A collaborative extraction method for the 5th-order WPT-high-resolution FFT of broadband harmonics in offshore wind turbine MMC systems.

Abstract

This invention relates to a method for collaborative extraction of db4 fundamental fifth-order WPT-high-resolution FFT of broadband harmonics in offshore wind turbine MMC systems. The method includes: acquiring the raw time-domain signal of the offshore wind turbine MMC system and preprocessing it to obtain a clean signal; selecting the db4 wavelet as the fundamental wavelet and performing fifth-order wavelet packet decomposition on the clean signal to divide the broadband into multiple sub-bands; calculating the energy proportion of each sub-band and selecting effective frequency bands based on a preset threshold; performing Fast Fourier Transform on the signal of each effective frequency band to extract harmonic parameters, including frequency, amplitude, and phase; integrating the harmonic parameters extracted from all effective frequency bands to form a complete broadband harmonic parameter list and calculating the total harmonic distortion rate. This invention combines the advantages of wavelet packet decomposition in frequency band division with the advantages of high-resolution FFT in parameter identification, effectively suppressing spectral leakage and improving the accuracy of frequency and amplitude identification, making it suitable for power quality monitoring of offshore wind power MMC systems.

Description

Technical Field

[0001] This invention belongs to the field of power system power quality assessment and harmonic detection technology, specifically involving a method for collaborative extraction of the db4 fundamental fifth-order WPT-high-resolution FFT of broadband harmonics in offshore wind turbine MMC systems. Background Technology

[0002] Offshore wind turbine MMC flexible DC transmission systems generate broadband harmonic signals in the 2~2500Hz range during operation. These harmonics are characterized by non-stationarity, a large frequency band, and susceptibility to fluctuations in system operating conditions. Accurately extracting these harmonic parameters is a crucial prerequisite for power quality assessment, harmonic mitigation scheme optimization, and system safety and stability operation evaluation.

[0003] In existing technologies, the mainstream approach for broadband harmonic extraction in offshore wind turbine MMC systems is based on a single Fourier transform (FFT), and its implementation process mainly includes three key stages: signal acquisition, direct transform analysis, and parameter output. However, this approach has the following technical drawbacks:

[0004] (1) The problem of spectrum leakage is prominent. The harmonic signal of the MMC system of offshore wind turbine is affected by wind speed fluctuations, turbine start-up and shutdown and converter switching, etc., and has significant non-stationary characteristics. The single FFT transformation is based on the assumption of signal stationarity and does not take effective window function preprocessing measures, which leads to energy diffusion of the signal during the spectrum transformation process, making it impossible to accurately locate the harmonic frequency, and the frequency identification error is as high as ±20Hz.

[0005] (2) Insufficient frequency band targeting. The 2~2500Hz wideband contains multiple harmonic-dense areas and noise-dominant areas. Existing technologies do not divide the frequency band and perform transformation analysis on the effective harmonic signal and noise signal together, resulting in severe noise interference, low amplitude identification accuracy, and an error of more than ±8%.

[0006] (3) Low frequency resolution. Existing technologies use fewer sampling points, and the frequency resolution can usually only reach 1 Hz or more. They cannot distinguish harmonic signals in adjacent frequency bands, especially in scenarios where harmonic frequencies are close, which can easily lead to harmonic parameter confusion.

[0007] (4) Poor adaptability to non-stationary operating conditions. Offshore wind power operation conditions are complex and changeable. The amplitude and frequency of harmonic signals will change dynamically with the operating conditions. A single FFT transformation cannot track the dynamic change characteristics of the signal, resulting in insufficient stability and reliability of the harmonic extraction results.

[0008] Therefore, how to suppress spectrum leakage, achieve accurate wideband division, and improve the accuracy of frequency and amplitude identification has become a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0009] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method for collaborative extraction of the db4 fundamental fifth-order WPT-high-resolution FFT of broadband harmonics in offshore wind turbine MMC systems.

[0010] To achieve the above objectives, the present invention adopts the following technical solution: a method for collaborative extraction of the db4 fundamental fifth-order WPT-high-resolution FFT of broadband harmonics in offshore wind turbine MMC systems, comprising the following steps:

[0011] Step S1: Acquire the raw time domain signal of the offshore wind turbine MMC system and preprocess the raw time domain signal, including windowing and leakage suppression, DC stripping and trend elimination, to obtain a clean signal;

[0012] Step S2: Select the db4 wavelet as the base wavelet and perform fifth-order wavelet packet decomposition on the clean signal to divide the wideband into multiple sub-bands;

[0013] Step S3: Calculate the energy percentage of each sub-band, and select sub-bands with an energy percentage greater than or equal to the preset energy percentage threshold as effective frequency bands;

[0014] Step S4: Perform a Fast Fourier Transform on the signal of each selected effective frequency band to extract the harmonic parameters in each effective frequency band. The harmonic parameters include frequency, amplitude and phase.

[0015] Step S5: Integrate the harmonic parameters extracted from all effective frequency bands to form a complete broadband harmonic parameter list, and calculate the total harmonic distortion rate.

[0016] Further, step S1 specifically includes:

[0017] Targeted data collection is achieved by optimizing monitoring points, sampling frequency, and data collection duration.

[0018] The Hanning window was selected and the window length was determined according to the acquisition parameters. The signal was then windowed to suppress spectral leakage.

[0019] The DC component in the signal is calculated and removed using the mean method.

[0020] The least squares method is used to fit and eliminate the trend term in the signal;

[0021] The preprocessed signal is quality checked to ensure that the signal-to-noise ratio and smoothness meet the standards.

[0022] Furthermore, in the windowing process, the length of the Hanning window is determined based on the product of the sampling frequency and the acquisition duration, achieving a perfect match between the window function and the signal duration; in the DC stripping process, the mean method obtains the DC component by summing with high precision and dividing by the total number of data points, and then subtracts it point by point; in the trend elimination process, a first-order or second-order trend term model is selected based on the signal change characteristics, and the least squares fitting parameters are solved iteratively.

[0023] Further, step S2 specifically includes:

[0024] Comparative analysis was used to verify the adaptability of the db4 wavelet to broadband non-stationary harmonic signals.

[0025] The decomposition order is set to fifth order, and the 2~2500Hz wideband is divided into 16 equal-width sub-bands;

[0026] The decomposition process is monitored in real time to ensure that there is no shift in the frequency range;

[0027] Verify the validity of the decomposition results, including frequency coverage integrity and non-overlap.

[0028] The decomposed sub-band data is stored in a standardized manner.

[0029] Furthermore, step S3 specifically includes:

[0030] An energy calculation model is constructed using the root mean square method to calculate the total energy of each sub-band.

[0031] The total energy of all sub-bands is summed and normalized to obtain the energy percentage of each sub-band.

[0032] Through multi-condition analysis and effect evaluation, the energy percentage threshold was determined to be 0.5%.

[0033] Sub-bands with an energy percentage of less than 0.5% are marked as noise bands and removed, while sub-bands with an energy percentage of ≥0.5% are retained as effective bands.

[0034] The screening results are verified to ensure that the main harmonic energy is retained.

[0035] Furthermore, the energy percentage threshold is 0.5%. This threshold is determined by collecting multi-condition operating data, analyzing energy distribution patterns, constructing an evaluation index system that includes effective harmonic energy retention rate and noise rejection rate, and verifying through multi-threshold comparison, so that the effective harmonic energy retention rate is not less than 95% and the noise rejection rate is not less than 80%.

[0036] Furthermore, step S4 specifically includes:

[0037] The number of sampling points for the Fast Fourier Transform is set to 32,768, so that the frequency resolution reaches 0.25Hz;

[0038] The effective frequency band signal is resampled to match the number of data points to the number of sampling points;

[0039] Perform a fast Fourier transform independently on each effective frequency band to generate its own spectrum.

[0040] The peak detection method is used to initially extract the frequency and amplitude from the spectrum, and the phase is extracted from the phase spectrum.

[0041] The initially extracted parameters are validated, and outliers are removed.

[0042] Furthermore, the number of sampling points for the Fast Fourier Transform is set to 32,768, which is calculated based on the frequency resolution requirement and the sampling frequency according to the relationship between frequency resolution = sampling frequency / number of sampling points. By comparing with 16,384 and 65,536 sampling points, it is verified that the optimal balance is achieved between frequency identification accuracy, amplitude calculation error and calculation time.

