Fan blade operation anomaly detection method, device, equipment, medium and product

Abstract

The embodiment of the invention discloses a fan blade operation anomaly detection method and device, equipment, a medium and a product, and the method comprises the steps: obtaining a to-be-analyzed audio generated during the operation of a target fan blade, and generating a to-be-analyzed spectrogram of the to-be-analyzed audio, and a to-be-analyzed average spectrogram corresponding to the to-be-analyzed spectrogram; segmenting the to-be-analyzed average spectrogram according to a preset time interval to obtain a to-be-analyzed sub-average spectrogram sequence; and calculating the similarity of adjacent to-be-analyzed sub-average spectrograms in the to-be-analyzed sub-average spectrogram sequence to obtain a similarity calculation result sequence, and determining an anomaly detection result of the to-be-analyzed audio according to the similarity calculation result sequence and a preset anomaly detection threshold. According to the technical scheme provided by the embodiment of the invention, the problem that the fan blade is difficult to perform anomaly detection at present is solved, the anomaly detection result can be obtained by automatically analyzing the audio in the running process of the fan blade, and the anomaly detection efficiency, accuracy and reliability are improved.

Description

Technical Field

[0001] Embodiments of the present invention relate to the technical field of equipment abnormality detection, and in particular, to a method, device, equipment, medium and product for detecting abnormal operation of fan blades. Background Art

[0002] As the global demand for renewable energy increases, wind power generation has been widely used as a clean energy source. With the continuous advancement of wind power generation technology, the scale and number of wind turbines have grown rapidly. The blades of wind turbines are key components, and their performance directly affects the efficiency and safety of the entire power generation system. During operation, the blades are affected by various factors such as wind, ice and snow, and lightning, and are prone to damage such as cracks, wear, and debonding. Blade damage will not only lead to a decrease in power generation efficiency, but may also cause serious safety accidents. Timely and accurate detection and diagnosis of blade damage is crucial to ensuring the normal operation of wind turbines and extending their service life. Traditional manual inspection methods have the disadvantages of high cost, low efficiency, and difficulty in comprehensive coverage. More advanced technical means are urgently needed to achieve online monitoring and automatic diagnosis. Summary of the invention

[0003] The embodiments of the present invention provide a method, device, equipment, medium and product for detecting abnormal operation of fan blades. The abnormal detection results can be obtained by automatically analyzing the audio during the operation of the fan blades, thereby improving the efficiency, accuracy and reliability of abnormal detection.

[0004] In a first aspect, an embodiment of the present invention provides a method for detecting abnormal operation of a fan blade, the method comprising:

[0005] Acquire the audio to be analyzed generated by the target fan blade during operation, and generate a spectrogram to be analyzed corresponding to the audio to be analyzed;

[0006] Determine the segmentation point sequence coordinates of the sweeping and sweeping intervals of the target fan blades during operation according to the spectrogram to be analyzed, segment the spectrogram to be analyzed according to the segmentation point sequence coordinates, and obtain the spectrogram segmentation result;

[0007] Perform transparency superposition on the spectrogram segmentation results to generate an average spectrogram to be analyzed corresponding to the spectrogram to be analyzed;

[0008] The average spectrogram to be analyzed is segmented according to a preset time interval to obtain a sequence of sub-average spectrograms to be analyzed;

[0009] The similarities of adjacent sub-average spectrograms to be analyzed in the sub-average spectrogram sequence to be analyzed are calculated to obtain a similarity calculation result sequence, and the anomaly detection result of the audio to be analyzed is determined according to the similarity calculation result sequence and a preset anomaly detection threshold.

[0010] In a second aspect, an embodiment of the present invention provides a device for detecting abnormal operation of a fan blade, the device comprising:

[0011] A spectrogram generation module is used to obtain the audio to be analyzed generated by the target fan blade during operation, and generate a spectrogram to be analyzed corresponding to the audio to be analyzed;

[0012] A spectrogram segmentation module is used to determine the segmentation point sequence coordinates of the sweeping and sweeping intervals of the target fan blades during operation according to the spectrogram to be analyzed, and to segment the spectrogram to be analyzed according to the segmentation point sequence coordinates to obtain the spectrogram segmentation result;

[0013] The average spectrogram generation module is used to perform transparency superposition on the spectrogram segmentation results to generate an average spectrogram to be analyzed corresponding to the spectrogram to be analyzed;

[0014] A sub-average spectrogram generation module is used to segment the average spectrogram to be analyzed according to a preset time interval to obtain a sequence of sub-average spectrograms to be analyzed;

[0015] The anomaly detection result determination module is used to calculate the similarity of adjacent sub-average spectrograms to be analyzed in the sub-average spectrogram sequence to be analyzed, obtain a similarity calculation result sequence, and determine the anomaly detection result of the audio to be analyzed based on the similarity calculation result sequence and a preset anomaly detection threshold.

[0016] In a third aspect, an embodiment of the present invention further provides a computer device, the computer device comprising:

[0017] one or more processors;

[0018] A memory for storing one or more programs;

[0019] When the above one or more programs are executed by one or more processors, the above one or more processors implement the method for detecting abnormal operation of wind turbine blades as provided in any embodiment of the present invention.

[0020] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a method for detecting abnormal operation of a wind turbine blade as provided in any embodiment of the present invention.

[0021] In a fifth aspect, an embodiment of the present invention further provides a computer program product, including a computer program, which, when executed by a processor, implements a method for detecting abnormal operation of a wind turbine blade as provided in any embodiment of the present invention.

