Wind turbine generator vibration frequency analysis method, device, equipment and medium
By employing thread pool technology and target spectrum analysis methods, the problem of insufficient identification of low-frequency components in wind turbine generators is solved, enabling efficient low-frequency vibration feature extraction and multi-dimensional analysis, and supporting health status assessment and fault diagnosis of wind turbine generators.
Patent Information
- Application Number
- CN202511775210.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-28
- Publication Date
- 2026-02-13
AI Technical Summary
In existing technologies, vibration testing of wind turbine generators cannot accurately identify frequency components in the low-frequency band, resulting in insufficient attention to the low-frequency band and low extraction accuracy.
The measured vibration data of the wind turbine was acquired in parallel using thread pool technology. After data preprocessing, the target low-frequency band was extracted using the target spectrum analysis method, divided into multiple sub-bands, and the target main frequency and energy characteristics of each sub-band were identified to generate a frequency analysis file.
It achieves efficient capture of key low-frequency vibration characteristics of wind turbine generators, solves the problems of insufficient attention to low-frequency bands and low extraction accuracy, and supports multi-dimensional analysis and fault diagnosis.
Smart Images
Figure CN121520139A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of wind turbine technology, and in particular to a method, apparatus, equipment and medium for analyzing the vibration frequency of wind turbine generators. Background Technology
[0002] Vibration monitoring is a crucial step in ensuring the safe and stable operation of wind power equipment during operation.
[0003] In existing technologies, vibration testing of wind turbine generators mainly relies on a combination of specialized data acquisition equipment and general-purpose data analysis software (such as MATLAB and LabVIEW). This method suffers from the inability to accurately identify frequency components in the low-frequency range. Therefore, improving the accuracy of low-frequency component identification has become a pressing technical problem to be solved. Summary of the Invention
[0004] In view of this, the purpose of this invention is to provide a method, apparatus, equipment, and medium for analyzing the vibration frequency of wind turbine generators. This method, by extracting the target low-frequency band and employing a target spectrum analysis method, solves the problems of insufficient focus on the low-frequency band and low extraction accuracy in existing technologies. The specific solution is as follows:
[0005] Firstly, this application provides a method for analyzing the vibration frequency of a wind turbine generator set, including:
[0006] Based on thread pool technology, several measured vibration data of the target wind turbine are obtained in parallel from the target data source, and the measured vibration data are preprocessed to obtain the corresponding target vibration data.
[0007] The power spectral density corresponding to the target vibration data is determined using the target spectrum analysis method, and the target low-frequency band corresponding to the target vibration data is extracted from the power spectral density; wherein, the target low-frequency band represents the frequency band where the vibration frequency of the wind turbine is less than a preset frequency threshold;
[0008] Select a target number of sub-frequency bands from the target low-frequency band, and determine the target main frequency and the target energy characteristics corresponding to each sub-frequency band; wherein, the target main frequency is the characteristic frequency characterizing the vibration component of the corresponding sub-frequency band;
[0009] Based on the target main frequencies and the target energy characteristics, a frequency analysis file corresponding to the target wind turbine is generated to analyze the vibration frequency of the target wind turbine.
[0010] Optionally, before acquiring several measured vibration data of the target wind turbine in parallel from the target data source based on thread pool technology, the method further includes:
[0011] Create a target spectrum cache dictionary; wherein the target spectrum cache dictionary is used to store data in the form of key-value pairs, the key of the target spectrum cache dictionary is the hash value corresponding to the vibration data, and the value of the target spectrum cache dictionary is the power spectral density corresponding to the vibration data.
[0012] Optionally, before acquiring several measured vibration data of the target wind turbine in parallel from the target data source based on thread pool technology, the method further includes...
[0013] Determine whether the target vibration data and the power spectral density exist in the target spectrum cache dictionary. If the target vibration data and the power spectral density exist in the target spectrum cache dictionary, then directly obtain the target vibration data and the power spectral density from the target spectrum cache dictionary.
[0014] If the power spectral density is not present in the target spectrum cache dictionary, then proceed to the step of obtaining several measured vibration data of the target wind turbine in parallel from the target data source based on thread pool technology.
[0015] Optionally, the step of preprocessing the measured vibration data to obtain the corresponding target vibration data includes:
[0016] The measured vibration data is validated to select vibration data with a data loss rate less than a preset loss rate threshold.
[0017] The filtered vibration data is subjected to outlier processing and missing value imputation to obtain the corresponding processed data, and the processed data is resampled based on the target frequency to obtain the target vibration data.
[0018] Optionally, determining the power spectral density corresponding to the target vibration data using the target spectrum analysis method includes:
[0019] The target parameters corresponding to the target spectrum analysis method are dynamically set according to the original sampling frequency corresponding to the measured vibration data and the target low-frequency band, and the power spectral density corresponding to the target vibration data is determined by using the target parameters and the target spectrum analysis method; wherein, the target parameters include the time window length and the time window function.
[0020] Optionally, determining the target primary frequency corresponding to each of the sub-frequency bands and the target energy characteristics corresponding to each of the target primary frequencies includes:
[0021] Determine the initial master frequency corresponding to each of the sub-frequency bands, and determine whether the initial energy characteristic corresponding to each initial master frequency is less than a preset energy threshold. If any initial energy characteristic is not less than the preset energy threshold, determine whether the index of the initial master frequency corresponding to any initial energy characteristic is within a preset index range. If the index of the initial master frequency corresponding to any initial energy characteristic is within the preset index range, determine any initial energy characteristic as the target energy characteristic of the corresponding sub-frequency band, and determine the initial master frequency corresponding to any initial energy characteristic as the target master frequency of the corresponding sub-frequency band.