[0043] Furthermore, the extraction accuracy of the harmonic parameters is optimized in the following ways: the frequency parameter is calculated by statistically averaging the results of multiple extractions of the same frequency and removing outliers, thereby controlling the identification error within ±5Hz; the amplitude parameter is calculated by weighted averaging by introducing the energy proportion of the effective frequency band as a weight, thereby controlling the identification error within ±2%.

[0044] Further, step S5 specifically includes:

[0045] The extracted harmonic parameters are classified and organized according to fundamental frequency and harmonics;

[0046] Verify the adaptability of the total harmonic distortion rate calculation formula to broadband harmonics in MMC systems;

[0047] A high-precision calculation method is used to calculate the total harmonic distortion rate step by step to control the error in each stage;

[0048] The completeness, accuracy, and reasonableness of the extracted results are verified from multiple dimensions.

[0049] The final results are output and stored in a standardized format.

[0050] Compared with the prior art, the present invention has the following beneficial effects:

[0051] Compared with the prior art, the present invention has the following technical advantages:

[0052] (1) The spectrum leakage suppression effect is significant and the adaptability of non-stationary signals is greatly improved. This invention adopts Hanning window collaborative descrambling preprocessing, and effectively reduces the interference of the original signal through operations such as signal directional acquisition optimization, window function adaptation, DC component stripping, trend term elimination and quality verification; at the same time, the broadband is divided into 16 precise sub-bands by db4 fundamental fifth-order wavelet packet decomposition, and then each effective sub-band is analyzed separately by FFT analysis to avoid energy diffusion caused by full-band mixed transformation.

[0053] (2) The accuracy of harmonic frequency identification is greatly improved, and the error is controlled within ±5Hz. In this invention, 32,768 sampling points are set in the FFT analysis stage to achieve a frequency resolution of 0.25Hz; at the same time, through precise frequency band division by wavelet packet decomposition, each FFT analysis is only for a specific frequency band, avoiding mutual interference between harmonics of different frequency bands.

[0054] (3) The accuracy of harmonic amplitude identification is significantly improved, and the error is controlled within ±2%. This invention uses energy ratio to drive effective frequency band screening, eliminates noise-dominated frequency bands with energy ratio <0.5%, and reduces the interference of noise on amplitude identification; in the amplitude calculation process, weighted optimization is performed in combination with the energy ratio of each effective frequency band to improve the accuracy of amplitude calculation.

[0055] (4) Wideband coverage is comprehensive and highly targeted, balancing harmonic extraction efficiency and reliability. This invention uses db4 fundamental fifth-order wavelet packet decomposition to accurately cover a wideband of 2~2500Hz, with no frequency omissions or overlaps in the 16 sub-bands; it focuses on core harmonic frequency bands by energy ratio screening to reduce unnecessary calculations; at the same time, it establishes a full-process quality verification mechanism to ensure that the processing results of each link meet the requirements. Attached Figure Description

[0056] Figure 1 This is a flowchart illustrating the implementation of the db4 fundamental fifth-order WPT-high-resolution FFT collaborative extraction method for broadband harmonics in offshore wind turbine MMC systems provided in this embodiment of the invention.

[0057] Figure 2 This is a flowchart illustrating the implementation of multi-dimensional interference source purification pretreatment in an embodiment of the present invention;

[0058] Figure 3 This is a flowchart illustrating the implementation of the fifth-order frequency band fine decomposition of the db4 fundamental wave in this embodiment of the invention.

[0059] Figure 4 This is a flowchart illustrating the implementation of energy percentage core frequency band selection in this embodiment of the invention.

[0060] Figure 5 This is a flowchart illustrating the implementation of high-resolution FFT harmonic parameter refinement in this embodiment of the invention.

[0061] Figure 6 This is a flowchart illustrating the implementation of harmonic index quantization output integration in this embodiment of the invention. Detailed Implementation

[0062] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0063] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of this application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0064] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to this application. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0065] This embodiment provides a db4 fundamental fifth-order WPT-high-resolution FFT collaborative extraction method for broadband harmonics in offshore wind turbine MMC systems. Addressing the problems of severe spectral leakage, poor adaptability to non-stationary signals, and low parameter identification accuracy in broadband harmonic extraction of offshore wind turbine MMC systems, it constructs a five-step closed-loop system of "signal purification - band segmentation - frequency selection - fine extraction - quantification." Starting with signal purification, it lays a foundation of clean data; then, it achieves precise broadband segmentation, building a refined analysis framework; next, it focuses on the core effective frequency bands to eliminate noise interference; subsequently, it uses high-resolution technology to complete precise parameter identification; finally, it achieves index quantification and standardized output. Each step is progressive and mutually reinforcing, forming a complete and efficient harmonic extraction process. Figure 1 As shown, its specific implementation process is as follows.

[0066] Step S1: Multi-dimensional interference source purification pretreatment

[0067] The raw time-domain signal of the offshore wind turbine MMC system is acquired and preprocessed, including windowing and leakage suppression, DC stripping and trend elimination, to obtain a clean signal.

[0068] The DC component, trend term, and truncation interference mixed in the original acquired signal can severely distort the harmonic characteristics. This step uses a progressive operation of "acquisition optimization - windowing and leakage suppression - DC stripping - trend elimination - quality verification" to gradually purify the signal. The processing results of the previous step provide high-quality input for the next step, ensuring continuous improvement in signal purity. Figure 2 As shown, the implementation steps are as follows.

[0069] (1) Optimization of directional multidimensional acquisition adaptation

[0070] To obtain complete raw data covering a wide frequency band of 2~2500Hz that accurately reflects harmonic characteristics, the following five steps are followed:

[0071] 1) In-depth analysis of the harmonic generation and transmission mechanism of the MMC flexible DC transmission system, clarifying that the converter output end is the key monitoring area with the most concentrated harmonic signals and the least attenuation, accurately locating the monitoring point in this area to avoid signal distortion due to improper selection of monitoring points;

[0072] 2) Based on the Nyquist sampling theorem and the highest frequency of the target harmonic band of 2~2500Hz, the minimum sampling frequency is calculated to be 5000Hz. Taking into account both computational efficiency and signal integrity, the sampling frequency is set to 8192Hz to ensure that no high-frequency harmonic signals are missed.

[0073] 3) Analyze the dynamic change period of harmonic signals under different operating conditions of offshore wind turbines (such as startup, stable operation, and shutdown), and determine that 10 seconds is the optimal acquisition time, which can completely capture the harmonic fluctuation characteristics under each operating condition;

[0074] 4) Select high-precision voltage and current sensors with an accuracy class of 0.2. Calibrate and debug the sensors before deployment to ensure the accuracy of the sensor output data;

[0075] 5) During the data acquisition process, monitor the sensor's operating voltage, temperature, and other status parameters in real time, and establish an abnormal early warning mechanism. Once the parameters are found to deviate from the normal range, immediately trigger the re-acquisition process.

[0076] This step, through multi-dimensional collaboration including precise location of monitoring points, scientific setting of sampling frequency, optimization of acquisition duration, accurate selection of sensors, and real-time monitoring of the acquisition process, ensures the integrity, authenticity, and validity of the original data from the source, laying a solid data foundation for subsequent signal purification and harmonic extraction.

[0077] (2) Hanning window parameter adaptation to suppress leakage

[0078] To address the spectral leakage problem caused by the non-stationary characteristics of harmonic signals and truncation interference, the following six steps should be taken:

[0079] 1) The system sorts out the spectral characteristics of commonly used window functions such as rectangular window, Hanning window, Hamming window, and Blackman window, compares and analyzes the performance differences of various window functions in suppressing spectral leakage and preserving signal amplitude accuracy. Combined with the non-stationary characteristics of harmonic signals in offshore wind turbine MMC system, the Hanning window is determined to be the optimal preprocessing window function. Its smooth transition amplitude characteristics can effectively reduce the interference caused by abrupt changes at the start and end of the signal.

[0080] 2) Based on the set sampling frequency of 8192Hz and the acquisition duration of 10 seconds, the length of the Hanning window is accurately calculated to be 81920 data points (8192Hz×10s) to ensure that the window function is completely matched with the acquisition signal duration and to avoid signal distortion caused by mismatch in window length.

[0081] 3) Based on the mathematical expression of the Hanning window, generate a sequence of window functions with the same length as the acquired signal to ensure that the amplitude variation of the window function meets the requirements for suppressing spectral leakage;

[0082] 4) By employing a point-by-point multiplication operation, the generated Hanning window function sequence is fused with the original time-domain signal to achieve the window function wrapping the original signal and reduce abrupt changes in the signal at the beginning and end.

