A method and system for extracting the rotational speed of a wind turbine generator

Abstract

The application provides a method and system for extracting the rotating speed of a wind turbine motor, comprising the following steps: step 1, obtaining the original waveform data of the vibration of the driving end of the wind turbine generator; step 2, processing the obtained original waveform data by using a pre-constructed Sine function to obtain a filtered waveform; step 3, dividing the filtered waveform into a plurality of segment waveforms with a length of L at a set step length, and calculating the rotating speed and time relationship corresponding to each segment waveform; then, all the rotating speed and time relationships are sorted to obtain the time-rotating speed relationship corresponding to the original waveform data; and the application solves the problem of inaccuracy in extracting short-time rotating speed through vibration signals in the industry.

Description

Technical Field

[0001] This invention belongs to the field of wind power, specifically relating to a method and system for extracting the rotational speed of a wind turbine motor. Background Technology

[0002] In recent years, with the introduction and implementation of the national "dual-carbon" policy, the wind power industry has experienced rapid development. Vibration monitoring equipment, as a dedicated device to ensure the safe operation of wind turbines, has increasingly become standard equipment. By analyzing the data collected by vibration monitoring equipment, the operating status of the unit can be determined, enabling timely maintenance. Rotational speed is a crucial parameter in vibration analysis. Due to issues such as the quality of the rotational speed sensor itself or damage during equipment operation, rotational speed measurements are often missing. Wind turbines have numerous vibration acceleration sensors; processing the vibration acceleration signals to extract the rotational speed over a period of time is of great significance for the vibration analysis of wind turbines.

[0003] For speed extraction, the main methods at present are short-time Fourier transform (SFT) and short-time autocorrelation analysis (SCA). SFT has three drawbacks: First, aliasing occurs. When the highest frequency in the spectrum is greater than half the sampling frequency, spectral overlap occurs, affecting the analysis. Second, leakage occurs. Short-time analysis requires truncating a segment of the original waveform, which is not a full-cycle signal; therefore, leakage often occurs when calculating without a window function. Third, the picket fence effect occurs. The frequency domain reflects values ​​at discrete points; if a peak exists between discrete points, it cannot be detected. SCA also suffers from noise significantly affecting the calculation results. Summary of the Invention

[0004] The purpose of this invention is to provide a method for extracting the rotational speed of a wind turbine motor, which solves the above-mentioned shortcomings of existing methods for providing the rotational speed of wind turbine motors. This invention filters out noise by constructing a singer function and performing a convolution operation with the original waveform of the motor; and obtains the short-time rotational speed by using short-time autocorrelation analysis, thereby completing the extraction of rotational speed information from the vibration signal.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] The present invention provides a method for extracting the rotational speed of a wind turbine motor, comprising the following steps:

[0007] Step 1: Obtain the raw waveform data of the vibration at the drive end of the wind turbine generator;

[0008] Step 2: Process the obtained raw waveform data using the pre-constructed Singer function to obtain the filtered waveform;

[0009] Step 3: Divide the filtered waveform into multiple waveform segments of length L with a set step size, and calculate the speed and time relationship corresponding to each waveform segment; then organize all the speed and time relationships to obtain the time-speed relationship corresponding to the original waveform data.

[0010] Preferably, in step 2, the pre-constructed singer function is specifically:

[0011]

[0012] Where g[n] is the pre-constructed Singer function; n is the x-coordinate of the Singer function. p is the upper limit of the filter.

[0013] Preferably, in step 3, the relationship between rotational speed and time corresponding to each waveform segment is calculated, specifically by:

[0014] S31, calculate the time delay and autocorrelation coefficient of each waveform segment;

[0015] S32, based on the obtained time delay and autocorrelation coefficient diagram, obtain the speed and time relationship corresponding to the waveform segment.

[0016] Preferably, in S31, the time delay and autocorrelation coefficient of each waveform segment are calculated, specifically by:

[0017] Construct an autocorrelation function, and calculate the time delay and autocorrelation coefficient of the waveform segment based on the obtained autocorrelation function.

[0018] Preferably, the constructed autocorrelation function expression is as follows:

[0019]

[0020] Among them, R x,x is the autocorrelation coefficient; s[n] is the waveform segment with length L; n indicates the nth element.

[0021] Preferably, in S32, the rotational speed versus time relationship corresponding to the waveform segment is obtained based on the obtained time delay and autocorrelation coefficient diagram. Specifically, the method is as follows:

[0022] First, set the time delay range value according to the rated speed and cut-in speed of the wind turbine generator;

[0023] Then, extract the segment corresponding to the set time delay range from the obtained time delay and autocorrelation coefficient graph; select the time delay with the autocorrelation coefficient close to 1 and the smallest value from the obtained segment graph;

[0024] Next, the rotational speed corresponding to this segment is calculated based on the selected minimum delay value;

[0025] Finally, the midpoint of this time segment is defined as the time corresponding to the rotational speed of that segment, thus obtaining the time-rotational speed relationship of that segment.