[0022] Optionally, the wind turbine vibration frequency analysis method further includes:
[0023] The data processing progress and frequency analysis file corresponding to the target wind turbine are displayed using a preset visual user interface.
[0024] Secondly, this application provides a wind turbine vibration frequency analysis device, comprising:
[0025] The data preprocessing module is used to acquire several measured vibration data of the target wind turbine in parallel from the target data source based on thread pool technology, and to perform data preprocessing on the measured vibration data to obtain the corresponding target vibration data.
[0026] The frequency band extraction module is used to determine the power spectral density corresponding to the target vibration data using a target spectrum analysis method, and to extract the target low-frequency band corresponding to the target vibration data from the power spectral density; wherein, the target low-frequency band represents the frequency band where the vibration frequency of the wind turbine is less than a preset frequency threshold;
[0027] An energy characteristic determination module is used to select a target number of sub-frequency bands from the target low-frequency band, and determine the target main frequency and the target energy characteristic corresponding to each sub-frequency band respectively; wherein, the target main frequency is the characteristic frequency characterizing the vibration component of the corresponding sub-frequency band;
[0028] The vibration frequency analysis module is used to generate a frequency analysis file corresponding to the target wind turbine based on each of the target main frequencies and each of the target energy characteristics, so as to analyze the vibration frequency of the target wind turbine.
[0029] Thirdly, this application provides an electronic device, comprising:
[0030] Memory, used to store computer programs;
[0031] A processor is used to execute the computer program to implement the aforementioned wind turbine vibration frequency analysis method.
[0032] Fourthly, this application provides a computer-readable storage medium for storing a computer program that, when executed by a processor, implements the aforementioned wind turbine vibration frequency analysis method.
[0033] This application first uses thread pool technology to acquire several measured vibration data of the target wind turbine in parallel from the target data source, and performs data preprocessing on the measured vibration data to obtain the corresponding target vibration data. Then, it uses a target spectrum analysis method to determine the power spectral density corresponding to the target vibration data, and extracts the target low-frequency band corresponding to the target vibration data from the power spectral density. The target low-frequency band represents the frequency band where the vibration frequency of the wind turbine is less than a preset frequency threshold. Then, a target number of sub-frequency bands are selected from the target low-frequency band, and the target main frequency and target energy characteristics corresponding to each sub-frequency band are determined respectively. The target main frequency is the characteristic frequency representing the vibration component of the corresponding sub-frequency band. Finally, a frequency analysis file corresponding to the target wind turbine is generated based on each target main frequency and each target energy characteristic to analyze the vibration frequency of the target wind turbine. Therefore, this application achieves efficient processing of large-scale data by using thread pool technology to acquire and preprocess measured vibration data in parallel, thus solving the problem of low data processing efficiency in existing technologies. By extracting the target low-frequency band and using the target spectrum analysis method, it achieves targeted capture of key low-frequency vibration characteristics of wind turbine generators, solving the problems of insufficient attention to the 0-1Hz low-frequency band and low extraction accuracy in existing technologies. By dividing the target low-frequency band into multiple sub-bands and identifying their target main frequency and target energy characteristics respectively, it achieves multi-dimensional analysis of vibration signals, solving the problem that existing technologies cannot effectively separate the frequency components of different components. Attached Figure Description
[0034] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.
[0035] Figure 1 This is a flowchart of a vibration frequency analysis method for wind turbine generators disclosed in this application;
[0036] Figure 2 This is a schematic diagram of the structure of a wind turbine vibration frequency analysis device disclosed in this application;
[0037] Figure 3This is a structural diagram of an electronic device disclosed in this application. Detailed Implementation
[0038] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0039] Current methods for analyzing the vibration of wind turbines suffer from the problem of accurately identifying frequency components in the low-frequency band. To address this, this application provides a method for analyzing the vibration frequency of wind turbines. By extracting the target low-frequency band and employing a target spectrum analysis method, this method solves the problems of insufficient attention to the low-frequency band and low extraction accuracy in existing technologies.
[0040] See Figure 1 As shown in the figure, an embodiment of the present invention discloses a method for analyzing the vibration frequency of a wind turbine, including:
[0041] Step S11: Based on thread pool technology, obtain several measured vibration data of the target wind turbine in parallel from the target data source, and perform data preprocessing on the measured vibration data to obtain the corresponding target vibration data.
[0042] In this embodiment, before acquiring several measured vibration data of the target wind turbine in parallel from the target data source based on thread pool technology, the method further includes: creating a target spectrum cache dictionary; wherein, the target spectrum cache dictionary is used to store data in the form of key-value pairs, the key of the target spectrum cache dictionary is the hash value corresponding to the vibration data, and the value of the target spectrum cache dictionary is the power spectral density corresponding to the vibration data.
[0043] That is, create a spectrum cache dictionary spectra_cache, with the key being a tuple (sig_adjusted) of the preprocessed signal and the value being (frequency array f, power spectral density array pxx).