[0022] The embodiments of the above invention have the following advantages or beneficial effects:

[0023] The embodiment of the present invention obtains the audio to be analyzed generated by the target fan blade during operation, and generates a spectrogram to be analyzed corresponding to the audio to be analyzed; determines the segmentation point sequence coordinates of the sweep and sweep interval of the target fan blade during operation according to the spectrogram to be analyzed, and segments the spectrogram to be analyzed according to the segmentation point sequence coordinates to obtain the spectrogram segmentation result; performs transparency superposition on the spectrogram segmentation result to generate an average spectrogram to be analyzed corresponding to the spectrogram to be analyzed; segments the average spectrogram to be analyzed according to a preset time interval to obtain a sequence of sub-average spectrograms to be analyzed; calculates the similarity of adjacent sub-average spectrograms to be analyzed in the sequence of sub-average spectrograms to be analyzed to obtain a sequence of similarity calculation results, and determines the abnormality detection result of the audio to be analyzed according to the sequence of similarity calculation results and a preset abnormality detection threshold. The technical solution of the embodiment of the present invention solves the problem that it is currently difficult to detect abnormalities in fan blades, and can obtain abnormality detection results by automatically analyzing the audio during the operation of the fan blades, thereby improving the efficiency, accuracy and reliability of abnormality detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 It is a flow chart of a method for detecting abnormal operation of fan blades provided by an embodiment of the present invention;

[0025] Figure 2 It is a flow chart of a method for detecting abnormal operation of fan blades provided by an embodiment of the present invention;

[0026] Figure 3 It is a flow chart of a method for detecting abnormal operation of fan blades provided by an embodiment of the present invention;

[0027] Figure 4 It is a flow chart of a method for detecting abnormal operation of fan blades provided by an embodiment of the present invention;

[0028] Figure 5 It is a structural schematic diagram of a fan blade operation abnormality detection device provided by an embodiment of the present invention;

[0029] Figure 6 It is a structural schematic diagram of a computer device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0030] The present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the present invention, rather than to limit the present invention. It should also be noted that, for ease of description, only parts related to the present invention, rather than all structures, are shown in the accompanying drawings.

[0031] Figure 1A flowchart of a method for detecting abnormal operation of a fan blade is provided in an embodiment of the present invention. This embodiment is applicable to the scenario of detecting abnormal operation of a power fan blade. The method can be performed by a fan blade abnormal operation detection device, which can be implemented by software and / or hardware and integrated into a computer device with application development function.

[0032] like Figure 1 As shown, the fan blade operation abnormality detection method of this embodiment includes the following steps:

[0033] S110, obtaining the audio to be analyzed generated by the target wind turbine blade during operation, and generating a spectrogram to be analyzed corresponding to the audio to be analyzed.

[0034] The target wind turbine blade can be any blade of a wind turbine generator set. The operation of the blade converts wind energy into mechanical energy, thereby driving the generator to generate electricity. However, in actual application, the blade may become abnormal due to various reasons. Therefore, in this embodiment, a sound collector is set to collect the audio of the wind turbine blade operation in real time as the audio to be analyzed, and a spectrogram of the audio to be analyzed is generated by a spectrogram generation technology.

[0035] S120, determining the segmentation point sequence coordinates of the sweep and sweep interval of the target fan blade during operation according to the spectrogram to be analyzed, and segmenting the spectrogram to be analyzed according to the segmentation point sequence coordinates to obtain the spectrogram segmentation result.

[0036] In this embodiment, an average spectrogram is generated after the spectrogram is segmented. The average spectrogram has stronger signals and weaker noises, and can more sensitively reflect abnormal situations, thereby improving the sensitivity of abnormality detection.

[0037] For example, cluster analysis is used to determine the segmentation point sequence coordinates of the sweeping and sweeping intervals of the target fan blade during operation. The cluster analysis may be a hierarchical clustering algorithm and a K-means algorithm, etc., which are not limited in this embodiment. The spectrogram to be analyzed is segmented according to the segmentation point sequence coordinates to obtain a spectrogram segmentation result, which may be a sub-spectrogram sequence.

[0038] S130, performing transparency superposition on the spectrogram segmentation results to generate an average spectrogram to be analyzed corresponding to the spectrogram to be analyzed.

[0039] The transparency of the sub-spectrogram sequence is superimposed, and the superposition calculation formula is as follows:

[0040]

[0041] Where t represents the number of segmentation coordinates, T i represents the i-th sub-spectrogram, p represents transparency, T col1 ,T col1 ,Tcol1 Matrix representation of the averaged spectrogram.

[0042] S140, dividing the average spectrogram to be analyzed according to preset time intervals to obtain a sequence of sub-average spectrograms to be analyzed.

[0043] The preset time interval may be a few minutes, a few seconds, etc., for example, 3 seconds. The average spectrogram to be analyzed is segmented according to the preset time interval to obtain a sequence of sub-average spectrograms to be analyzed in chronological order, representing the same time interval.

[0044] S150, calculating similarities of adjacent sub-average spectrograms to be analyzed in the sub-average spectrogram sequence to be analyzed, obtaining a similarity calculation result sequence, and determining anomaly detection results of the audio to be analyzed according to the similarity calculation result sequence and a preset anomaly detection threshold.

[0045] By analyzing the changes of the adjacent sub-average spectrograms to be analyzed in time, the abnormality detection result of the audio to be analyzed is determined according to the changes. In this embodiment, the changes of the adjacent sub-average spectrograms to be analyzed in the sub-average spectrogram sequence to be analyzed are determined by calculating the similarity of the adjacent sub-average spectrograms to be analyzed. The similarity can be MSSIM (Mean Structural Similarity Index Measurement).

[0046] The preset abnormality detection threshold may include a normal threshold under normal operation and an abnormal threshold of at least one abnormal type. The normal threshold and the thresholds of various abnormal types are substituted into the evaluation function to obtain the evaluation function of each state, the evaluation function is used to obtain the evaluation value of each state corresponding to the similarity calculation result, and then the probability of the similarity calculation result relative to each state is obtained by importing it into the probability function, so as to judge whether the state of the fan blade is normal or abnormal, and the abnormal type.

[0047] The technical solution of this embodiment is to obtain the audio to be analyzed generated by the target fan blade during operation, and generate a spectrogram to be analyzed corresponding to the audio to be analyzed; determine the segmentation point sequence coordinates of the sweep and sweep interval of the target fan blade during operation according to the spectrogram to be analyzed, segment the spectrogram to be analyzed according to the segmentation point sequence coordinates, and obtain the spectrogram segmentation result; perform transparency superposition on the spectrogram segmentation result to generate an average spectrogram to be analyzed corresponding to the spectrogram to be analyzed; segment the average spectrogram to be analyzed according to a preset time interval to obtain a sequence of sub-average spectrograms to be analyzed; calculate the similarity of adjacent sub-average spectrograms to be analyzed in the sequence of sub-average spectrograms to be analyzed, obtain a sequence of similarity calculation results, and determine the abnormality detection result of the audio to be analyzed according to the similarity calculation result sequence and a preset abnormality detection threshold. The technical solution of the embodiment of the present invention solves the problem that it is currently difficult to detect abnormalities in fan blades, and can obtain abnormality detection results by automatically analyzing the audio during the operation of the fan blades, thereby improving the efficiency, accuracy and reliability of abnormality detection.