[0026] A system for extracting the rotational speed of a wind turbine motor, comprising:

[0027] The data acquisition unit is used to acquire the original waveform data of the vibration at the drive end of the wind turbine generator;

[0028] The filtering unit is used to process the obtained raw waveform data using a pre-constructed singer function to obtain the filtered waveform.

[0029] The calculation unit divides the filtered waveform into multiple waveform segments of length L with a set step size, calculates the rotational speed and time relationship corresponding to each waveform segment, and then organizes all the rotational speed and time relationships to obtain the time-rotational speed relationship corresponding to the original waveform data.

[0030] Compared with the prior art, the beneficial effects of the present invention are:

[0031] This invention provides a method for extracting the rotational speed of a wind turbine motor. It uses a pre-constructed Singer function as a filter and performs filtering by time-domain convolution to avoid aliasing, leakage, and picket fence phenomena caused by short-time Fourier transforms. Then, it combines short-time autocorrelation analysis to filter out the influence of noise, which greatly improves the accuracy of rotational speed extraction. This invention makes up for the problem of inaccurate extraction of short-time rotational speed from vibration signals in the industry. Attached Figure Description

[0032] Figure 1 This is a flowchart illustrating the present invention;

[0033] Figure 2 This is the original waveform of the generator drive end vibration;

[0034] Figure 3 It is the Singer function;

[0035] Figure 4 This is the waveform after filtering;

[0036] Figure 5 This is the first segment delay-autocorrelation coefficient plot;

[0037] Figure 6 This is the first time-speed relationship graph. Detailed Implementation

[0038] The present invention will now be described in further detail with reference to the accompanying drawings.

[0039] like Figure 1 As shown, the present invention provides a method for extracting the rotational speed of a wind turbine motor, comprising the following steps:

[0040] Step 1: Based on the existing CMS monitoring system of the wind turbine, acquire the original waveform data f[n] of the vibration at the drive end of the wind turbine generator, such as... Figure 2 As shown;

[0041] Step 2, based on the generator's rated speed (rpm) 额定 Construct the singer function according to the following formula, such as Figure 3 As shown:

[0042]

[0043] Where g[n] is the preconstructed Singer function; n is the x-coordinate of the Singer function, and its relationship with the original generator waveform time T is as follows: p is the upper limit of the filter, which is greater than the generator's rated speed (rpm). 额定 / 60), in this embodiment, p is set to rpm. 额定 / 60+1.

[0044] Step 3: Use the Singer function from Step 2 to convolve the original waveform from Step 1 to obtain the filtered waveform, as shown below. Figure 4 As shown.

[0045] Using the Singer function as a window, convolving the original waveform is essentially performing low-pass filtering at the time-domain waveform level. The convolution calculation is based on the following formula:

[0046]

[0047] Where f[n] is the original waveform with length N; g[n] is the constructed singer function; n indicates the nth element.

[0048] Step 4: Set the segment length to L and obtain the initial segment (i,j) of the filtered waveform, where i represents the i-th element of the waveform, which is initially 0; j represents the j-th element of the waveform, which is initially L.

[0049] Step 5: Construct the autocorrelation function, and calculate the time delay and autocorrelation coefficient of the waveform segment based on the obtained autocorrelation function:

[0050]

[0051] Among them, R x,x is the autocorrelation coefficient; s[n] is the waveform segment with length L; n indicates the nth element.

[0052] Step 6: Based on the time delay and autocorrelation coefficient graph calculated in Step 5, calculate the rotational speed corresponding to this segment and correlate it with time. Specifically:

[0053] First, based on the rated speed and cut-in speed of the wind turbine generator, the time delay range (a, b) is set, where a is calculated from the rated speed and b is calculated from the cut-in speed.

[0054] Then, in step 5, a segment corresponding to the set time delay range is extracted from the obtained time delay and autocorrelation coefficient graph; from the obtained segment graph, the time delay lag with the autocorrelation coefficient close to 1 and the smallest value is selected, such as... Figure 5 As shown;

[0055] Next, the rotational speed corresponding to this segment is calculated based on the selected minimum delay value:

[0056] rpm=f s / lag*60

[0057] Where rpm is the rotational speed corresponding to this segment; f s 1 is the sampling frequency of the original waveform; lag is the time delay.

[0058] Finally, the midpoint of this time segment is defined as the time corresponding to the rotational speed of that segment, thus obtaining the time-rotational speed relationship for that segment, as follows: Figure 6 As shown.