[0044] In addition, in this embodiment, before acquiring several measured vibration data of the target wind turbine in parallel from the target data source based on thread pool technology, the method further includes: determining whether the target vibration data and the power spectral density exist in the target spectrum cache dictionary; if the target vibration data and the power spectral density exist in the target spectrum cache dictionary, the target vibration data and the power spectral density are directly acquired from the target spectrum cache dictionary; if the power spectral density does not exist in the target spectrum cache dictionary, the method jumps to the step of acquiring several measured vibration data of the target wind turbine in parallel from the target data source based on thread pool technology.
[0045] The above process is as follows: first, check if the current signal already exists in the cache to avoid duplicate calculations; if the tuple(sig_adjusted) is in the key of spectra_cache, directly read the cached f (full-band frequency array) and pxx (full-band power spectral density array); if not found, execute the Welch power spectrum calculation step.
[0046] This implementation uses thread pool technology to read multiple data files in parallel, supporting text files with different delimiters (commas, semicolons, spaces, tabs, etc.).
[0047] In this embodiment, before processing the vibration data, the following parameters need to be preliminarily determined: the original sampling frequency fs (Hz, e.g., 62.5Hz), the target analysis frequency band (0-1Hz); the dual-interval division parameters: the lower bound of the low-frequency sub-interval freq_low (e.g., 0.0Hz), the midpoint of the interval freq_mid (e.g., 0.2Hz), and the upper bound of the high-frequency sub-interval freq_high (e.g., 0.3Hz); and the energy threshold energy_thresh (e.g., ...). ), noise filtering threshold noise_thresh (e.g.) ); Welch method parameters: window length nperseg (e.g., 2048 points), overlap_ratio (e.g., 50%), window type (Hanning window).
[0048] Furthermore, in this embodiment, the process of preprocessing the measured vibration data to obtain the corresponding target vibration data may specifically include: performing data verification on the measured vibration data to filter out the filtered vibration data with a data loss rate less than a preset loss rate threshold; performing outlier processing and missing value imputation on the filtered vibration data to obtain the corresponding processed data; and resampling the processed data based on the target frequency to obtain the target vibration data.
[0049] Specifically, data preprocessing includes data verification, outlier handling, and missing value imputation, filtering valid data based on a set loss rate threshold; resampling involves resampling the original signal according to the target frequency to unify the data format. Through data filtering and resampling, data usability is ensured.
[0050] In other words, the preprocessed vibration acceleration signal in this embodiment (such as the data after X / Y / Z direction correction, denoted as sig_adjusted) must satisfy the following conditions: no NaN values and no extreme outliers (absolute value not greater than 1). The data length should be no less than 1024 points.
[0051] It should be noted that this embodiment can replace the multi-threaded processing mechanism with a distributed computing framework (such as Apache Spark), which is suitable for analysis scenarios of ultra-large-scale data (tens of thousands of files) and can further improve processing efficiency.
[0052] By employing thread pool technology, tasks such as data reading, preprocessing, and spectrum calculation are distributed to multiple threads for parallel processing, improving the efficiency of processing large-scale data. The degree of parallelism can be adjusted based on hardware performance by setting the maximum number of threads.
[0053] Step S12: Determine the power spectral density corresponding to the target vibration data using the target spectrum analysis method, and extract the target low-frequency band corresponding to the target vibration data from the power spectral density; wherein, the target low-frequency band represents the frequency band where the vibration frequency of the wind turbine is less than a preset frequency threshold.
[0054] This embodiment requires the Welch method (i.e., target spectrum analysis method) to calculate the power spectral density of the vibration signal, improve the spectrum estimation accuracy through sliding window and overlap processing, and perform special processing on the 0-1Hz frequency band (i.e. target low frequency band) to improve the extraction accuracy of frequency features in this band.
[0055] In this embodiment, the process of determining the power spectral density corresponding to the target vibration data using the target spectrum analysis method may specifically include: dynamically setting the target parameters corresponding to the target spectrum analysis method based on the original sampling frequency and target low-frequency band corresponding to the measured vibration data, and determining the power spectral density corresponding to the target vibration data using the target parameters and the target spectrum analysis method; wherein, the target parameters include the time window length and the time window function.
[0056] Specifically, the calculation parameters are dynamically adjusted based on the original sampling frequency fs and the target frequency band (0-1Hz):
[0057] Window length nperseg: take min(2048, len(sig_adjusted)) to ensure that the window covers enough data points to guarantee low-frequency resolution (e.g. when fs=62.5Hz, nperseg=2048 corresponds to a frequency resolution of approximately 0.03Hz).
[0058] Overlap count: Calculated with overlap_ratio=50%, i.e., overlap = nperseg / / 2, to reduce the variance of the spectral estimation;
[0059] Window function: The Hanning window (window='hann') is used, whose side lobe attenuation (-31dB) is much higher than that of the rectangular window, suppressing the interference of high-frequency side lobes on the low-frequency band.
[0060] Power spectral density calculation:
[0061] The scipy.signal.welch function is called to perform the calculation:
[0062] Python execution:
[0063] import signal from scipy;
[0064] f, pxx = signal.welch(
[0065] x = sig_adjusted;
[0066] fs=fs,
[0067] window='hann';
[0068] nperseg=nperseg;
[0069] noverlap = noverlap;
[0070] scaling='density'# Output power spectral density (unit: m² / s) 4 / Hz);
[0071] After the calculation is completed, (f, pxx) is stored in spectra_cache with the key tuple(sig_adjusted) for later reuse.