[0048] Figure 2 A flowchart of a method for detecting abnormal operation of a fan blade is provided in an embodiment of the present invention. This embodiment and the method for detecting abnormal operation of a fan blade in the above embodiment belong to the same inventive concept, and further describes the process of generating a spectrogram to be analyzed. The method can be executed by a device for detecting abnormal operation of a fan blade, which can be implemented by software and / or hardware and integrated into a computer device with application development function.

[0049] like Figure 2 As shown, the fan blade operation abnormality detection method of this embodiment includes the following steps:

[0050] S210, obtaining the audio to be analyzed generated by the target wind turbine blade during operation, and performing frame processing on the audio to be analyzed according to preset frame parameters to obtain a frame processing result.

[0051] The preset framing parameters may include frame length and frame shift, where frame length is the number of sampling points contained in each frame, and frame shift is the number of sampling points between the starting points of two adjacent frames. Then, starting from the starting position of the audio to be analyzed, a section is intercepted as a frame according to the frame length, and then the next frame is intercepted according to the frame shift until the end of the signal cannot be completely intercepted to obtain the framing processing result.

[0052] In an optional implementation, before generating a spectrogram to be analyzed corresponding to the audio to be analyzed, wind noise in the audio to be analyzed is removed by a Butterworth high-pass filter to obtain the audio to be analyzed with the wind noise removed; the signal incident direction of the audio to be analyzed with the wind noise removed is determined by a sparse Bayesian learning direction finding algorithm; the signal in the signal incident direction is enhanced by a conventional beamforming algorithm, and interference in non-signal incident directions is weakened to obtain signal-enhanced audio to be analyzed.

[0053] In the time domain, a Butterworth high-pass filter is used to remove low-frequency wind noise, and the audio to be analyzed is obtained without wind noise. The formula for generating the Butterworth high-pass filter is:

[0054]

[0055] Where j represents the imaginary unit, ω represents the frequency, and ω c Indicates the cutoff frequency.

[0056] In space, the sparse Bayesian learning direction finding algorithm (SBL) is first used to estimate the signal incident direction, and then the conventional beam forming algorithm (CBF) is used to enhance the signal in the signal incident direction and weaken the interference in other directions to improve the signal-to-noise ratio of the received signal. This embodiment effectively eliminates the interference of environmental noise through filtering and signal enhancement, and enhances the robustness and stability of subsequent data processing.

[0057] S220 , multiply each frame spectrum in the frame processing result by the window function to obtain a windowing processing result.

[0058] The window function is multiplied by the signal spectrum of each frame in the frame processing result to obtain a windowed frame signal.

[0059] S230, performing Fourier transform on the windowing processing result to obtain a Fourier transform result.

[0060] Perform a fast Fourier transform (FFT) on the windowed frame signal to obtain the spectrum of each frame after the Fourier transform.

[0061] S240, calculating the energy value of each frequency component in the Fourier transform result, and mapping the energy value to a color value, to obtain a spectrogram to be analyzed corresponding to the audio to be analyzed.

[0062] For the spectrum of each frame after Fourier transformation, the energy of each frequency component is calculated and mapped to color and plotted into a two-dimensional image, with the horizontal axis representing time, the vertical axis representing frequency, and the color representing energy.

[0063] For example, the audio to be analyzed is divided into several frames with a frame length of 20ms to 40ms and a frame shift of 10ms to 20ms. Each frame is windowed to reduce spectrum leakage. The windowed signal of each frame is subjected to a fast Fourier transform to convert the time domain signal to the frequency domain to obtain the spectrum of each frame. The energy of each spectral component is calculated, and the energy can be expressed as the square of the amplitude of the spectral component. The energy value of each frequency component is mapped to a color value. A two-dimensional image is constructed, with the horizontal axis representing time, that is, the position of each frame, the vertical axis representing frequency, and the color representing energy. The calculated energy value is filled into the corresponding time and frequency position to generate a spectrogram.

[0064] S250, determining the segmentation point sequence coordinates of the sweeping and sweeping intervals of the target fan blades during operation according to the spectrogram to be analyzed, and segmenting the spectrogram to be analyzed according to the segmentation point sequence coordinates to obtain the spectrogram segmentation result.

[0065] According to the clustering analysis algorithm, the distance between the eigenvectors represented by the matrix of the spectrogram to be analyzed is calculated, the audio to be analyzed is divided into puff audio and puff interval audio, the segmentation point sequence coordinates of puff and puff interval are obtained, the spectrogram to be analyzed is segmented according to the segmentation point sequence coordinates, and the spectrogram segmentation result is obtained.

[0066] In an optional embodiment, the coordinate sequence of segmentation points of the sweep and sweep interval of the target fan blades during operation is determined according to the spectrogram to be analyzed, and the spectrogram matrix to be analyzed corresponding to the spectrogram to be analyzed can be determined according to the energy of each frequency component; the characteristic vectors in the spectrogram matrix to be analyzed are clustered to obtain the clustering analysis results; the dimensions that affect the clustering effect are determined according to the clustering analysis results, and the spectrogram matrix to be analyzed is dimensionally reduced according to the dimensions that affect the clustering effect to obtain the reduced-dimensional spectrogram matrix; cluster analysis is performed on the reduced-dimensional spectrogram matrix to obtain a preliminary label sequence of the sweep and sweep interval of the target fan blades during operation; the preliminary label sequence is filtered for abnormal points to obtain the coordinate sequence of segmentation points of the sweep and sweep interval of the target fan blades during operation.

[0067] The spectrogram matrix to be analyzed corresponding to the spectrogram to be analyzed is determined according to the energy of each frequency component. The calculation formula of the spectrogram matrix to be analyzed is as follows:

[0068] S (t,f) =log 10 (|FFT(x (t) *ω)| 2 );

[0069] Among them, t represents time, f represents frequency, ω represents window function, and x (t) Represents the original signal.

[0070] The spectrogram matrix to be analyzed is flattened into a two-dimensional matrix, where each row is a feature vector, representing a time frame, and each column represents a frequency component. The K-means algorithm is used to cluster these time frames into two categories, and the cluster analysis results are obtained, namely the clustering results of the purge audio and the purge interval audio.