[0059] Step 7: Obtain the next segment according to the set step size. The step size is set to half the segment length, i.e., L / 2, so the next segment should be the segment with a length of (i+L / 2, j+L / 2).

[0060] Step 8: Determine whether the next segment exceeds the length of the original waveform. If it does not exceed the length, continue to calculate the segment time-speed relationship; if it exceeds the length, stop the calculation and output the time-speed relationship of the overall waveform.

[0061] This invention utilizes a window constructed using the Singer function as a low-pass filter, and performs convolution calculation with the Singer function to perform low-pass filtering. The rotational speed of wind turbines is generally in the low-frequency range, with most doubly-fed wind turbines operating below 30Hz. Low-pass filtering can remove high-frequency noise. Simultaneously, the autocorrelation function is used, which is the cross-correlation between a signal and itself at different time points, providing a mathematical tool to find the fundamental frequency implicit in the signal. After low-pass filtering the original signal, only low-frequency signals remain, with the generator rotational speed being the primary signal. Based on autocorrelation analysis, this rotational speed can be calculated relatively easily.

[0062] This invention provides a system for extracting the rotational speed of a wind turbine motor, comprising:

[0063] The data acquisition unit is used to acquire the original waveform data of the vibration at the drive end of the wind turbine generator;

[0064] The filtering unit is used to process the obtained raw waveform data using a pre-constructed singer function to obtain the filtered waveform.

[0065] The calculation unit is used to divide the filtered waveform into multiple waveform segments of length L with a set step size, calculate the speed and time relationship corresponding to each waveform segment, and then organize all the speed and time relationships to obtain the time-speed relationship corresponding to the original waveform data.

Claims

1. A method for extracting the rotational speed of a wind turbine motor, characterized in that, Includes the following steps: Step 1: Obtain the raw waveform data of the vibration at the drive end of the wind turbine generator; Step 2: Process the obtained raw waveform data using the pre-constructed Singer function to obtain the filtered waveform; Step 3: Divide the filtered waveform into multiple waveform segments of length L with a set step size, and calculate the relationship between rotational speed and time for each waveform segment. Then, all the speed and time relationships are sorted out to obtain the time-speed relationship corresponding to the original waveform data; In step 3, the relationship between rotational speed and time corresponding to each waveform segment is calculated. The specific method is as follows: S31, calculate the time delay and autocorrelation coefficient of each waveform segment; S32, Based on the obtained time delay and autocorrelation coefficient diagram, obtain the relationship between rotational speed and time corresponding to the waveform segment; In S32, based on the obtained time delay and autocorrelation coefficient diagrams, the relationship between rotational speed and time corresponding to the waveform segment is obtained. The specific method is as follows: First, set the time delay range value according to the rated speed and cut-in speed of the wind turbine generator; Then, extract the segment corresponding to the set time delay range from the obtained time delay and autocorrelation coefficient graph; select the time delay with the autocorrelation coefficient close to 1 and the smallest value from the obtained segment graph; Next, the rotational speed corresponding to this segment is calculated based on the selected minimum delay value; Finally, the midpoint of this time segment is defined as the time corresponding to the rotational speed of that segment, thus obtaining the time-rotational speed relationship of that segment.

2. The method for extracting the speed of a wind turbine motor according to claim 1, characterized in that, In step 2, the pre-constructed singer function is specifically: in, For pre-constructed singer functions; The x-coordinate of the Singer function. , The original waveform time of the generator; This is the upper limit of the filter.

3. The method for extracting the speed of a wind turbine motor according to claim 1, characterized in that, In S31, the time delay and autocorrelation coefficient of each waveform segment are calculated. The specific method is as follows: Construct an autocorrelation function, and calculate the time delay and autocorrelation coefficient of the waveform segment based on the obtained autocorrelation function.

4. The method for extracting the speed of a wind turbine motor according to claim 3, characterized in that, The constructed autocorrelation function is expressed as follows: in, This is the autocorrelation coefficient; This is a waveform segment with a length of [length missing]. ; Indicates the nth element.

5. A system for extracting the rotational speed of a wind turbine motor, characterized in that, A method for extracting the speed of a wind turbine motor according to claim 1 includes: The data acquisition unit is used to acquire the original waveform data of the vibration at the drive end of the wind turbine generator; The filtering unit is used to process the obtained raw waveform data using a pre-constructed singer function to obtain the filtered waveform. The calculation unit is used to divide the filtered waveform into multiple waveform segments of length L with a set step size, calculate the speed and time relationship corresponding to each waveform segment, and then organize all the speed and time relationships to obtain the time-speed relationship corresponding to the original waveform data.

Images