[0072] Low-frequency data extraction:
[0073] Precisely filter data in the 0-1Hz frequency band using masking technology to eliminate high-frequency interference:
[0074] Python execution:
[0075] low_freq_mask = (f >= 0.0) & (f <= 1.0) # Generate a 0-1Hz frequency band mask;
[0076] f_low = f[low_freq_mask] # Frequency array for the 0-1Hz band;
[0077] pxx_low = pxx[low_freq_mask] # Power spectral density array for the 0-1Hz frequency band;
[0078] If len(f_low) == 0, it is determined that there is no valid low-frequency data for the current signal, the anomaly is recorded, and the spectrum analysis process for the signal is terminated.
[0079] Noise filtering:
[0080] Low-frequency noise is eliminated through multi-dimensional filtering, while retaining the effective signal.
[0081] 1. Power threshold filtering: Remove data in pxx_low that is less than noise_thresh (e.g., 1e-12), and mark the corresponding frequency points as invalid.
[0082] Python execution:
[0083] valid_pxx_mask = pxx_low >= noise_thresh;
[0084] f_valid = f_low[valid_pxx_mask];
[0085] pxx_valid = pxx_low[valid_pxx_mask];
[0086] 2. Outlier removal: using 3... The criteria filter out extreme noise points. The mean (pxx_mean) and standard deviation (pxx_std) of pxx_valid are calculated, and data exceeding the range [pxx_mean - 3*pxx_std, pxx_mean + 3*pxx_std] are removed.
[0087] Python execution:
[0088] import numpy as np;
[0089] pxx_mean = np.mean(pxx_valid);
[0090] pxx_std = np.std(pxx_valid);
[0091] outlier_mask = (pxx_valid >= pxx_mean - 3*pxx_std) & (pxx_valid <=pxx_mean + 3*pxx_std);
[0092] f_clean = f_valid[outlier_mask];
[0093] pxx_clean = pxx_valid[outlier_mask];
[0094] If len(f_clean) < 10 (too few valid data points), the low-frequency signal quality is deemed insufficient, a warning is recorded, and subsequent main frequency identification is terminated.
[0095] Spectral smoothing (optional):
[0096] To address high-frequency noise residue, a moving average is used for smoothing (window length is 5-10 points), balancing smoothness and feature preservation.
[0097] Python execution:
[0098] window_size = 5;
[0099] pxx_smoothed = np.convolve(pxx_clean, np.ones(window_size) / window_size, mode='same');
[0100] The final optimized low-frequency data are: f_clean (0-1Hz clean frequency array) and pxx_smoothed (smoothed power spectral density array).
[0101] Step S13: Select a target number of sub-frequency bands from the target low-frequency band, and determine the target main frequency and the target energy characteristics corresponding to each sub-frequency band; wherein, the target main frequency is the characteristic frequency characterizing the vibration component of the corresponding sub-frequency band.
[0102] In this embodiment, the main frequency is identified and its corresponding energy value is calculated within the preset low-frequency range (freq_low-freq_mid) and high-frequency range (freq_mid-freq_high), i.e., sub-frequency bands. Combining the design frequency of the analyzed component and the wind turbine rotor frequency range, effective separation of the component frequency from the rotor frequency and its harmonics can be achieved by setting high and low frequency ranges. Simultaneously, based on the frequency energy values corresponding to each range, the main frequency and its energy characteristics within the range can be accurately identified, providing data support for component frequency analysis. A complete spectrum analysis is performed on the test time-series data, generating statistical data tables and corresponding spectrum diagrams. By comparing the spectrum diagrams of different time periods, the trend of the main frequency changing over time can be intuitively analyzed, thereby achieving accurate location of abnormal moments. This analysis result provides reliable data support for the health status assessment and fault diagnosis of the wind turbine. Subsequently, spectrum caching: the calculated spectrum data is cached to avoid redundant calculations and improve processing efficiency.
[0103] Specifically, the 0-1Hz frequency band is divided into two sub-intervals based on preset parameters, and the dominant frequency is identified for each sub-interval (designed for different low-frequency vibration sources in wind power generation equipment):
[0104] Low-frequency sub-range: freq_low_range = (f_clean >= freq_low) & (f_clean <= freq_mid) (e.g., 0.0-0.2Hz, corresponding to equipment base vibration);
[0105] High-frequency sub-range: freq_high_range = (f_clean >= freq_mid) & (f_clean <= freq_high) (e.g., 0.2-0.3Hz, corresponding to the vibration of the transmission system).