[0083] 5) Perform preliminary spectral analysis on the windowed signal, and quantitatively evaluate the suppression effect of spectral leakage by comparing the spectrum diagrams of the signal before and after windowing;

[0084] 6) If the spectral leakage suppression effect does not meet the preset standard, readjust the window function parameters or change the window function type until the requirements are met.

[0085] This step, through a complete process of scientific window function selection, precise parameter calculation, window function generation, signal windowing fusion, effect evaluation, and parameter optimization, effectively suppresses spectral leakage caused by signal truncation, while preserving the original characteristics of harmonic signals to the maximum extent, creating favorable conditions for subsequent interference removal work.

[0086] (3) Mean value method for fine stripping of DC component

[0087] To eliminate the interference of DC components on the accuracy of low-order harmonic identification, follow these five steps:

[0088] 1) Extract the complete signal data sequence after windowing processing to ensure that there are no missing data or outliers;

[0089] 2) The DC component is calculated using the mean method. First, all data points after windowing are summed one by one. High-precision calculation is used in the summation process to avoid summation errors caused by data overflow.

[0090] 3) Divide the summation result by the total number of data points (81920) to obtain the accurate DC component value. Six decimal places are retained during the calculation process to improve the calculation accuracy of the DC component.

[0091] 4) Subtract the windowed signal data sequence from the calculated DC component value point by point to achieve the initial stripping of the DC component;

[0092] 5) Perform baseline level detection on the signal after removing the DC component. Calculate the mean, variance and other statistical quantities of the signal to determine whether there is any residual DC component. If there is any residual, repeat the above steps until the signal baseline level is stable within the preset range.

[0093] This step, through a closed-loop operation of data extraction, high-precision summation, DC component calculation, point-by-point stripping, and residual detection, completely removes the DC component from the original signal, preventing the DC component from masking the characteristics of low-order harmonic signals. This makes the characteristics of the 2~2500Hz broadband harmonic signal more prominent, providing a cleaner signal source for subsequent trend term elimination work.

[0094] (4) Self-elimination of least squares trend term

[0095] To address trend term interference caused by factors such as sensor drift and changes in ambient temperature, the following six steps should be taken:

[0096] 1) Extract the signal data after stripping the DC component, smooth the data, and remove random noise from the signal to facilitate more accurate identification of trend terms;

[0097] 2) Analyze the trend of the signal and determine whether the trend term is linear or nonlinear. If the signal changes approximately linearly with time, construct a first-order trend term model; if it changes nonlinearly, construct a second-order trend term model.

[0098] 3) Based on the principle of least squares, establish a trend term fitting objective function, and determine the parameters to be estimated in the trend term model with the goal of minimizing the fitting error;

[0099] 4) Solve for the parameters to be estimated by iterative calculation. Set a convergence threshold during the iteration process. Stop the iteration when the difference between two iteration results is less than the convergence threshold and obtain the optimal parameters.

[0100] 5) Substitute the optimal parameters into the trend term model to generate the fitted trend term. Subtract the fitted trend term point by point from the signal after removing the DC component to achieve adaptive elimination of the trend term.

[0101] 6) Perform a smoothness test on the signal after eliminating the trend term, calculate the smoothness index of the signal, and if the index does not meet the preset standard, readjust the trend term model type or fitting parameters until the signal baseline is stable and there is no obvious drift.

[0102] This step, through systematic operations such as data smoothing, trend term type identification, model building, parameter iterative solution, trend term elimination, and smoothness verification, effectively eliminates the interference of trend terms on harmonic signals, further improves the purity of harmonic signals in the offshore wind turbine MMC system, and provides a reliable guarantee for subsequent frequency band division and screening.

[0103] (5) Preprocessing signal quality verification optimization

[0104] To ensure that the preprocessed signal meets the requirements of subsequent processing, follow these five steps:

[0105] 1) Define the core indicators for signal quality verification, including signal-to-noise ratio, smoothness, and harmonic component integrity. The signal-to-noise ratio should be ≥30dB, the signal should have no obvious abrupt changes or glitches, and there should be no loss of 2~2500Hz wideband harmonic components.

[0106] 2) A signal-to-noise ratio (SNR) calculation algorithm is used to calculate the SNR of the preprocessed signal. By comparing the ratio of signal power to noise power, an accurate SNR value is obtained.

[0107] 3) Calculate the first and second derivatives of the signal using a signal smoothness analysis algorithm, and determine whether there are obvious abrupt changes or glitches in the signal based on the changes in the derivatives;

[0108] 4) Perform spectral pre-analysis on the preprocessed signal to generate a spectrum diagram of the signal, and visually verify whether the harmonic components in the 2~2500Hz wideband are completely preserved;

[0109] 5) If the verification result does not meet the requirements, trace back to the corresponding preprocessing step, such as re-optimizing the Hanning window length, adjusting the fitting order of the trend term, replacing the sensor, etc. After adjusting the parameters, re-execute the preprocessing process until the signal quality reaches the preset standard.

[0110] This step, through defining verification indicators, accurately calculating the signal-to-noise ratio, analyzing signal smoothness, performing spectrum pre-analysis, and closed-loop optimization, ensures that the pre-processed offshore wind turbine MMC system signal can accurately reflect the true characteristics of broadband harmonics, laying a solid foundation for subsequent wavelet packet decomposition and FFT refinement analysis.

[0111] Step S2: Precise decomposition of the fifth-order frequency band of the db4 fundamental frequency

[0112] The db4 wavelet is selected as the base wavelet, and the clean signal is decomposed into a fifth-order wavelet packet, dividing the wideband into multiple subbands.

[0113] The preprocessed clean signal needs to undergo precise wideband segmentation. This step uses db4 fundamental wavelet fifth-order wavelet packet decomposition to achieve a scientific segmentation of the 2~2500Hz wideband. The process proceeds logically according to "fundamental wavelet selection - parameter optimization - process monitoring - result verification - data storage," with each step providing technical support for the next, ensuring the accuracy and completeness of the band segmentation. Figure 3 As shown, the implementation steps are as follows.

[0114] (1) Verification of DB4 fundamental frequency compatibility

[0115] To select the fundamental wavelet that best fits a broadband non-stationary harmonic signal, the following five steps are performed:

[0116] 1) The characteristics of commonly used wavelets such as the db series, sym series, and coif series are systematically reviewed. The key indicators such as the tight support, time domain resolution, and frequency domain division capability of various wavelets are analyzed. In combination with the non-stationarity and large frequency band span of the 2~2500Hz wideband harmonic signal of the offshore wind turbine MMC system, the db4 wavelet is initially selected as the candidate basis wavelet.

[0117] 2) Select multiple sets of preprocessed signals of offshore wind turbine MMC system under different operating conditions, and perform trial decomposition using db4 wavelet, db2 wavelet, sym4 wavelet and coif3 wavelet respectively, and obtain the decomposition results of each wavelet;

[0118] 3) Construct an evaluation index system for decomposition effect, including frequency band division accuracy, harmonic feature retention degree, energy concentration, etc., and quantify and score the decomposition results of different wavelets;

[0119] 4) Compare the scoring results of each wavelet to verify that the db4 wavelet has higher frequency band division accuracy than other wavelets, can capture the frequency domain characteristics of harmonic signals more accurately, and performs best in terms of harmonic feature preservation and energy concentration.

[0120] 5) The db4 wavelet was determined as the final basis wavelet, providing reliable basis wavelet support for the subsequent fifth-order wavelet packet decomposition.

[0121] This step, through a complete process of wavelet characteristic analysis, multi-wavelet trial decomposition, construction of an evaluation index system, comparison of decomposition results, and determination of the fundamental wavelet, ensures that the selected fundamental wavelet can accurately meet the decomposition requirements of broadband non-stationary harmonic signals in the offshore wind turbine MMC system, laying the foundation for accurate frequency band division.

[0122] (2) Fifth-order decomposition parameter optimization design

[0123] To achieve complete coverage and precise division of the 2~2500Hz wideband, follow these six steps:

[0124] 1) Based on the frequency range of 2~2500Hz wideband and combined with the frequency band division principle of wavelet packet decomposition, the number of sub-bands and frequency width corresponding to different decomposition orders are calculated. Through theoretical derivation, it is found that the 5th order wavelet packet decomposition can divide the original signal into 16 equal-width sub-bands, and the frequency width of each sub-band is 125Hz (2500Hz / 16), which can completely cover the 2~2500Hz wideband without frequency omission;

[0125] 2) Determine the order of wavelet packet decomposition as 5, set the number of decomposition iterations as 5, and clarify that the decomposition logic of each iteration is to separate the high and low frequencies of the signal based on the scaling function and wavelet function of db4 wavelet.

