the voltage acquisition values ​​within the latest time period. In step S3, the data buffer matrix... The expression is:

[0040] In the formula: This refers to the data update interval. for The number of new voltage acquisition values ​​obtained by the internal voltage acquisition device.

[0041] More preferably, the The range of values ​​is Second.

[0042] Example 4 This embodiment further elaborates and supplements the implementation of the present invention based on Embodiment 1, Embodiment 2 or Embodiment 3.

[0043] In step S4, the data evaluation matrix The expression is:

[0044] In the formula: It is a data evaluation matrix. Number.

[0045] Example 5 This embodiment further elaborates and supplements the implementation of the present invention based on Embodiment 1, Embodiment 2, Embodiment 3 or Embodiment 4.

[0046] In step S5, the harmonic phase matrix The expression is:

[0047] In the formula: For the first Data evaluation matrix The Middle Column data Second harmonic phase value; variable The corresponding fundamental wave and its odd harmonic orders ;parameter The value corresponding to the highest odd harmonic order is considered, which depends on the generator operating parameters and the large odd harmonics present in actual operating conditions.

[0048] Example 6 This embodiment further elaborates and supplements the implementation of the present invention based on Embodiment 1, Embodiment 2, Embodiment 3, Embodiment 4 or Embodiment 5.

[0049] In step S6, the phase difference matrix The expression is:

[0050] In the formula: Phase difference matrix The Okay, number The column element values ​​represent the harmonic phase matrix. Middle elements and The normalized value of the difference; variable Values ,variable Values ; for The first of the matrix Okay, number Column element values.

[0051] In step S7, the dynamic quantile array The calculation expression is:

[0052] In the formula: For matrix The List all element values.

[0053] More preferably, in step S8, the abnormal upper and lower limits are... and The calculation expression is:

[0054] In the formula: This represents the upper limit of the box plot; This represents the lower limit of the box plot. for The upper quartiles; for The lower quartile; parameter The adjustment factor depends on the error level of the voltage acquisition device, which typically has an acquisition error value. This is the upper limit of the abnormal value; This is the lower limit of the abnormal value.

[0055] Example 7 This embodiment is a further detailed description and supplement to the implementation of the present invention based on Embodiment 6.

[0056] The online fault diagnosis method for hydro-generators based on the potential gradation distance of the through-bolt includes the following steps: S1, at any magnetic pole of the hydro-generator A voltage acquisition device is installed on each of the through-hole screws to acquire voltage data in real time. The induced electromotive force of the through-hole screw; S2, based on the sampling rate of the voltage acquisition device Set the length of time for storing historical data. , build × dimensional historical data matrix :

[0057]

[0058] In the formula: For the first The first core screw was monitored and acquired. The voltage acquisition value of each; for The number of voltage values ​​acquired by the voltage acquisition device within a given time period; Second.

[0059] S3, every other time At that moment, make use of that period of time × Weixin monitoring data Slide to update historical data matrix Obtain the data buffer matrix :

[0060] In the formula: This refers to the data update interval. for The number of newly acquired voltage values ​​by the internal voltage acquisition device; The range of values ​​is Second.

[0061] S4. Buffer the data matrix Divide into equal rows Data evaluation matrix :

[0062] In the formula: It is a data evaluation matrix. Number.

[0063] S5. Evaluate the matrix for each data point. of Each column is subjected to Fourier decomposition to extract the phase values ​​of the fundamental wave and its odd harmonics, forming a harmonic phase matrix. :

[0064] In the formula: For the first Data evaluation matrix The Middle Column data Second harmonic phase value; variable The corresponding fundamental wave and its odd harmonic orders ;parameter The value corresponding to the highest odd harmonic order is considered, which depends on the generator operating parameters.

[0065] S6. Calculate the phase matrix for each harmonic. The phase difference between two adjacent columns is calculated and normalized using the mean of the fundamental phase difference as a reference, to obtain the phase difference matrix. ;

[0066] In the formula: Phase difference matrix The Okay, number The column element values ​​represent the harmonic phase matrix. Middle elements and The normalized value of the difference; variable Values ,variable Values ; for The first of the matrix Okay, number Column element values.

[0067] S7, Phase difference matrix Convert to column vector array and calculate the array Upper and lower quartiles , and its interquartile range ,Will The interquartile ranges constitute a dynamic interquartile range array. :

[0068] In the formula: For matrix The List all element values.

[0069] S8. Calculate the dynamic percentile array. upper and lower quartiles and Obtain the upper and lower limits of the box plot. and Then, the error level of the voltage acquisition device is introduced to obtain the upper and lower limits of abnormality. and :

[0070] In the formula: This represents the upper limit of the box plot; The lower limit of the box plot; for The upper quartiles; for The lower quartile; parameter This is an adjustment factor, which depends on the error level of the voltage acquisition device; This is the upper limit of the abnormal value; This is the lower limit of the abnormal value.