[0106] Preliminary determination of main frequency:
[0107] Find the frequencies corresponding to the maximum power spectral density in each of the two sub-intervals, and use them as candidate dominant frequencies:
[0108] Candidate dominant frequencies for low-frequency sub-intervals:
[0109] Python execution:
[0110] if np.any(freq_low_range): # Find the index of the maximum power value within the subrange;
[0111] low_max_idx = np.argmax(pxx_smoothed[freq_low_range]) # Corresponds to the frequency and power value;
[0112] candidate_freq_low = f_clean[freq_low_range][low_max_idx];
[0113] candidate_energy_low = pxx_smoothed[freq_low_range][low_max_idx];
[0114] else:
[0115] candidate_freq_low = 0.0;
[0116] candidate_energy_low = 0.0;
[0117] Candidate dominant frequencies for high-frequency sub-intervals:
[0118] Python execution:
[0119] if np.any(freq_high_range):
[0120] high_max_idx = np.argmax(pxx_smoothed[freq_high_range]);
[0121] candidate_freq_high = f_clean[freq_high_range][high_max_idx];
[0122] candidate_energy_high= pxx_smoothed[freq_high_range][high_max_idx];
[0123] else:
[0124] candidate_freq_high = 0.0;
[0125] candidate_energy_high = 0.0;
[0126] The process of acquiring energy characteristics described above is handled by the statistical analysis module. The statistical analysis module performs the following: Time-domain feature calculation: calculates the maximum, minimum, average, and standard deviation of the vibration signal; Frequency feature statistics: performs statistical analysis on the main frequencies and corresponding energies in each direction, including the calculation of average, maximum, minimum, and standard deviation; TopN analysis: sorts the data files according to the statistical parameters and extracts the top N extreme value files for focused analysis.
[0127] The process of generating the above statistical charts is handled by the visualization module. The visualization module includes: time-domain plotting: generating time-domain waveforms of the vibration signal and marking key feature points; spectrum plotting: plotting the power spectral density of the 0-1Hz frequency band and marking the main frequency position and energy value; statistical chart generation: generating histograms, scatter plots, and average comparison charts of the main frequency and energy; and chart optimization: supporting logarithmic / linear scale switching, automatically adjusting the coordinate axis range, and optimizing label display.
[0128] It should be noted that this embodiment prioritizes a logarithmic scale for spectrogram plotting, providing a clearer view of energy differences. When the logarithmic scale is unsuitable, it automatically switches to a linear scale. Furthermore, it can automatically adjust the coordinate axis range to ensure key features are clearly visible and dynamically calculate the position of the main frequency labels, avoiding deviations from the coordinate axis range or overlaps. This embodiment can also incorporate a deep learning model (such as a convolutional neural network) into the frequency analysis module. By training the model, it can automatically identify key frequency features, making it suitable for feature extraction in complex vibration environments. However, it requires a large amount of labeled data for model training.
[0129] Furthermore, the process of saving the analysis results is handled by the data storage module. This module can save the analysis results by: saving statistical data as Excel files, supporting multi-worksheet categorization; saving charts as images; and logging errors that occur during data processing and analysis for easy troubleshooting. In addition, this embodiment can replace Excel file storage with database storage (such as MySQL or MongoDB), making it suitable for scenarios requiring long-term storage and multi-user sharing of analysis results, thus improving data query and management efficiency.
[0130] Accordingly, the process of determining the target main frequency corresponding to each of the sub-frequency bands and the target energy characteristics corresponding to each of the target main frequencies may specifically include: determining the initial main frequency corresponding to each sub-frequency band, and determining whether the initial energy characteristics corresponding to each initial main frequency are less than a preset energy threshold; if any initial energy characteristic is not less than the preset energy threshold, then determining whether the index of the initial main frequency corresponding to any initial energy characteristic is within a preset index range; if the index of the initial main frequency corresponding to any initial energy characteristic is within the preset index range, then determining any initial energy characteristic as the target energy characteristic of the corresponding sub-frequency band, and determining the initial main frequency corresponding to any initial energy characteristic as the target main frequency of the corresponding sub-frequency band.
[0131] Specifically, main frequency verification and correction:
[0132] The validity of the candidate master frequency is ensured through a two-step verification process to avoid false positives due to noise:
[0133] 1. Energy Threshold Verification: A candidate dominant frequency is considered a valid dominant frequency only if its energy (power spectral density value) is not less than energy_thresh (i.e., the preset energy threshold); otherwise, it is marked as invalid (set to 0.0).
[0134] Python execution:
[0135] if candidate_energy_low >= energy_thresh: # Low-frequency sub-interval verification;
[0136] valid_freq_low = candidate_freq_low;
[0137] valid_energy_low = candidate_energy_low;
[0138] else:
[0139] valid_freq_low = 0.0;
[0140] valid_energy_low = 0.0;
[0141] if candidate_energy_high >= energy_thresh: # High-frequency sub-interval verification;
[0142] valid_freq_high = candidate_freq_high;
[0143] valid_energy_high = candidate_energy_high;
[0144] else:
[0145] valid_freq_high = 0.0;
[0146] valid_energy_high = 0.0;
[0147] 2. Index validity correction: If the index corresponding to the candidate primary frequency exceeds the range of f_clean (extreme case), it is corrected using the nearest neighbor method:
[0148] Python execution:
[0149] if valid_freq_low != 0.0: # Example of low-frequency master frequency correction;
[0150] `correct_idx = np.argmin(np.abs(f_clean - valid_freq_low));` # Find the valid index closest to the candidate frequency;
[0151] valid_freq_low = f_clean[correct_idx];
[0152] valid_energy_low = pxx_smoothed[correct_idx];
[0153] Output results:
[0154] The verified effective main frequencies and corresponding information are stored in a FrequencyResult class instance for subsequent statistical analysis and visualization.