[0126] 3) Optimize the computational efficiency of the decomposition algorithm by adopting the fast wavelet transform algorithm. By improving the algorithm's iterative steps and data storage method, the amount of computation is reduced and the decomposition speed is increased.

[0127] 4) Set the precision control parameters during the decomposition process to ensure that the amplitude error of the decomposed sub-band signal is controlled within 0.5% and the frequency range error is controlled within 1Hz.

[0128] 5) Verify the rationality of the decomposition parameters through simulation experiments. Select typical harmonic signals for simulated decomposition, compare the decomposition results with the theoretical values, and adjust the parameters until the accuracy requirements are met.

[0129] 6) Establish a dynamic adjustment mechanism for decomposition parameters. If deviations in frequency band division are found during subsequent processing, the decomposition order or iteration parameters can be fine-tuned according to the actual situation.

[0130] This step, through systematic operations including decomposition order calculation, iteration number setting, algorithm efficiency optimization, accuracy parameter control, simulation verification, and dynamic adjustment mechanism construction, ensures that the fifth-order wavelet packet decomposition can accurately and stably divide the 2~2500Hz broadband harmonic signal of the offshore wind turbine MMC system into 16 effective sub-bands, balancing decomposition accuracy and efficiency.

[0131] (3) Real-time monitoring and adjustment of the decomposition process

[0132] To ensure the stability and accuracy of the wavelet packet decomposition process, the following five steps are performed:

[0133] 1) Set up real-time monitoring nodes during the decomposition process. After each decomposition is completed, immediately collect data on the sub-band signals obtained from that decomposition and record key parameters such as the frequency range, signal amplitude, and energy distribution of the sub-band.

[0134] 2) Compare the actual frequency range of each sub-band after decomposition with the theoretical frequency range, calculate the frequency offset, and if the offset exceeds the preset threshold, adjust the scale parameter of the db4 wavelet in time to correct the decomposition direction.

[0135] 3) Analyze the signal amplitude variation trend of the sub-frequency band and compare it with the amplitude characteristics of the preprocessed original signal. If amplitude distortion is found, adjust the decomposition iteration step size and optimize the decomposition algorithm.

[0136] 4) Calculate the energy proportion of each sub-band to make a preliminary judgment on the distribution of harmonic energy. If the energy distribution is abnormally concentrated or dispersed, investigate the data transmission and calculation errors in the decomposition process.

[0137] 5) Establish an abnormal early warning and emergency handling mechanism for the decomposition process. When a serious abnormality is detected, immediately stop the decomposition process, backtrack to the previous decomposition result, and restart the decomposition operation.

[0138] This step ensures the stability and accuracy of the fifth-order wavelet packet decomposition process through real-time monitoring operations, including setting monitoring nodes, frequency range verification, amplitude distortion correction, energy distribution analysis, and handling of anomalies. It avoids frequency band division deviations caused by decomposition parameter drift or calculation errors, and provides high-quality intermediate products for subsequent decomposition result verification.

[0139] (4) Validation of decomposition results

[0140] To ensure that the sub-bands obtained from the decomposition meet the requirements for subsequent screening and analysis, follow these six steps:

[0141] 1) Define the verification dimensions for the validity of the decomposition results, including frequency coverage integrity, signal integrity, non-overlap, and energy rationality;

[0142] 2) Summarize and analyze the frequency range of the 16 sub-bands to verify that the frequency range of all sub-bands covers a wide frequency band of 2~2500Hz without any frequency omissions. If there are omissions, readjust the decomposition parameters and perform a second decomposition.

[0143] 3) Using spectral analysis, the spectral features of the signal in each sub-band are extracted and compared with the spectral features of the preprocessed original signal to ensure that the harmonic signals in the sub-band are not distorted and retain the key features of the original harmonics such as amplitude and phase.

[0144] 4) Verify the frequency range of each sub-band one by one to ensure that there is no frequency overlap between any two sub-bands, so as to avoid the problem of repeated calculation of harmonic parameters in the subsequent screening and analysis process;

[0145] 5) Calculate the energy value of each sub-band, analyze the rationality of the energy distribution, and if the energy value of a certain sub-band deviates significantly from the normal range, it is judged as an abnormal decomposition and the decomposition is repeated.

[0146] 6) Summarize all verification results and generate a validity verification report for the decomposition results. If the verification is successful, proceed to the subsequent data storage stage; if it fails, return to the decomposition stage for re-decomposition.

[0147] This step involves a comprehensive verification process, including setting multi-dimensional verification indicators, verifying frequency coverage integrity, signal integrity, non-overlap, energy rationality, and generating a verification report. This ensures that all 16 sub-bands obtained from the decomposition meet the validity requirements, providing high-quality decomposition results for subsequent effective frequency band selection.

[0148] (5) Decomposed data and standardized storage

[0149] To achieve efficient access and management of decomposed data, follow these five steps:

[0150] 1) Clearly define the storage content of the decomposed data, including key parameters such as the unique identifier number of each sub-band, frequency range, time domain signal data, amplitude information, energy data, and decomposed timestamp;

[0151] 2) Design a structured data storage format, adopt a relational database storage model, and establish related tables such as sub-frequency band information table, time domain data table, and energy data table to ensure the standardization and relevance of data storage;

[0152] 3) Assign a unique identifier to each sub-band. The numbering rules combine information such as decomposition order and frequency band order to facilitate quick identification of sub-band attributes;

[0153] 4) Establish a data indexing system, build an index based on key parameters such as the frequency range and energy value of sub-bands, improve the efficiency of finding effective frequency bands in the subsequent screening process, and shorten the data retrieval time;

[0154] 5) Establish a data backup and security mechanism, regularly back up the stored decomposed data, set data access permissions, and prevent data loss or leakage.

[0155] This step, through standardized operations such as defining storage content, designing storage format, assigning unique identifiers, establishing data indexes, and building security mechanisms, ensures that the decomposed data of broadband harmonics in the offshore wind turbine MMC system can be accessed efficiently and securely. This lays a solid data foundation for subsequent energy proportion calculations and effective frequency band selection, and ensures the smoothness and efficiency of the entire harmonic extraction process.

[0156] Step S3: Energy Proportion Core Frequency Band Screening

[0157] Calculate the energy percentage of each sub-band, and select sub-bands with an energy percentage greater than or equal to the preset energy percentage threshold as effective frequency bands.

[0158] The 16 sub-bands after decomposition contain low-energy frequency bands dominated by noise. This step follows the logic of "energy modeling - normalization processing - threshold determination - screening execution - verification and optimization," focusing on the core harmonic energy frequency bands and eliminating noise interference. The processing results of the previous step provide a basis for judgment in the next step, ensuring that the selected effective frequency bands accurately reflect the core harmonic characteristics. Figure 4 As shown, the implementation steps are as follows.

[0159] (1) Subband energy calculation modeling

[0160] To accurately calculate the harmonic energy of each sub-band, the following five steps are performed:

[0161] 1) Based on the non-stationary characteristics of harmonic signals in offshore wind turbine MMC systems, the influencing factors of time-domain signal energy calculation are analyzed, and the root mean square method is determined to be the core method for energy calculation. This method can effectively reflect the energy distribution characteristics of non-stationary signals.

[0162] 2) Construct a mathematical model for energy calculation, specifying that the input to the model is the time-domain signal data of the sub-frequency band, and the output is the total energy of the sub-frequency band. The model expression is as follows: , where x i This represents the i-th time-domain data point of the sub-band.

[0163] 3) Optimize the weights of the data points in the model. Based on the amplitude variation law of the harmonic signal, assign higher weights to data points with larger amplitudes and lower weights to data points with smaller amplitudes to ensure that the energy calculation results can accurately reflect the true distribution of harmonic energy in the sub-band.

[0164] 4) Set the accuracy threshold for energy calculation, and verify it through multiple simulation experiments to ensure that the error of the energy calculation results is controlled within 0.1%, which meets the accuracy requirements of subsequent energy proportion analysis.