[0071] According to the cluster analysis results, the influence index of each dimension on the clustering effect is determined. According to the influence index threshold, the dimensions that need to be retained are determined, and the dimensions that do not need to be retained are intercepted. The medium and high dimensions can be intercepted to retain the meaningful frequency range and reduce the data dimension. The matrix after dimensionality reduction is input into the K-means algorithm again, and the data is clustered into two categories again to obtain the preliminary label sequence of the target fan blades during operation during the sweep and sweep interval.

[0072] The preliminary label sequence after the second K-means clustering is post-processed to filter out abnormal points, i.e. short-term glitches and spikes, by means of sliding window or median filtering. The filtered preliminary label sequence is then smoothed and converged to determine the final segmentation point. Common methods include running average filtering or voting.

[0073] S260, performing transparency superposition on the spectrogram segmentation results to generate an average spectrogram to be analyzed corresponding to the spectrogram to be analyzed.

[0074] In this embodiment, after the average spectrogram to be analyzed is generated, the average spectrogram to be analyzed can also be converted from RGB format to YCbCr format that does not interfere with the color balance of the image and balances the brightness value, and grayscale histogram equalization operation is performed on the intensity plane, and then the YCbCr format is converted back to RGB format. For example, the generated average spectrogram of leaves is converted into an RGB image, the brightness (Y plane) and chrominance (Cb and Cr planes) of the image are separated, and the RGB image is converted into YCbCr format using OpenCV to facilitate independent processing of brightness information. The Y brightness plane is extracted from the YCbCr image, and the brightness information is processed to enhance the image contrast and details. The contrast of the Y plane is enhanced using the histogram equalization method, and the dynamic range of the brightness is stretched, so that the image is visually clearer and more balanced. The processed Y plane is merged with the original Cb and Cr planes, the original chrominance information is maintained, and the brightness contrast is enhanced at the same time. The merged YCbCr image is converted back to RGB format to obtain the final color-balanced and brightness-enhanced image, which is convenient for visualization and further analysis.

[0075] Among them, the calculation formula for converting RGB format to YCbCr format is:

[0076] Y=0.257*R+0.504*G+0.098*B+16;

[0077] Cb=-0.148*R-0.291*G+0.439*B+128;

[0078] Cr=0.439*R-0.368*G-0.071*B+128;

[0079] The grayscale histogram equalization operation is to first use the transformation function to calculate the grayscale level S output after mapping. k , and then the new gray level S after statistical mapping k The number of pixels n k , and finally calculate the histogram of the output image. The calculation formula of the transformation function is:

[0080]

[0081] Where n represents the total number of pixels, n k represents the number of pixels at the kth gray level, p(r k ) represents the number of occurrences of the kth gray level.

[0082] The calculation formula for converting YCbCr format to RGB format is:

[0083] R=1.164*(Y-16)+1.596*(Cr-128);

[0084] G=1.164*(Y-16)-0.813*(Cr-128)-0.392*(Cb-128);

[0085] B=1.164*(Y-16)+2.017*(Cb-128).

[0086] S270 , dividing the average spectrogram to be analyzed according to preset time intervals to obtain a sequence of sub-average spectrograms to be analyzed.

[0087] S280, calculating the similarities of adjacent sub-average spectrograms to be analyzed in the sub-average spectrogram sequence to be analyzed, obtaining a similarity calculation result sequence, and determining an anomaly detection result of the audio to be analyzed according to the similarity calculation result sequence and a preset anomaly detection threshold.

[0088] The technical solution of this embodiment is to obtain the audio to be analyzed generated by the target fan blade during operation, and perform frame processing on the audio to be analyzed according to preset frame parameters to obtain a frame processing result; multiply each frame spectrum in the frame processing result by a window function to obtain a window processing result; perform Fourier transform on the window processing result to obtain a Fourier transform result; calculate the energy value of each frequency component in the Fourier transform result, and map the energy value to a color value to obtain a spectrogram to be analyzed corresponding to the audio to be analyzed; determine the sweeping and coloring of the target fan blade during operation according to the spectrogram to be analyzed The segmentation point sequence coordinates of the sweeping interval are used to segment the spectrogram to be analyzed according to the segmentation point sequence coordinates to obtain the spectrogram segmentation result; the spectrogram segmentation result is subjected to transparency superposition to generate the average spectrogram to be analyzed corresponding to the spectrogram to be analyzed; the average spectrogram to be analyzed is segmented according to the preset time interval to obtain a sequence of sub-average spectrograms to be analyzed; the similarity of adjacent sub-average spectrograms to be analyzed in the sequence of sub-average spectrograms to be analyzed is calculated to obtain a sequence of similarity calculation results, and the abnormality detection result of the audio to be analyzed is determined according to the sequence of similarity calculation results and a preset abnormality detection threshold. The technical solution of the embodiment of the present invention solves the problem that it is currently difficult to detect abnormalities in fan blades. The abnormality detection result can be obtained by automatically analyzing the audio during the operation of the fan blades, especially by converting the audio into a spectrogram through audio processing, and obtaining the abnormality detection result through spectrogram analysis, thereby improving the efficiency, accuracy and reliability of abnormality detection.

[0089] Figure 3 A flowchart of a method for detecting abnormal operation of a fan blade is provided in an embodiment of the present invention. This embodiment and the method for detecting abnormal operation of a fan blade in the above embodiment belong to the same inventive concept, and further describes the process of similarity calculation. The method can be executed by a device for detecting abnormal operation of a fan blade, which can be implemented by software and / or hardware and integrated into a computer device with application development function.

[0090] like Figure 3 As shown, the fan blade operation abnormality detection method of this embodiment includes the following steps:

[0091] S310, obtaining the audio to be analyzed generated by the target wind turbine blade during operation, and generating a spectrogram to be analyzed corresponding to the audio to be analyzed.

[0092] S320, determining the segmentation point sequence coordinates of the sweep and sweep interval of the target fan blade during operation according to the spectrogram to be analyzed, and segmenting the spectrogram to be analyzed according to the segmentation point sequence coordinates to obtain the spectrogram segmentation result.

[0093] S330 , performing transparency superposition on the spectrogram segmentation results to generate an average spectrogram to be analyzed corresponding to the spectrogram to be analyzed.

[0094] S340 , segmenting the average spectrogram to be analyzed according to preset time intervals to obtain a sequence of sub-average spectrograms to be analyzed.

[0095] S350: Divide adjacent sub-average spectrograms to be analyzed in the sequence of sub-average spectrograms to be analyzed into a preset first number of block images of the same size.