[0155] Python execution:
[0156] class FrequencyResult:
[0157] def __init__(self): # Low-frequency sub-interval results;
[0158] self.main_freq_low = 0.0;
[0159] self.main_energy_low = 0.0 # High-frequency sub-interval results;
[0160] self.main_freq_high = 0.0;
[0161] self.main_energy_high = 0.0;
[0162] freq_result = FrequencyResult() # Assign the result;
[0163] freq_result.main_freq_low = round(valid_freq_low, 6) # Retain 6 decimal places to ensure precision;
[0164] freq_result.main_energy_low = round(valid_energy_low, 6);
[0165] freq_result.main_freq_high = round(valid_freq_high, 6);
[0166] freq_result.main_energy_high = round(valid_energy_high, 6);
[0167] By specifically optimizing for the 0-1Hz low-frequency band, it can capture low-frequency vibration characteristics that are easily overlooked by traditional methods. Combining the design frequency of the analyzed components with the wind turbine rotor frequency range, effective separation of component frequencies from the rotor frequency and its harmonics can be achieved by setting high and low frequency intervals. Simultaneously, based on the frequency energy values corresponding to each interval, the main frequencies and their energy characteristics within each interval can be accurately identified, providing data support for component frequency analysis. A complete spectral analysis is performed on the test time-series data, generating statistical data tables and corresponding spectrum diagrams. By comparing the spectrum diagrams of different time periods, the trend of the main frequency changing over time can be intuitively analyzed, thereby achieving precise location of abnormal moments. The analysis results can provide reliable data support for the health status assessment and fault cause diagnosis of wind turbines.
[0168] In addition, this embodiment also has a complete error handling mechanism that can identify and record abnormal data, ensuring the stability of the system when processing real data containing noise or abnormalities.
[0169] Step S14: Generate a frequency analysis file corresponding to the target wind turbine based on each of the target main frequencies and each of the target energy characteristics, so as to analyze the vibration frequency of the target wind turbine.
[0170] The wind turbine vibration frequency analysis method in this embodiment further includes: displaying the data processing progress and frequency analysis file corresponding to the target wind turbine using a preset visual user interface.
[0171] Specifically, this embodiment requires result recording: associating freq_result with the corresponding signal file information (such as file name and direction) and storing it in the global statistics set; exception log: recording detailed logs (including file name, error type, and data length) for cases with no effective low-frequency data or insufficient signal quality, to facilitate subsequent troubleshooting; cache update: if a new power spectrum calculation is performed in the process, ensuring that the result has been stored in spectra_cache for subsequent reuse of the same signal.
[0172] The steps described above for displaying the data processing progress and frequency analysis files are handled by the user interaction module, which has the following functions: Parameter setting interface: Provides an intuitive parameter setting interface, including sampling frequency, data duration, frequency range, etc.; Progress display: Displays the data processing and chart generation progress in real time; Result display: Provides a log and error message display area to help users understand the processing status; Result navigation: Provides a result directory navigation function for easy viewing of analysis results.
[0173] In summary, the data processing flow in this embodiment is as follows:
[0174] 1. Users set analysis parameters through the interactive interface, including data directory, sampling frequency, target frequency, frequency range, etc.
[0175] 2. The system initializes the processing thread pool and reads data files from the specified directory;
[0176] 3. Perform the following operations on each data file:
[0177] a. Read the file content and parse the data according to the specified delimiter;
[0178] b. Verify data validity and filter out files that exceed the loss rate threshold;
[0179] c. Preprocess and resample the valid data;
[0180] d. Calculate time-domain statistical parameters (maximum, minimum, average, etc.);
[0181] e. The Welch method is used to calculate the power spectral density, with a focus on the 0-1Hz frequency band;
[0182] f. Identify the dominant frequency and calculate the corresponding energy within the set frequency range;
[0183] g. Generate time-domain and spectrogram plots, and label key features;
[0184] 4. Summarize and statistically analyze the results of all documents:
[0185] a. Calculate the statistical parameters (mean, standard deviation, etc.) of the dominant frequency and energy in each direction;
[0186] b. Generate distribution charts of the dominant frequency and energy;
[0187] c. Sort the files according to statistical parameters and generate TopN analysis charts;
[0188] 5. Save all analysis results (statistical data, charts) to the specified directory;
[0189] 6. Generate analysis reports and record any abnormal information during the processing.
[0190] In addition, this embodiment can transform desktop applications into web application architectures, adopting a front-end and back-end separation design, allowing users to access the system through a browser, thereby improving the system's accessibility and cross-platform compatibility.
[0191] Therefore, this application achieves efficient processing of large-scale data by using thread pool technology to acquire and preprocess measured vibration data in parallel, thus solving the problem of low data processing efficiency in existing technologies. By extracting the target low-frequency band and using the target spectrum analysis method, it achieves targeted capture of key low-frequency vibration characteristics of wind turbine generators, solving the problems of insufficient attention to the 0-1Hz low-frequency band and low extraction accuracy in existing technologies. By dividing the target low-frequency band into multiple sub-bands and identifying their target main frequency and target energy characteristics respectively, it achieves multi-dimensional analysis of vibration signals, solving the problem that existing technologies cannot effectively separate the frequency components of different components.
[0192] See Figure 2 As shown, an embodiment of the present invention discloses a wind turbine vibration frequency analysis device, comprising:
[0193] The data preprocessing module 11 is used to acquire several measured vibration data of the target wind turbine in parallel from the target data source based on thread pool technology, and to perform data preprocessing on the measured vibration data to obtain the corresponding target vibration data.