[0165] 5) Develop an energy calculation program module to perform batch calculations on 16 sub-frequency band signals, improve calculation efficiency, and add an abnormal data detection function during the calculation process to automatically remove abnormal values ​​in the time domain signal to avoid affecting the energy calculation results.

[0166] This step, through systematic operations including selection of calculation methods, construction of mathematical models, optimization of data point weights, setting of precision thresholds, and development of program modules, establishes an accurate and efficient sub-band energy calculation system. This provides reliable data support for subsequent energy proportion analysis and avoids errors in effective frequency band selection due to energy calculation errors.

[0167] (2) Total energy normalization treatment

[0168] To convert the energy levels of each sub-band into intuitively comparable relative proportions, the following six steps are performed:

[0169] 1) Extract the energy calculation results of 16 sub-bands and perform consistency verification on all energy data to ensure that there are no calculation errors or outliers in the data;

[0170] 2) Calculate the total energy of the 16 sub-bands using a high-precision summation algorithm to avoid errors in total energy due to data overflow or insufficient calculation precision;

[0171] 3) Calculate the ratio of the energy value of each sub-band to the sum of the total energy to obtain the initial energy percentage of each sub-band;

[0172] 4) Round the initial energy percentage to two decimal places to ensure data simplicity and readability, and to facilitate subsequent threshold comparison.

[0173] 5) Sum and verify the energy percentage of all normalized sub-bands to ensure that the summation result is 100%. If there is a deviation, trace back to the energy calculation stage to check for calculation errors.

[0174] 6) Link and store the normalized energy percentage data with the corresponding sub-band identifier number, frequency range and other information to establish an energy percentage information table.

[0175] This step involves a complete process of data consistency verification, high-precision calculation of total energy, energy percentage determination, data formatting, summation verification, and associated storage. It transforms the absolute energy of each sub-band into a relative percentage, making it easier to intuitively judge the energy contribution of each sub-band in the broadband harmonics of the offshore wind turbine MMC system. This provides a clear and accurate basis for selecting effective frequency bands.

[0176] (3) Verification of effective frequency band threshold determination

[0177] To scientifically set the screening threshold for effective frequency bands, the following seven steps should be followed:

[0178] 1) Collect multiple sets of offshore wind turbine MMC system operation data under different operating conditions, covering typical operating conditions such as startup, stable operation, and shutdown, to ensure the comprehensiveness and representativeness of the data;

[0179] 2) Perform the aforementioned preprocessing, wavelet packet decomposition, and energy normalization on each set of data to obtain the sub-band energy proportion distribution under each working condition;

[0180] 3) Analyze the energy proportion distribution pattern under all operating conditions, statistically analyze the energy proportion range of the sub-frequency bands where harmonic energy is mainly concentrated, and preliminarily determine the candidate range of the effective frequency band energy proportion threshold as 0.3%~0.7%;

[0181] 4) Select multiple threshold points (0.3%, 0.4%, 0.5%, 0.6%, 0.7%) within the candidate interval, and perform effective frequency band filtering for each group of data;

[0182] 5) Construct an evaluation index system for the screening effect, including effective harmonic energy retention rate, noise removal rate, number of frequency bands after screening, etc., and quantify and score the screening results under different thresholds;

[0183] 6) Comparing the scoring results of each threshold, it was found that when the threshold is set to 0.5%, the effective harmonic energy retention rate is ≥95%, the noise removal rate is ≥80%, and the number of frequency bands after screening is moderate. It can effectively remove low-energy frequency bands dominated by noise while retaining the main harmonic energy, thus avoiding the problem of effective harmonic energy loss due to excessively high thresholds or the inability to suppress noise interference due to excessively low thresholds.

[0184] 7) Verify the practicality of this threshold through field experiments. Select the MMC system of an actual offshore wind turbine for testing. Compare the screening results with the actual harmonic distribution to confirm that 0.5% is the optimal effective frequency band energy ratio threshold.

[0185] This step, through a systematic operation involving data collection, multi-condition analysis, candidate interval determination, multi-threshold screening, evaluation index construction, optimal threshold selection, and on-site verification, scientifically determined the energy proportion threshold of the effective frequency band, providing a reliable judgment standard for subsequent frequency band screening.

[0186] (4) Execution based on threshold frequency band filtering

[0187] To accurately select the effective frequency band containing major harmonic energy, follow these six steps:

[0188] 1) Call the standardized storage sub-band energy percentage information table to obtain the identification number, frequency range and energy percentage data of 16 sub-bands;

[0189] 2) Sort all sub-bands in ascending order of frequency range to facilitate orderly screening and subsequent analysis;

[0190] 3) Compare the energy percentage of each sub-band with the 0.5% threshold one by one. If the energy percentage is ≥0.5%, mark it as an effective frequency band and record its identification number, frequency range, energy percentage and other key information; if the energy percentage is <0.5%, mark it as an invalid frequency band (noise-dominated).

[0191] 4) Establish a screening log to record in detail the screening results, screening time, judgment criteria, and other information for each sub-band, so as to facilitate subsequent traceability and verification;

[0192] 5) Remove invalid frequency bands from the subsequent analysis queue, retaining only the relevant data of valid frequency bands to reduce the computational load of subsequent FFT analysis;

[0193] 6) Perform preliminary sorting of the selected effective frequency band data and sort them by frequency range to form an effective frequency band list.

[0194] This step, through standardized operations such as data retrieval, sub-band sorting, threshold comparison and judgment, log record filtering, invalid frequency band removal, and valid frequency band organization, accurately focuses on the core energy frequency band of broadband harmonics in the offshore wind turbine MMC system, effectively removes noise interference, provides a highly targeted and high-quality analysis object for subsequent high-resolution FFT analysis, and improves the efficiency of harmonic parameter extraction.

[0195] (5) Optimization of screening result verification

[0196] To ensure that the selected effective frequency bands meet the requirements of subsequent analysis, the following six steps shall be followed:

[0197] 1) Define the verification dimensions of the screening results, including frequency coverage integrity, total energy ratio, harmonic feature retention, and frequency band rationality;

[0198] 2) Analyze the frequency range of the effective frequency band to verify that it can cover the main harmonic distribution area in the 2~2500Hz wideband and that no key harmonic frequency bands are missed. If there are omissions, appropriately lower the energy percentage threshold and re-screen.

[0199] 3) Calculate the total energy percentage of all effective frequency bands and ensure that the total is ≥95%. If the requirement is not met, adjust the threshold and re-screen to ensure that most of the harmonic energy is retained.

[0200] 4) Perform preliminary spectrum analysis on the time-domain signal of the effective frequency band, extract the amplitude and phase characteristics of harmonics, and compare them with the harmonic characteristics of the preprocessed original signal to verify that the harmonic characteristics have not been distorted;

[0201] 5) Based on the harmonic generation mechanism of the offshore wind turbine MMC system, analyze whether the frequency band distribution of the effective frequency band is reasonable. If there are obviously unreasonable frequency bands (such as frequency bands that exceed the range of possible harmonic generation), check for errors in the screening process.

[0202] 6) If a problem is found during the verification, adjust the energy percentage threshold or correct the screening logic, and re-screen until the screening results meet all verification requirements.

[0203] This step, through multi-dimensional verification, frequency coverage integrity verification, energy ratio summation verification, harmonic feature retention verification, frequency band rationality analysis, and closed-loop optimization, ensures that the selected effective frequency bands can accurately reflect the core characteristics of broadband harmonics in the offshore wind turbine MMC system, providing a high-quality analysis object for subsequent high-resolution FFT analysis.

[0204] Step S4: High-resolution FFT harmonic parameter refinement

[0205] A Fast Fourier Transform is performed on the signal of each selected effective frequency band to extract the harmonic parameters within each effective frequency band. The harmonic parameters include frequency, amplitude, and phase.

[0206] After screening, the effective frequency bands need to undergo precise parameter identification. This step follows the logic of "sampling point optimization - signal resampling - frequency band FFT - initial parameter extraction - optimization and integration" to achieve high-precision extraction of harmonic frequencies, amplitudes, and phases. The previous step provides adaptability data for the next step, ensuring the accuracy and stability of parameter identification. Figure 5 As shown, the implementation steps are as follows.