[0071] S9. Traverse the dynamic percentile array If there exists an element with a value greater than the upper limit or less than the lower limit If so, it is determined that the hydro-generator unit has malfunctioned.

Claims

1. A method for online fault diagnosis of hydro-generators based on the potential distribution distance of the through-bolt screw, characterized in that: Includes the following steps: S1, at any magnetic pole of the hydro-generator A voltage acquisition device is installed on each of the through-hole screws to acquire voltage data in real time. The induced electromotive force of the through-hole screw; S2, based on the sampling rate of the voltage acquisition device Set the length of time for storing historical data. , build × dimensional historical data matrix ; for The number of voltage values ​​acquired by the voltage acquisition device within a given time period; S3, every other time At that moment, make use of that period of time × Weixin monitoring data Slide to update historical data matrix Obtain the data buffer matrix ; for The number of newly acquired voltage values ​​by the internal voltage acquisition device; S4. Buffer the data matrix Divide into equal rows Data evaluation matrix ; S5. Evaluate the matrix for each data point. of Each column is subjected to Fourier decomposition to extract the phase values ​​of the fundamental wave and its odd harmonics, forming a harmonic phase matrix. ; S6. Calculate the phase matrix for each harmonic. The phase difference between two adjacent columns is calculated and normalized using the mean of the fundamental phase difference as a reference, to obtain the phase difference matrix. ; S7, Phase difference matrix Convert to column vector array and calculate the array Upper and lower quartiles , and its interquartile range ,Will The interquartile ranges constitute a dynamic interquartile array. ; S8. Calculate the dynamic percentile array. Upper and lower quartiles and Obtain the upper and lower limits of the box plot. and Then, the error level of the voltage acquisition device is introduced to obtain the upper and lower limits of abnormality. and ; S9. Traverse the dynamic percentile array If there exists an element with a value greater than the upper limit or less than the lower limit If so, it is determined that the hydro-generator unit has malfunctioned.

2. The online fault diagnosis method for hydro-generators based on the potential distribution distance of the through-screw as described in claim 1, characterized in that: In step S2, the historical data matrix The expression is: In the formula: For the first The first core screw was monitored and acquired. Each voltage measurement value; for The number of voltage values ​​acquired by the voltage acquisition device within a given time period.

3. The online fault diagnosis method for hydro-generators based on the potential distribution distance of the through-screw as described in claim 2, characterized in that: The Second.

4. The online fault diagnosis method for hydro-generators based on the potential distribution distance of the through-bolt as described in claim 1, characterized in that: In step S3, the data buffer matrix... The expression is: In the formula: This refers to the data update interval. for The number of new voltage acquisition values ​​obtained by the internal voltage acquisition device.

5. The online fault diagnosis method for hydro-generators based on the potential distribution distance of the through-bolt as described in claim 4, characterized in that: The The range of values ​​is Second.

6. The online fault diagnosis method for hydro-generators based on the potential distribution distance of the through-bolt as described in claim 1, characterized in that: In step S4, the data evaluation matrix The expression is: In the formula: It is a data evaluation matrix. Number.

7. The online fault diagnosis method for hydro-generators based on the potential distribution distance of the through-screw as described in claim 1, characterized in that: In step S5, the harmonic phase matrix The expression is: In the formula: For the first Data evaluation matrix The Middle Column data Second harmonic phase value; variable The corresponding fundamental wave and its odd harmonic orders ;parameter The value corresponding to the highest odd harmonic order is considered, which depends on the generator operating parameters.

8. The online fault diagnosis method for hydro-generators based on the potential distribution distance of the through-bolt as described in claim 1, characterized in that: In step S6, the phase difference matrix The expression is: In the formula: Phase difference matrix The Okay, number The column element values ​​represent the harmonic phase matrix. Middle elements and The normalized value of the difference; variable Values ,variable Values ; for The first of the matrix Okay, number Column element values.

9. The online fault diagnosis method for hydro-generators based on the potential distribution distance of the through-bolt as described in claim 1, characterized in that: In step S7, the dynamic quantile array The calculation expression is: In the formula: For matrix The List all element values.

10. The online fault diagnosis method for hydro-generators based on the potential distribution distance of the through-bolt as described in claim 1, characterized in that: In step S8, the abnormal upper and lower limits are... and The calculation expression is: In the formula: This represents the upper limit of the box plot; This represents the lower limit of the box plot. for The upper quartiles; for The lower quartile; parameters This is an adjustment factor, which depends on the error level of the voltage acquisition device; This is the upper limit of the abnormal value; This is the lower limit of the abnormal value.

Images