[0096] Considering that the effect of local SSIM (Structural Similarity Index) is better than global, SSIM is calculated first and then the average structural similarity MSSIM is calculated based on SSIM. Every two adjacent sub-average spectrograms to be analyzed in the sub-average spectrogram sequence to be analyzed are divided into a sub-average spectrogram group. The sub-average spectrograms in each sub-average spectrogram group are divided into blocks using a sliding window, and the total number of blocks is set to a preset first number, and the preset first number is represented as N.

[0097] S360, calculating the mean, variance and covariance of each block image by Gaussian weighting.

[0098] Taking into account the influence of window shape on blocking, Gaussian weighting is used to calculate the mean, variance and covariance of each window, and then the structural similarity SSIM of the corresponding block is calculated. Finally, the average value of all block images is used as the structural similarity measure of the two images in the sub-average speech spectrum group, namely the average structural similarity MSSIM.

[0099] The calculation formula for the SSIM value obtained by Gaussian weighted calculation is as follows:

[0100] The formula for calculating brightness similarity is:

[0101]

[0102] The formula for calculating contrast similarity is:

[0103]

[0104] The formula for calculating structural similarity is:

[0105]

[0106] Where x represents the block image of the first image in the sub-average spectrogram group, y represents the block image of the second image in the sub-average spectrogram group, and μ x represents the mean of x, μ y represents the mean of y, σ x represents the variance of x, σ y represents the variance of y, σ xy represents the covariance between x and y, c1, c2 and c3 are constants to avoid the denominator being 0. Among them, c1=(k1L)2 , c2=(k2L) 2 , The default value of k1 is 0.01, the default value of k2 is 0.03, and L is the pixel value range. The above three comparison functions constitute the SSIM function.

[0107] S370, calculating the structural similarity of the corresponding block image groups in the adjacent sub-average spectrograms to be analyzed according to the mean, variance and covariance.

[0108] The structural similarity of the corresponding block image groups is:

[0109]

[0110] S380, calculating the average structural similarity of all block image groups, taking the average as the similarity of adjacent sub-average spectrograms to be analyzed, obtaining a similarity calculation result sequence, and determining the anomaly detection result of the audio to be analyzed based on the similarity calculation result sequence and a preset anomaly detection threshold.

[0111] The average structural similarity of all block image groups is calculated to obtain the average structural similarity MSSIM, and MSSIM is used as the similarity of every two adjacent sub-average spectrograms to be analyzed. The anomaly detection result of the audio to be analyzed is determined based on the similarity calculation result sequence, the preset anomaly detection threshold and the evaluation function.

[0112] The technical solution of this embodiment is to obtain the audio to be analyzed generated by the target fan blades during operation, and generate a spectrogram to be analyzed corresponding to the audio to be analyzed; determine the segmentation point sequence coordinates of the sweep and sweep interval of the target fan blades during operation according to the spectrogram to be analyzed, segment the spectrogram to be analyzed according to the segmentation point sequence coordinates, and obtain the spectrogram segmentation result; perform transparency superposition on the spectrogram segmentation result to generate an average spectrogram to be analyzed corresponding to the spectrogram to be analyzed; segment the average spectrogram to be analyzed according to a preset time interval to obtain a sequence of sub-average spectrograms to be analyzed; and separate the sub-average spectrograms to be analyzed into a sequence of sub-average spectrograms to be analyzed. The adjacent sub-average spectrograms to be analyzed in the spectrogram sequence are divided into a preset first number of block images of the same size; the mean, variance and covariance of each block image are calculated by Gaussian weighting; the structural similarity of the corresponding block image group in the adjacent sub-average spectrograms to be analyzed is calculated according to the mean, variance and covariance; the average value of the structural similarities of all block image groups is calculated, and the average value is used as the similarity of the adjacent sub-average spectrograms to be analyzed to obtain a similarity calculation result sequence, and the abnormal detection result of the audio to be analyzed is determined according to the similarity calculation result sequence and a preset abnormal detection threshold. The technical solution of the embodiment of the present invention solves the problem that it is currently difficult to detect abnormalities in fan blades. The abnormal detection result can be obtained by automatically analyzing the audio during the operation of the fan blades, determining the change of the audio according to the average structural similarity, and improving the efficiency, accuracy and reliability of abnormal detection.

[0113] Figure 4 A flowchart of a method for detecting abnormal operation of a fan blade is provided in an embodiment of the present invention. This embodiment and the method for detecting abnormal operation of a fan blade in the above embodiment belong to the same inventive concept, and further describes the process of determining the abnormal detection result. The method can be executed by a fan blade abnormal operation detection device, which can be implemented by software and / or hardware and integrated into a computer device with application development function.

[0114] like Figure 4 As shown, the fan blade operation abnormality detection method of this embodiment includes the following steps:

[0115] S410, obtaining the audio to be analyzed generated by the target wind turbine blade during operation, and generating a spectrogram to be analyzed corresponding to the audio to be analyzed.

[0116] S420, determining the segmentation point sequence coordinates of the sweeping and sweeping intervals of the target fan blades during operation according to the spectrogram to be analyzed, and segmenting the spectrogram to be analyzed according to the segmentation point sequence coordinates to obtain the spectrogram segmentation result.

[0117] S430, performing transparency superposition on the spectrogram segmentation results to generate an average spectrogram to be analyzed corresponding to the spectrogram to be analyzed.

[0118] S440, dividing the average spectrogram to be analyzed according to preset time intervals to obtain a sequence of sub-average spectrograms to be analyzed.

[0119] S450, calculating similarities of adjacent sub-average spectrograms to be analyzed in the sub-average spectrogram sequence to be analyzed, and obtaining a similarity calculation result sequence.

[0120] S460: According to the preset abnormality detection threshold and the preset evaluation function, a target evaluation function corresponding to each preset abnormality detection threshold is obtained.

[0121] The preset abnormality detection threshold includes a normal operation threshold and an abnormal operation threshold of at least one abnormal type. There are at least two target evaluation functions, including a target evaluation function corresponding to the normal operation threshold and a target evaluation function corresponding to the abnormal operation threshold of at least one abnormal type.

[0122] The blade operation status is judged according to the sequence of similarity calculation results, and the calculated MSSIM value is used to judge whether the blade is operating normally. The normal operation is the highest probability after the MSSIM value sequence is evaluated. The abnormal operation is the highest probability after the MSSIM value sequence is evaluated.