[0207] (1) Optimization setting of FFT sampling points

[0208] To improve frequency resolution and ensure accurate identification of harmonic parameters, follow these six steps:

[0209] 1) Analyze the frequency range and harmonic characteristics of the effective frequency band, clarify the frequency resolution requirements, and combine the characteristics of harmonic signals in the MMC system of offshore wind turbines to determine that the frequency resolution needs to be within 0.25Hz in order to accurately distinguish harmonic signals in adjacent frequency bands;

[0210] 2) Based on the relationship between sampling frequency and frequency resolution (frequency resolution = sampling frequency / number of sampling points), given that the sampling frequency is 8192Hz, the required number of sampling points is calculated to be 8192Hz / 0.25Hz = 32768. Considering both computational efficiency and hardware resources, 32768 is selected as the number of FFT sampling points to meet the requirements for accurate recognition.

[0211] 3) Compare the FFT analysis results under different numbers of sampling points (16384, 32768, 65536), including indicators such as frequency identification accuracy, amplitude calculation error, and calculation time;

[0212] 4) Through comparison, it was found that the 32,768 sampling points performed well in terms of frequency identification accuracy (error ≤ ±5Hz) and amplitude calculation error (error ≤ ±2%), and the calculation time was moderate, without causing low calculation efficiency due to too many sampling points.

[0213] 5) In conjunction with the real-time monitoring requirements of the offshore wind turbine MMC system, the computational efficiency of 32,768 sampling points was verified to meet the real-time requirements;

[0214] 6) The number of FFT sampling points was determined to be 32768, which will be used as a fixed parameter for subsequent frequency-band FFT analysis.

[0215] This step, through a systematic operation involving frequency resolution requirement analysis, theoretical calculation of the number of sampling points, comparison of multiple sampling points, balance between accuracy and efficiency, and real-time verification, optimizes the setting of the number of FFT sampling points, providing technical support for high-precision identification of harmonic parameters and effectively solving the parameter confusion problem caused by insufficient frequency resolution in existing technologies.

[0216] (2) Effective frequency band signal resampling processing

[0217] To ensure the effective frequency band signal is compatible with the FFT analysis requirements of 32,768 sampling points, follow these seven steps:

[0218] 1) Extract the effective frequency band time domain signal data after filtering, and record key parameters such as the sampling frequency and number of data points of the original signal;

[0219] 2) Analyze the difference between the number of data points in the original signal and the number of 32,768 sampling points. If the number of data points is insufficient, interpolation is required to expand the signal. If the number of data points is excessive, downsampling is required.

[0220] 3) Considering the need for continuity and feature preservation of harmonic signals, linear interpolation is selected as the core method for interpolation expansion. This method can quickly expand data points while ensuring that the signal features are not distorted.

[0221] 4) Based on the principle of linear interpolation, construct an interpolation model, clarify the interpolation interval and interpolation calculation logic, and ensure that the interpolated signal can completely retain the harmonic characteristics of the original effective frequency band;

[0222] 5) Perform interpolation on the effective frequency band signal to expand the number of data points to 32,768, and smooth the interpolated signal to avoid signal abrupt changes caused by interpolation;

[0223] 6) Perform frequency verification on the resampled signal and verify through spectrum analysis that the frequency range of the signal is consistent with the original effective frequency band and no frequency band shift has occurred;

[0224] 7) Perform amplitude verification on the resampled signal to ensure that the amplitude change trend is consistent with the original signal and there is no amplitude distortion.

[0225] This step, through the entire process of original signal analysis, interpolation method selection, interpolation model construction, interpolation calculation and smoothing, and frequency and amplitude verification, realizes the resampling processing of the effective frequency band signal, providing the required signal data for high-resolution FFT analysis and ensuring the accuracy and effectiveness of FFT transform.

[0226] (3) Frequency-segmented FFT transformation execution

[0227] To avoid mutual interference between harmonic signals of different frequency bands and improve the accuracy of FFT analysis, the following six steps should be followed:

[0228] 1) Group the resampled effective frequency band signals in ascending order of frequency range, with each effective frequency band serving as an independent analysis unit;

[0229] 2) For each effective frequency band, set the frequency analysis interval for FFT transformation to ensure that the analysis interval completely covers the frequency range of the effective frequency band and avoid parameter loss due to harmonic signals exceeding the analysis interval;

[0230] 3) Configure the core parameters of the FFT transform, including the number of sampling points (32768), frequency analysis interval, and window function type (using the Hanning window), to ensure the consistency and adaptability of parameter settings;

[0231] 4) The Fast Fourier Transform (FFT) algorithm is used to perform FFT transformation on each effective frequency band signal separately to generate the spectrum (amplitude spectrum and phase spectrum) of each frequency band.

[0232] 5) Record the key parameters in the FFT transformation process of each effective frequency band, such as the number of transformation iterations, the number of spectral points, and the calculation time, to facilitate subsequent result verification and traceability;

[0233] 6) Perform a preliminary quality check on the FFT-transformed spectrum to ensure that the spectrum has no obvious distortion and low noise interference. If any abnormalities are found, perform the FFT transformation again.

[0234] This step, through standardized operations such as signal grouping, analysis interval setting, parameter configuration, individual FFT transformation, process parameter recording, and spectrum quality inspection, avoids the spectrum leakage and interference problems caused by transforming the entire frequency band signal at the same time in existing technologies. It enables the harmonic signals of each effective frequency band to fully exhibit their spectral characteristics in an independent transformation process, creating favorable conditions for the accurate extraction of broadband harmonic parameters of offshore wind turbine MMC systems.

[0235] (4) Preliminary extraction and verification of harmonic parameters

[0236] To obtain preliminary information on the frequency, amplitude, and phase of harmonics and ensure data reliability, follow these seven steps:

[0237] 1) Define the methods for extracting harmonic parameters: frequency extraction uses peak detection method, amplitude extraction uses peak amplitude method and converts it to effective value, and phase extraction is based on phase spectrum reading;

[0238] 2) Perform peak detection on the spectrum of each effective frequency band, identify the peak points in the spectrum, and record the frequency values ​​corresponding to the peak points, which is the preliminary result of harmonic frequency extraction;

[0239] 3) Read the amplitude data corresponding to the peak point, and calculate the amplitude based on the conversion relationship between effective value and peak value (effective value = peak value / peak value). The peak values ​​are converted into effective values ​​to obtain preliminary results of harmonic amplitude extraction.

[0240] 4) Based on the phase spectrum of the spectrogram, read the phase angle of the corresponding harmonic frequency to obtain the preliminary harmonic phase extraction result;

[0241] 5) Verify the initially extracted parameters. Frequency verification ensures that the extracted harmonic frequencies are within the effective frequency band. Amplitude verification ensures that the amplitude data conforms to the actual amplitude range of the harmonics in the offshore wind turbine MMC system (by comparing historical data with theoretical calculations). Phase verification ensures that the phase angle is within the range of 0~360°.

[0242] 6) If abnormal parameters are found, such as frequency exceeding the frequency band range, amplitude significantly deviating from the normal range, or phase angle being abnormal, repeat the FFT transformation or adjust the extraction algorithm until the parameters meet the verification requirements.

[0243] 7) Classify and organize the qualified preliminary parameters according to harmonic frequency to form a preliminary harmonic parameter list.

[0244] This step, through methods such as extraction, preliminary parameter extraction, multi-dimensional verification, and anomaly handling, yielded reliable preliminary results for harmonic parameters, providing high-quality foundational data for subsequent parameter optimization and integration.

[0245] (5) Harmonic parameter optimization and integration output

[0246] To improve the stability and accuracy of harmonic parameters and generate a complete parameter list, the following eight steps should be followed:

[0247] 1) Extract the preliminary harmonic parameters that have passed the verification of each effective frequency band, classify them according to harmonic frequency, and group harmonic parameters of the same frequency together;

[0248] 2) For frequency parameters, statistical analysis methods are used to compare the extraction results of the same harmonic frequency in different effective frequency bands, calculate the average frequency and standard deviation, remove outliers exceeding 3 times the standard deviation, and then recalculate the average value to control the frequency identification error within ±5Hz.

[0249] 3) For the amplitude parameter, a weighted calculation is performed based on the energy proportion of each effective frequency band, with a higher weight for the frequency band with a higher energy proportion. The weighting coefficient is calculated as: (Energy proportion of the frequency band / Sum of the energy proportions of all frequency bands containing the harmonic frequency). The final harmonic amplitude is obtained through weighted averaging, controlling the amplitude identification error within ±2%.