[0194] The frequency band extraction module 12 is used to determine the power spectral density corresponding to the target vibration data using a target spectrum analysis method, and extract the target low-frequency band corresponding to the target vibration data from the power spectral density; wherein, the target low-frequency band represents the frequency band where the vibration frequency of the wind turbine is less than a preset frequency threshold;
[0195] The energy characteristic determination module 13 is used to select a target number of sub-frequency bands from the target low-frequency band, and determine the target main frequency and the target energy characteristic corresponding to each sub-frequency band respectively; wherein, the target main frequency is the characteristic frequency characterizing the vibration component of the corresponding sub-frequency band;
[0196] The vibration frequency analysis module 14 is used to generate a frequency analysis file corresponding to the target wind turbine based on each of the target main frequencies and each of the target energy characteristics, so as to analyze the vibration frequency of the target wind turbine.
[0197] In some specific embodiments, the data preprocessing module 11 further includes:
[0198] A cache field creation unit is used to create a target spectrum cache dictionary; wherein, the target spectrum cache dictionary is used to store data in the form of key-value pairs, the key of the target spectrum cache dictionary is the hash value corresponding to the vibration data, and the value of the target spectrum cache dictionary is the power spectral density corresponding to the vibration data.
[0199] In some specific embodiments, the data preprocessing module 11 further includes:
[0200] The data acquisition unit is used to determine whether the target vibration data and the power spectral density exist in the target spectrum cache dictionary. If the target vibration data and the power spectral density exist in the target spectrum cache dictionary, the target vibration data and the power spectral density are directly acquired from the target spectrum cache dictionary.
[0201] The step jump unit is used to jump to the step of acquiring several measured vibration data of the target wind turbine in parallel from the target data source based on thread pool technology if the power spectral density is not found in the target spectrum cache dictionary.
[0202] In some specific embodiments, the data preprocessing module 11 may specifically include:
[0203] The data verification unit is used to verify the measured vibration data so as to select the vibration data with a data loss rate of less than a preset loss rate threshold from the measured vibration data.
[0204] The data resampling unit is used to perform outlier processing and missing value imputation on the filtered vibration data to obtain the corresponding processed data, and to resample the processed data based on the target frequency to obtain the target vibration data.
[0205] In some specific embodiments, the frequency band extraction module 12 may specifically include:
[0206] The power spectral density determination unit is used to dynamically set the target parameters corresponding to the target spectrum analysis method based on the original sampling frequency corresponding to the measured vibration data and the target low-frequency band, and to determine the power spectral density corresponding to the target vibration data using the target parameters and the target spectrum analysis method; wherein, the target parameters include the time window length and the time window function.
[0207] In some specific embodiments, the energy characteristic determination module 13 may specifically include:
[0208] An energy feature determination unit is used to determine the initial master frequency corresponding to each of the sub-frequency bands, and to determine whether the initial energy feature corresponding to each initial master frequency is less than a preset energy threshold. If any initial energy feature is not less than the preset energy threshold, then it is determined whether the index of the initial master frequency corresponding to any initial energy feature is within a preset index range. If the index of the initial master frequency corresponding to any initial energy feature is within the preset index range, then the initial energy feature is determined as the target energy feature of the corresponding sub-frequency band, and the initial master frequency corresponding to the initial energy feature is determined as the target master frequency of the corresponding sub-frequency band.
[0209] In some specific embodiments, the wind turbine vibration frequency analysis device further includes:
[0210] The data display module is used to display the data processing progress and the frequency analysis file corresponding to the target wind turbine using a preset visual user interface.
[0211] Furthermore, embodiments of this application also disclose an electronic device, Figure 3 This is a structural diagram of an electronic device 20 according to an exemplary embodiment. The content of the diagram should not be construed as limiting the scope of this application.
[0212] Figure 3 This is a schematic diagram of the structure of an electronic device 20 provided in an embodiment of this application. Specifically, the electronic device 20 may include: at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25, and a communication bus 26. The memory 22 stores a computer program, which is loaded and executed by the processor 21 to implement the relevant steps in the wind turbine vibration frequency analysis method disclosed in any of the foregoing embodiments. Alternatively, the electronic device 20 in this embodiment may specifically be an electronic computer.
[0213] In this embodiment, the power supply 23 is used to provide operating voltage for each hardware device on the electronic device 20; the communication interface 24 can create a data transmission channel between the electronic device 20 and external devices, and the communication protocol it follows can be any communication protocol applicable to the technical solution of this application, and is not specifically limited here; the input / output interface 25 is used to acquire external input data or output data to the outside world, and its specific interface type can be selected according to specific application needs, and is not specifically limited here.
[0214] In addition, the memory 22, as a carrier for resource storage, can be a read-only memory, random access memory, disk or optical disk, etc. The resources stored thereon can include operating system 221, computer program 222, etc., and the storage method can be temporary storage or permanent storage.
[0215] The operating system 221 is used to manage and control the various hardware devices on the electronic device 20 and the computer program 222, which may be Windows Server, Netware, Unix, Linux, etc. In addition to including a computer program capable of performing the wind turbine vibration frequency analysis method executed by the electronic device 20 as disclosed in any of the foregoing embodiments, the computer program 222 may further include computer programs capable of performing other specific tasks.
[0216] Furthermore, this application also discloses a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the aforementioned wind turbine vibration frequency analysis method. Specific steps of this method can be found in the corresponding content disclosed in the foregoing embodiments, and will not be repeated here.
[0217] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to in the method section.
[0218] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.
[0219] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.
[0220] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.