[0123] The preset evaluation function can be expressed by the following formula:

[0124]

[0125] Where i represents the number of thresholds, ω i represents the i-th threshold, and x represents the MSSIM value in the MSSIM value sequence.

[0126] Substitute the threshold value under normal operation and the threshold value of various abnormal states into the evaluation function to obtain the target evaluation function of each state.

[0127] S470 , substituting each similarity calculation result in the similarity calculation result sequence into at least two target evaluation functions to obtain at least two evaluation values ​​for each similarity calculation result.

[0128] The evaluation function is used to obtain each similarity calculation result, that is, the evaluation value of each state corresponding to each MSSIM value to be evaluated, that is, the evaluation value corresponding to each target evaluation function.

[0129] S480. Calculate probability values ​​corresponding to the evaluation values ​​according to the evaluation values, and determine the maximum probability value among the probability values.

[0130] The evaluation value is introduced into the probability function to obtain the probability of the MSSIM value to be evaluated relative to each state, so as to judge the state of the wind turbine blade.

[0131] The probability function can be expressed by the following formula:

[0132]

[0133] Where: r i : The evaluation value of the i-th objective evaluation function corresponding to the MSSIM value to be evaluated.

[0134] S490: When there are a continuous preset second number of maximum probability values ​​corresponding to the same target evaluation function, determine a preset anomaly detection threshold corresponding to the type of the target evaluation function as the target anomaly detection threshold.

[0135] When there are a continuous preset second number of maximum probability values ​​corresponding to the same target evaluation function in the MSSIM value sequence, the preset anomaly detection threshold corresponding to the type of the target evaluation function is determined as the target anomaly detection threshold. The preset second number can be set according to actual needs and is not limited in this embodiment.

[0136] S4100: When the target abnormality detection threshold is the normal operation threshold, determine that the abnormality detection result of the audio to be analyzed is no abnormality.

[0137] S4110: When the target abnormality detection threshold is the abnormal operation threshold, determine that the abnormality detection result of the audio to be analyzed is an abnormality of an abnormal type corresponding to the abnormal operation threshold.

[0138] Types of anomalies can include wear, stalling, icing, and loose connections.

[0139] In a specific example of abnormal operation detection of wind turbine blades, the preset abnormality detection threshold can be determined by obtaining audio samples required for the threshold determination process, performing audio processing on the audio samples to obtain test audio samples, calculating the average structural similarity of the test audio samples by a method as described in any embodiment, and determining the preset abnormality detection threshold based on the distribution characteristics of the average structural similarity data and a preset evaluation function and probability function.

[0140] The process of determining the audio sample may be to obtain the original audio sample by collecting the sound of the fan blade operation on-site through a sound collector, and classify and organize the original audio sample to form an original audio sample library, including normal operation audio samples, fault operation audio samples, noise interference audio samples, and test audio samples. The test audio sample includes a normal operation audio sample for testing and a fault operation audio sample for testing, and the fault operation audio sample for testing corresponds to at least one abnormal type.

[0141] The normal operation audio samples for testing are obtained by fusing the normal operation audio samples and the noise interference audio samples, and the fault operation audio samples for testing are obtained by fusing the fault operation audio samples and the noise interference audio samples. The noise interference audio samples contain a variety of interference information, such as lightning strikes, car sounds, bird calls, strong wind noise, foreign object impacts, and sounds produced by people, and the influence of interference information on the results is eliminated as much as possible. By making the test audio samples contain unknown noise interference and audio of various blade damage forms, the accuracy of threshold determination and the accuracy of anomaly detection can be improved.

[0142] The technical solution of this embodiment is to obtain the audio to be analyzed generated by the target fan blades during operation, and generate a spectrogram to be analyzed corresponding to the audio to be analyzed; determine the segmentation point sequence coordinates of the sweep and sweep interval of the target fan blades during operation according to the spectrogram to be analyzed, segment the spectrogram to be analyzed according to the segmentation point sequence coordinates, and obtain the spectrogram segmentation result; perform transparency superposition on the spectrogram segmentation result to generate an average spectrogram to be analyzed corresponding to the spectrogram to be analyzed; segment the average spectrogram to be analyzed according to preset time intervals to obtain a sequence of sub-average spectrograms to be analyzed; calculate the similarities of adjacent sub-average spectrograms to be analyzed in the sequence of sub-average spectrograms to be analyzed, obtain a sequence of similarity calculation results, and obtain a target evaluation function corresponding to each preset abnormality detection threshold according to a preset abnormality detection threshold and a preset evaluation function, wherein the preset abnormality detection threshold includes a normal operating threshold and at least one abnormal The target evaluation function is at least two; each similarity calculation result in the similarity calculation result sequence is substituted into at least two target evaluation functions to obtain at least two evaluation values ​​for each similarity calculation result; the probability value corresponding to the evaluation value is calculated according to the evaluation value, and the maximum probability value in the probability value is determined; when there are a continuous preset second number of maximum probability values ​​corresponding to the same target evaluation function, the preset abnormality detection threshold corresponding to the type of the target evaluation function is determined as the target abnormality detection threshold; when the target abnormality detection threshold is the normal operation threshold, the abnormality detection result of the audio to be analyzed is determined to be no abnormality; when the target abnormality detection threshold is the abnormal operation threshold, the abnormality detection result of the audio to be analyzed is determined to be an abnormality of the abnormal type corresponding to the abnormal operation threshold; the abnormality detection result of the audio to be analyzed is determined according to the similarity calculation result sequence and the preset abnormality detection threshold. The technical solution of the embodiment of the present invention solves the problem that it is currently difficult to detect abnormalities in fan blades. The abnormality detection result can be obtained by automatically analyzing the audio during the operation of the fan blades, calculating the similarity of the sub-average spectrogram of the audio, and determining the evaluation value and probability value according to the similarity, and improving the efficiency, accuracy and reliability of abnormality detection through logical algorithms and precise thresholds.

[0143] Figure 5The present invention provides a schematic diagram of a fan blade abnormality detection device, which can be applied to the scenario of power fan blade abnormality detection. The fan blade abnormality detection device can be implemented by software and / or hardware, and integrated into a computer terminal device with application development function.

[0144] like Figure 5 As shown, the wind turbine blade operation abnormality detection device includes: a spectrogram generation module 510, a spectrogram segmentation module 520, an average spectrogram generation module 530, a sub-average spectrogram generation module 540 and an abnormality detection result determination module 550.