[0250] 4) For phase parameters, the phase consistency verification method is used to analyze the phase change trend of the same harmonic frequency in different effective frequency bands, eliminate abnormal data with abrupt phase changes, calculate the average value of the remaining phase data, and ensure the rationality of the phase parameters.

[0251] 5) Perform consistency verification on all optimized harmonic parameters to ensure the matching between frequency, amplitude, and phase parameters (based on the mathematical relationship of harmonic signals).

[0252] 6) Integrate all optimized effective frequency band harmonic parameters and arrange them in ascending order of frequency to form a complete list of 2~2500Hz wideband harmonic parameters for offshore wind turbine MMC system;

[0253] 7) Standardize the format of the parameter list, and specify the units, precision, and other information of the parameters;

[0254] 8) Store the standardized parameter list in a temporary database to provide comprehensive and accurate parameter support for the calculation of total harmonic distortion rate.

[0255] This step, through systematic operations such as parameter classification, frequency optimization, amplitude weighted calculation, phase consistency verification, parameter consistency verification, integration sorting, format standardization, and storage, significantly improves the stability and accuracy of harmonic parameters, laying a solid foundation for subsequent harmonic index calculations.

[0256] Step S5: Quantitative Output Integration of Harmonic Indicators

[0257] The harmonic parameters extracted from all effective frequency bands are integrated to form a complete broadband harmonic parameter list, and the total harmonic distortion rate is calculated.

[0258] Based on the extracted harmonic parameters, the power quality assessment indicators are quantified. This step follows the logic of "parameter classification - formula adaptation - step-by-step calculation - result verification - standardized output" to quantify and store the extracted results. Each step provides complete parameters for the next, ensuring the accuracy of the indicator calculation and the usability of the results. Figure 6 As shown, the implementation steps are as follows.

[0259] (1) Classification and organization of fundamental harmonic parameters

[0260] To clearly distinguish between fundamental and harmonic parameters and facilitate the calculation of total harmonic distortion (THD), follow these six steps:

[0261] 1) From the integrated list of harmonic parameters, filter out the fundamental parameters. The fundamental frequency is fixed at 50Hz. Extract the corresponding effective value of the fundamental voltage (U1).

[0262] 2) Filter out the parameters of all integer harmonics in the range of 2~2500Hz, including harmonic frequency and RMS voltage (U). h and phase;

[0263] 3) Mark the fundamental frequency parameters separately and set a dedicated identifier field to ensure that there is no confusion between the fundamental frequency and harmonic parameters;

[0264] 4) Sort the harmonic parameters in ascending order of harmonic order, based on the harmonic frequency (harmonic order = harmonic frequency / 50Hz).

[0265] 5) Establish a parameter classification index, and construct a fundamental wave parameter index and a harmonic parameter index (indexed by harmonic order) to improve the efficiency of parameter lookup and retrieval;

[0266] 6) Store the categorized and organized fundamental and harmonic parameters in a dedicated database table to ensure data security and traceability.

[0267] This step, through fundamental parameter filtering, harmonic parameter filtering, fundamental labeling, harmonic sorting, index construction, and categorized storage, achieves clear distinction and orderly management of fundamental and harmonic parameters, laying a clear parameter foundation for the calculation of total harmonic distortion rate and ensuring the smoothness of the calculation process.

[0268] (2) Verification of the compatibility of the THD formula

[0269] To ensure that the Total Harmonic Distortion (THD) calculation formula accurately reflects the harmonic distortion level of the offshore wind turbine MMC system, the following six steps are followed:

[0270] 1) Clearly adopt the standard THD calculation formula ,in U1 is the sum of the squares of the effective values ​​of each harmonic voltage, and U2 is the effective value of the fundamental voltage.

[0271] 2) Analyze the characteristics of broadband harmonics in offshore wind turbine MMC systems, including multiple harmonic orders and uneven amplitude distribution, and verify that the formula can be adapted to the distortion rate calculation of such harmonics.

[0272] 3) Select multiple sets of simulated data with different harmonic distortion levels, substitute them into the formula for calculation, compare the calculation results with the actual distortion level, and verify the calculation accuracy of the formula.

[0273] 4) Select actual operating data of multiple sets of offshore wind turbine MMC systems, use the formula to calculate the THD value, and combine it with other harmonic evaluation indicators (such as single harmonic distortion rate) to verify the rationality of the formula calculation results.

[0274] 5) Clearly define the value range and calculation accuracy requirements of each parameter in the formula, U1 and U... h The calculation precision is retained to three decimal places, and the calculation precision of the sum of squares and square root is retained to four decimal places to ensure the accuracy of the THD calculation results;

[0275] 6) Establish a scope of application for the formula, clarifying that the formula is applicable to the calculation of the total distortion rate of 2~2500Hz broadband harmonics in the MMC system of offshore wind turbines, providing a clear formula basis for subsequent THD calculations.

[0276] This step, through formula determination, adaptability analysis, simulation data verification, actual data verification, clarification of accuracy requirements, and explanation of applicable scope, ensures the rationality and effectiveness of the THD calculation formula, providing a guarantee for the accurate calculation of total harmonic distortion rate.

[0277] (3) THD step-by-step calculation error control

[0278] To achieve accurate THD calculation and effectively control calculation errors, follow these seven steps:

[0279] 1) Extract the effective value of the fundamental voltage (U1) and the effective values ​​of each harmonic voltage (U) after classification and sorting. h All parameters are checked for consistency to ensure that there are no missing parameters or outliers.

[0280] 2) Calculate the square of the effective value of each harmonic voltage, for each U h To perform squaring operations, a high-precision calculation method is used to ensure that the error in squaring calculations is controlled within 0.01%.

[0281] 3) Sum the squares of all harmonic voltage effective values ​​using an accumulator, performing an error check after each summation to avoid calculation errors caused by data overflow;

[0282] 4) Regarding the summation result ( To perform square root calculations, a high-precision square root algorithm is used to ensure that the calculation error is controlled within 0.01%.

[0283] 5) The total effective value of harmonic voltage ( The initial THD value is obtained by comparing the ratio of the fundamental voltage effective value (U1) with the fundamental voltage effective value (U1).

[0284] 6) Correct the initial THD value for error. Combine the calculation errors of the previous steps and fine-tune the result to ensure that the error of the final THD value is controlled within 0.1%.

[0285] 7) Keep two decimal places for the THD value to form the final total harmonic distortion result.

[0286] This step, through a complete process including parameter verification, square operation error control, summation operation error control, square root operation error control, ratio calculation, error correction, and result formatting, achieves accurate THD calculation, effectively controls calculation errors in each step, and ensures that the calculation results can truly reflect the power quality status of the offshore wind turbine MMC system.

[0287] (4) Verification of the integrity of extraction results

[0288] To ensure the comprehensiveness, accuracy, and rationality of the broadband harmonic extraction results, the following seven steps are followed:

[0289] 1) Clearly define the verification content for the completeness of the extracted results, including the completeness of harmonic parameters, the accuracy of THD calculation, the reasonableness of the results, and the consistency of the data;

[0290] 2) Harmonic parameter integrity verification: Check the frequency, amplitude, and phase parameters of all integer harmonics in the range of 2 to 2500 Hz one by one to ensure that no parameters are missing;

[0291] 3) Verify the accuracy of THD calculations. Recalculate the THD value using different calculation methods (such as step-by-step manual calculation or third-party software calculation), compare the consistency of the results, and ensure that the calculation results are error-free.

[0292] 4) Verification of the reasonableness of the results: Combined with the operating conditions of the offshore wind turbine MMC system (such as wind speed and load), the harmonic extraction results under similar historical operating conditions were compared to verify that the results of this test are consistent with the actual operating conditions and there are no obvious abnormalities.

[0293] 5) Data consistency verification: Check the mathematical relationship between harmonic parameters and THD values ​​to ensure that it conforms to the logic of the THD calculation formula and that there are no contradictions.

[0294] 6) Establish a verification report, which records in detail the verification content, verification method, verification results, and handling of any anomalies;

[0295] 7) If problems are found during the verification, return to the corresponding step for correction, such as supplementing missing harmonic parameters, recalculating THD values, etc., until the extracted results meet the requirements of completeness, accuracy and rationality.

[0296] This step, through multi-dimensional verification content setting, parameter integrity check, calculation accuracy verification, result rationality analysis, data consistency verification, verification report generation, and closed-loop correction, ensures the high quality of the extracted results and provides a reliable basis for power quality assessment.