[0221] The technical solutions provided in this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.
Claims
1. A method for analyzing the vibration frequency of a wind turbine generator set, characterized in that, include: Based on thread pool technology, several measured vibration data of the target wind turbine are obtained in parallel from the target data source, and the measured vibration data are preprocessed to obtain the corresponding target vibration data. The power spectral density corresponding to the target vibration data is determined using the target spectrum analysis method, and the target low-frequency band corresponding to the target vibration data is extracted from the power spectral density; wherein, the target low-frequency band represents the frequency band where the vibration frequency of the wind turbine is less than a preset frequency threshold; Select a target number of sub-frequency bands from the target low-frequency band, and determine the target main frequency and the target energy characteristics corresponding to each sub-frequency band; wherein, the target main frequency is the characteristic frequency characterizing the vibration component of the corresponding sub-frequency band; Based on the target main frequencies and the target energy characteristics, a frequency analysis file corresponding to the target wind turbine is generated to analyze the vibration frequency of the target wind turbine.
2. The method for analyzing the vibration frequency of a wind turbine generator according to claim 1, characterized in that, Before acquiring several measured vibration data of the target wind turbine in parallel from the target data source based on thread pool technology, the following steps are also included: Create a target spectrum cache dictionary; wherein the target spectrum cache dictionary is used to store data in the form of key-value pairs, the key of the target spectrum cache dictionary is the hash value corresponding to the vibration data, and the value of the target spectrum cache dictionary is the power spectral density corresponding to the vibration data.
3. The method for analyzing the vibration frequency of a wind turbine generator according to claim 2, characterized in that, Before acquiring several measured vibration data of the target wind turbine in parallel from the target data source based on thread pool technology, it also includes... Determine whether the target vibration data and the power spectral density exist in the target spectrum cache dictionary. If the target vibration data and the power spectral density exist in the target spectrum cache dictionary, then directly obtain the target vibration data and the power spectral density from the target spectrum cache dictionary. If the power spectral density is not present in the target spectrum cache dictionary, then proceed to the step of obtaining several measured vibration data of the target wind turbine in parallel from the target data source based on thread pool technology.
4. The method for analyzing the vibration frequency of a wind turbine generator according to claim 1, characterized in that, The step of preprocessing the measured vibration data to obtain the corresponding target vibration data includes: The measured vibration data is validated to select vibration data with a data loss rate less than a preset loss rate threshold. The filtered vibration data is subjected to outlier processing and missing value imputation to obtain the corresponding processed data, and the processed data is resampled based on the target frequency to obtain the target vibration data.
5. The method for analyzing the vibration frequency of a wind turbine generator according to claim 1, characterized in that, The method of determining the power spectral density corresponding to the target vibration data using target spectrum analysis includes: The target parameters corresponding to the target spectrum analysis method are dynamically set according to the original sampling frequency corresponding to the measured vibration data and the target low-frequency band, and the power spectral density corresponding to the target vibration data is determined by using the target parameters and the target spectrum analysis method; wherein, the target parameters include the time window length and the time window function.
6. The method for analyzing the vibration frequency of a wind turbine generator according to claim 1, characterized in that, The step of determining the target main frequency corresponding to each of the sub-frequency bands and the target energy characteristics corresponding to each of the target main frequencies includes: Determine the initial master frequency corresponding to each of the sub-frequency bands, and determine whether the initial energy characteristic corresponding to each initial master frequency is less than a preset energy threshold. If any initial energy characteristic is not less than the preset energy threshold, determine whether the index of the initial master frequency corresponding to any initial energy characteristic is within a preset index range. If the index of the initial master frequency corresponding to any initial energy characteristic is within the preset index range, determine any initial energy characteristic as the target energy characteristic of the corresponding sub-frequency band, and determine the initial master frequency corresponding to any initial energy characteristic as the target master frequency of the corresponding sub-frequency band.
7. The method for analyzing the vibration frequency of a wind turbine generator according to any one of claims 1 to 6, characterized in that, Also includes: The data processing progress and frequency analysis file corresponding to the target wind turbine are displayed using a preset visual user interface.
8. A vibration frequency analysis device for wind turbine generators, characterized in that, include: The data preprocessing module is used to acquire several measured vibration data of the target wind turbine in parallel from the target data source based on thread pool technology, and to perform data preprocessing on the measured vibration data to obtain the corresponding target vibration data. The frequency band extraction module is used to determine the power spectral density corresponding to the target vibration data using a target spectrum analysis method, and to extract the target low-frequency band corresponding to the target vibration data from the power spectral density; wherein, the target low-frequency band represents the frequency band where the vibration frequency of the wind turbine is less than a preset frequency threshold; An energy characteristic determination module is used to select a target number of sub-frequency bands from the target low-frequency band, and determine the target main frequency and the target energy characteristic corresponding to each sub-frequency band respectively; wherein, the target main frequency is the characteristic frequency characterizing the vibration component of the corresponding sub-frequency band; The vibration frequency analysis module is used to generate a frequency analysis file corresponding to the target wind turbine based on each of the target main frequencies and each of the target energy characteristics, so as to analyze the vibration frequency of the target wind turbine.
9. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor for executing the computer program to implement the wind turbine vibration frequency analysis method as described in any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that, Used to store a computer program, which, when executed by a processor, implements the wind turbine vibration frequency analysis method as described in any one of claims 1 to 7.