[0145] Among them, the spectrogram generation module 510 is used to obtain the audio to be analyzed generated by the target fan blades during operation, and generate a spectrogram to be analyzed corresponding to the audio to be analyzed; the spectrogram segmentation module 520 is used to determine the segmentation point sequence coordinates of the sweep and sweep intervals of the target fan blades during operation according to the spectrogram to be analyzed, and segment the spectrogram to be analyzed according to the segmentation point sequence coordinates to obtain the spectrogram segmentation result; the average spectrogram generation module 530 is used to perform transparency superposition on the spectrogram segmentation result to generate an average spectrogram to be analyzed corresponding to the spectrogram to be analyzed; the sub-average spectrogram generation module 540 is used to segment the average spectrogram to be analyzed according to a preset time interval to obtain a sequence of sub-average spectrograms to be analyzed; the abnormality detection result determination module 550 is used to calculate the similarity of adjacent sub-average spectrograms to be analyzed in the sequence of sub-average spectrograms to be analyzed, obtain a sequence of similarity calculation results, and determine the abnormality detection result of the audio to be analyzed according to the similarity calculation result sequence and a preset abnormality detection threshold.

[0146] The technical solution of this embodiment is to obtain the audio to be analyzed generated by the target fan blade during operation, and generate a spectrogram to be analyzed corresponding to the audio to be analyzed; determine the segmentation point sequence coordinates of the sweep and sweep interval of the target fan blade during operation according to the spectrogram to be analyzed, segment the spectrogram to be analyzed according to the segmentation point sequence coordinates, and obtain the spectrogram segmentation result; perform transparency superposition on the spectrogram segmentation result to generate an average spectrogram to be analyzed corresponding to the spectrogram to be analyzed; segment the average spectrogram to be analyzed according to a preset time interval to obtain a sequence of sub-average spectrograms to be analyzed; calculate the similarity of adjacent sub-average spectrograms to be analyzed in the sequence of sub-average spectrograms to be analyzed, obtain a sequence of similarity calculation results, and determine the abnormality detection result of the audio to be analyzed according to the similarity calculation result sequence and a preset abnormality detection threshold. The technical solution of the embodiment of the present invention solves the problem that it is currently difficult to detect abnormalities in fan blades, and can obtain abnormality detection results by automatically analyzing the audio during the operation of the fan blades, thereby improving the efficiency, accuracy and reliability of abnormality detection.

[0147] In an optional implementation, the spectrogram generation module 510 is specifically used for:

[0148] The audio to be analyzed is framed according to preset frame parameters to obtain a frame processing result; each frame spectrum in the frame processing result is multiplied by a window function to obtain a windowing processing result; the windowing processing result is Fourier transformed to obtain a Fourier transform result; the energy value of each frequency component in the Fourier transform result is calculated, and the energy value is mapped to a color value to obtain a spectrogram to be analyzed corresponding to the audio to be analyzed.

[0149] In an optional implementation, the spectrogram segmentation module 520 is specifically used for:

[0150] Determine the spectrogram matrix to be analyzed corresponding to the spectrogram to be analyzed according to the energy of each frequency component; perform cluster analysis on the eigenvectors in the spectrogram matrix to be analyzed to obtain the cluster analysis results; determine the dimension that affects the clustering effect according to the cluster analysis results, perform dimensionality reduction processing on the spectrogram matrix to be analyzed according to the dimension that affects the clustering effect to obtain the reduced-dimensional spectrogram matrix; perform cluster analysis on the reduced-dimensional spectrogram matrix to obtain a preliminary label sequence of the sweep and sweep interval of the target fan blades during operation; perform abnormal point filtering on the preliminary label sequence to obtain the sequence coordinates of the sweep and sweep interval segmentation points of the target fan blades during operation.

[0151] In an optional implementation, the abnormal detection result determination module 550 is specifically used to:

[0152] Adjacent sub-average spectrograms to be analyzed in the sequence of sub-average spectrograms to be analyzed are divided into a preset first number of block images of the same size; the mean, variance and covariance of each block image are calculated by Gaussian weighting; the structural similarity of the corresponding block image groups in the adjacent sub-average spectrograms to be analyzed is calculated according to the mean, variance and covariance; the average value of the structural similarities of all block image groups is calculated, and the average value is used as the similarity of the adjacent sub-average spectrograms to be analyzed.

[0153] In an optional implementation, the preset abnormality detection threshold includes a normal operation threshold and an abnormal operation threshold of at least one abnormal type, and the abnormality detection result determination module 550 is further used to:

[0154] According to the preset abnormality detection threshold and the preset evaluation function, a target evaluation function corresponding to each preset abnormality detection threshold is obtained, wherein there are at least two target evaluation functions; each similarity calculation result in the similarity calculation result sequence is substituted into at least two target evaluation functions to obtain at least two evaluation values ​​for each similarity calculation result; a probability value corresponding to the evaluation value is calculated according to the evaluation value, and a maximum probability value among the probability values ​​is determined; when there are a continuous preset second number of maximum probability values ​​corresponding to the same target evaluation function, the preset abnormality detection threshold corresponding to the type of the target evaluation function is determined to be the target abnormality detection threshold; when the target abnormality detection threshold is the normal operating threshold, the abnormality detection result of the audio to be analyzed is determined to be no abnormality; when the target abnormality detection threshold is the abnormal operating threshold, the abnormality detection result of the audio to be analyzed is determined to be an abnormality of the abnormal type corresponding to the abnormal operating threshold.

[0155] In an optional embodiment, the device further comprises:

[0156] The audio processing module to be analyzed is used to remove wind noise in the audio to be analyzed through a Butterworth high-pass filter to obtain the audio to be analyzed with the wind noise removed; determine the signal incident direction of the audio to be analyzed with the wind noise removed through a sparse Bayesian learning direction-finding algorithm; enhance the signal in the signal incident direction through a conventional beamforming algorithm, and weaken the interference in the non-signal incident direction, to obtain the audio to be analyzed with signal enhancement.

[0157] The fan blade operation abnormality detection device provided in the embodiment of the present invention can execute the fan blade operation abnormality detection method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0158] Figure 6 A schematic diagram of the structure of a computer device provided in an embodiment of the present invention. Figure 6 A block diagram of an exemplary computer device 12 suitable for use in implementing embodiments of the present invention is shown. Figure 6 The computer device 12 shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention. The computer device 12 can be any terminal device with computing capabilities, such as an intelligent controller and server, a mobile phone and other terminal devices.