[0297] (5) Standardized result output and storage

[0298] To achieve efficient sharing and long-term traceability of extraction results, follow these eight steps:

[0299] 1) Clearly define the content of the standardized output, including key information such as the harmonic parameter list (frequency, amplitude, phase), THD value, extraction time, operating conditions (wind speed, load, unit status), sensor information, preprocessing parameters, and decomposition parameters;

[0300] 2) Determine the output format, adopt a common data format (such as Excel, CSV), and also support JSON format output to facilitate subsequent use and analysis by power quality assessment software;

[0301] 3) Design the naming rules for the output files, including information such as the unit number, extraction date, and extraction time, to ensure the uniqueness and identifiability of the files;

[0302] 4) Generate an output file from the extracted results according to the set format and naming rules, and perform an integrity check on the file to ensure that there is no missing data or format errors;

[0303] 5) Establish a dedicated database for harmonic extraction results, and adopt a distributed storage architecture to ensure the security and scalability of data storage;

[0304] 6) Store the output files and related metadata in the database, and classify them according to extraction time, unit number, operating conditions, etc.

[0305] 7) Establish a data retrieval index to support data queries based on multiple conditions (such as time range, unit number, THD value range) and improve data retrieval efficiency;

[0306] 8) Develop data backup and archiving strategies, regularly back up the data in the database, and archive historical data that exceeds a certain number of years to ensure long-term traceability of the data.

[0307] This step, through systematic operations including clear output content, defined format, standardized file naming, file generation and inspection, database construction, categorized storage, index creation, and backup archiving, achieves standardized output and storage of extracted results. It provides comprehensive and systematic data support for power quality assessment, harmonic control, and long-term operation monitoring of offshore wind turbine MMC systems, thereby promoting the improvement of power quality management and control in the offshore wind power industry.

[0308] This embodiment also provides an electronic device, including: at least one processor, at least one memory, and computer program instructions stored in the memory, which implement the above-described method when executed by the processor.

[0309] This embodiment also provides a computer-readable storage medium storing computer program instructions that, when executed by a processor, implement the above-described method.

[0310] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0311] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0312] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0313] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0314] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A method for collaborative extraction of the db4 fundamental fifth-order WPT-high-resolution FFT of broadband harmonics in an offshore wind turbine MMC system, characterized in that, Includes the following steps: Step S1: Acquire the raw time domain signal of the offshore wind turbine MMC system and preprocess the raw time domain signal, including windowing and leakage suppression, DC stripping and trend elimination, to obtain a clean signal; Step S2: Select the db4 wavelet as the base wavelet and perform fifth-order wavelet packet decomposition on the clean signal to divide the wideband into multiple sub-bands; Step S3: Calculate the energy percentage of each sub-band, and select sub-bands with an energy percentage greater than or equal to the preset energy percentage threshold as effective frequency bands; Step S4: Perform a Fast Fourier Transform on the signal of each selected effective frequency band to extract the harmonic parameters in each effective frequency band. The harmonic parameters include frequency, amplitude and phase. Step S5: Integrate the harmonic parameters extracted from all effective frequency bands to form a complete broadband harmonic parameter list, and calculate the total harmonic distortion rate.

2. The method for collaborative extraction of the db4 fundamental fifth-order WPT-high-resolution FFT of broadband harmonics in offshore wind turbine MMC systems according to claim 1, characterized in that, Step S1 specifically includes: Targeted data collection is achieved by optimizing monitoring points, sampling frequency, and data collection duration. The Hanning window was selected and the window length was determined according to the acquisition parameters. The signal was then windowed to suppress spectral leakage. The DC component in the signal is calculated and removed using the mean method. The least squares method is used to fit and eliminate the trend term in the signal; The preprocessed signal is quality checked to ensure that the signal-to-noise ratio and smoothness meet the standards.

3. The method for collaborative extraction of the db4 fundamental fifth-order WPT-high-resolution FFT of broadband harmonics in offshore wind turbine MMC systems according to claim 2, is characterized in that, In the windowing process, the length of the Hanning window is determined based on the product of the sampling frequency and the acquisition duration, achieving a perfect match between the window function and the signal duration; in the DC stripping process, the mean method obtains the DC component by summing with high precision and dividing by the total number of data points, and then subtracts it point by point; in the trend elimination process, a first-order or second-order trend term model is selected based on the signal change characteristics, and the least squares fitting parameters are solved iteratively.

4. The method for collaborative extraction of the db4 fundamental fifth-order WPT-high-resolution FFT of broadband harmonics in offshore wind turbine MMC systems according to claim 1, characterized in that, Step S2 specifically includes: Comparative analysis was used to verify the adaptability of the db4 wavelet to broadband non-stationary harmonic signals. The decomposition order is set to fifth order, and the 2~2500Hz wideband is divided into 16 equal-width sub-bands; The decomposition process is monitored in real time to ensure that there is no shift in the frequency range; Verify the validity of the decomposition results, including frequency coverage integrity and non-overlap. The decomposed sub-band data is stored in a standardized manner.

5. The method for collaborative extraction of the db4 fundamental fifth-order WPT-high-resolution FFT of broadband harmonics in offshore wind turbine MMC systems according to claim 1, characterized in that, Step S3 specifically includes: An energy calculation model is constructed using the root mean square method to calculate the total energy of each sub-band. The total energy of all sub-bands is summed and normalized to obtain the energy percentage of each sub-band. Through multi-condition analysis and effect evaluation, the energy percentage threshold was determined to be 0.5%. Sub-bands with an energy percentage of less than 0.5% are marked as noise bands and removed, while sub-bands with an energy percentage of ≥0.5% are retained as effective bands. The screening results are verified to ensure that the main harmonic energy is retained.

6. The method for collaborative extraction of the db4 fundamental fifth-order WPT-high-resolution FFT of broadband harmonics in offshore wind turbine MMC systems according to claim 5, is characterized in that, The energy percentage threshold is 0.5%. This threshold is determined by collecting multi-condition operating data, analyzing energy distribution patterns, constructing an evaluation index system that includes effective harmonic energy retention rate and noise rejection rate, and verifying through multi-threshold comparison, so that the effective harmonic energy retention rate is not less than 95% and the noise rejection rate is not less than 80%.

7. The method for collaborative extraction of the db4 fundamental fifth-order WPT-high-resolution FFT of broadband harmonics in offshore wind turbine MMC systems according to claim 1, characterized in that, Step S4 specifically includes: The number of sampling points for the Fast Fourier Transform is set to 32,768, so that the frequency resolution reaches 0.25Hz; The effective frequency band signal is resampled to match the number of data points to the number of sampling points; Perform a fast Fourier transform independently on each effective frequency band to generate its own spectrum. The peak detection method is used to initially extract the frequency and amplitude from the spectrum, and the phase is extracted from the phase spectrum. The initially extracted parameters are validated, and outliers are removed.

8. The method for collaborative extraction of the db4 fundamental fifth-order WPT-high-resolution FFT of broadband harmonics in offshore wind turbine MMC systems according to claim 7, is characterized in that, The number of sampling points for the Fast Fourier Transform is set to 32,768, which is calculated based on the frequency resolution requirement and the sampling frequency according to the relationship between frequency resolution = sampling frequency / number of sampling points. It is verified by comparison with 16,384 and 65,536 sampling points to achieve an optimal balance between frequency identification accuracy, amplitude calculation error and calculation time.

9. The method for collaborative extraction of the db4 fundamental fifth-order WPT-high-resolution FFT of broadband harmonics in offshore wind turbine MMC systems according to claim 7, is characterized in that, The extraction accuracy of the harmonic parameters is optimized in the following ways: for the frequency parameter, the identification error is controlled within ±5Hz by statistically averaging the results of multiple extractions of the same frequency and removing outliers; for the amplitude parameter, the identification error is controlled within ±2% by weighted averaging by introducing the energy proportion of the effective frequency band as a weight.

10. The method for collaborative extraction of the db4 fundamental fifth-order WPT-high-resolution FFT of broadband harmonics in offshore wind turbine MMC systems according to claim 1, characterized in that, Step S5 specifically includes: The extracted harmonic parameters are classified and organized according to fundamental frequency and harmonics; Verify the adaptability of the total harmonic distortion rate calculation formula to broadband harmonics in MMC systems; A high-precision calculation method is used to calculate the total harmonic distortion rate step by step to control the error in each stage; The completeness, accuracy, and reasonableness of the extracted results are verified from multiple dimensions. The final results are output and stored in a standardized format.

Images