[0159] like Figure 6 As shown, the computer device 12 is in the form of a general-purpose computing device. The components of the computer device 12 may include, but are not limited to: one or more processors or processing units 16, a system memory 28, and a bus 18 that connects various system components (including the system memory 28 and the processing unit 16).

[0160] Bus 18 represents one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processor or a local bus using any of a variety of bus architectures. By way of example, these architectures include, but are not limited to, an Industry Standard Architecture (ISA) bus, a Micro Channel Architecture (MAC) bus, an Enhanced ISA bus, a Video Electronics Standards Association (VESA) local bus, and a Peripheral Component Interconnect (PCI) bus.

[0161] The computer device 12 typically includes a variety of computer system readable media. These media can be any available media that can be accessed by the computer device 12, including volatile and non-volatile media, removable and non-removable media.

[0162] The system memory 28 may include computer system readable media in the form of volatile memory, such as random access memory (RAM) 30 and / or cache memory 32. The computer device 12 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, the storage system 34 may be used to read and write non-removable, non-volatile magnetic media ( Figure 6 not shown, usually called a "hard drive"). Although Figure 6 Not shown in the figure, a disk drive for reading and writing to a removable non-volatile disk (e.g., a "floppy disk"), and an optical disk drive for reading and writing to a removable non-volatile optical disk (e.g., a CD-ROM, a DVD-ROM, or other optical media) may be provided. In these cases, each drive may be connected to the bus 18 via one or more data medium interfaces. The system memory 28 may include at least one program product having a set (e.g., at least one) of program modules that are configured to perform the functions of the various embodiments of the present invention.

[0163] A program / utility 40 having a set (at least one) of program modules 42 may be stored, for example, in system memory 28, such program modules 42 including, but not limited to, an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment. Program modules 42 generally perform the functions and / or methods of the embodiments described herein.

[0164] The computer device 12 may also communicate with one or more external devices 14 (e.g., keyboards, pointing devices, displays 24, etc.), one or more devices that enable a user to interact with the computer device 12, and / or any device that enables the computer device 12 to communicate with one or more other computing devices (e.g., network cards, modems, etc.). Such communication may be performed via an input / output (I / O) interface 22. Furthermore, the computer device 12 may also communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) via a network adapter 20. As shown, the network adapter 20 communicates with the other modules of the computer device 12 via the bus 18. It should be understood that although Figure 6 Not shown, other hardware and / or software modules may be used in conjunction with computer device 12, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RFID systems, tape drives, and data backup storage systems.

[0165] The processing unit 16 executes various functional applications and data processing by running the program stored in the system memory 28, for example, implementing the abnormal operation detection method of the fan blade provided in the embodiment of the present invention, the method comprising:

[0166] Acquire the audio to be analyzed generated by the target fan blade during operation, and generate a spectrogram to be analyzed corresponding to the audio to be analyzed;

[0167] Determine the segmentation point sequence coordinates of the sweeping and sweeping intervals of the target fan blades during operation according to the spectrogram to be analyzed, segment the spectrogram to be analyzed according to the segmentation point sequence coordinates, and obtain the spectrogram segmentation result;

[0168] Perform transparency superposition on the spectrogram segmentation results to generate an average spectrogram to be analyzed corresponding to the spectrogram to be analyzed;

[0169] The average spectrogram to be analyzed is segmented according to a preset time interval to obtain a sequence of sub-average spectrograms to be analyzed;

[0170] The similarities of adjacent sub-average spectrograms to be analyzed in the sub-average spectrogram sequence to be analyzed are calculated to obtain a similarity calculation result sequence, and the anomaly detection result of the audio to be analyzed is determined according to the similarity calculation result sequence and a preset anomaly detection threshold.

[0171] An embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon. When the program is executed by a processor, the method for detecting abnormal operation of a fan blade provided in any embodiment of the present invention is implemented. The method includes:

[0172] Acquire the audio to be analyzed generated by the target fan blade during operation, and generate a spectrogram to be analyzed corresponding to the audio to be analyzed;

[0173] Determine the segmentation point sequence coordinates of the sweeping and sweeping intervals of the target fan blades during operation according to the spectrogram to be analyzed, segment the spectrogram to be analyzed according to the segmentation point sequence coordinates, and obtain the spectrogram segmentation result;

[0174] Perform transparency superposition on the spectrogram segmentation results to generate an average spectrogram to be analyzed corresponding to the spectrogram to be analyzed;

[0175] The average spectrogram to be analyzed is segmented according to a preset time interval to obtain a sequence of sub-average spectrograms to be analyzed;

[0176] The similarities of adjacent sub-average spectrograms to be analyzed in the sub-average spectrogram sequence to be analyzed are calculated to obtain a similarity calculation result sequence, and the anomaly detection result of the audio to be analyzed is determined according to the similarity calculation result sequence and a preset anomaly detection threshold.

[0177] The computer storage medium of the embodiment of the present invention can adopt any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium can be, for example, but not limited to: an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples (non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this document, a computer-readable storage medium can be any tangible medium containing or storing a program, which can be used by an instruction execution system, device or device or used in combination with it.

[0178] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, which carry computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0179] The program code embodied on the computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0180] Computer program code for performing the operation of the present invention may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, Python, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0181] An embodiment of the present invention further provides a computer program product, including a computer program, which, when executed by a processor, implements a method for detecting abnormal operation of a wind turbine blade as provided in any embodiment of the present application.

[0182] In the process of implementation, the computer program product can be written in one or more programming languages ​​or a combination thereof to perform the computer program code of the present invention, including object-oriented programming languages, such as Java, Smalltalk, Python, C++, and conventional procedural programming languages, such as "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (for example, using an Internet service provider to connect through the Internet).

[0183] It should be understood by those skilled in the art that the modules or steps of the present invention described above can be implemented by a general-purpose computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, optionally, they can be implemented by a program code executable by a computer device, so that they can be stored in a storage device and executed by the computing device, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present invention is not limited to any specific combination of hardware and software.

[0184] Note that the above are only preferred embodiments of the present invention and the technical principles used. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and that various obvious changes, readjustments and substitutions can be made by those skilled in the art without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in more detail through the above embodiments, the present invention is not limited to the above embodiments, and may include more other equivalent embodiments without departing from the concept of the present invention, and the scope of the present invention is determined by the scope of the appended claims.

Claims

1. A method for detecting abnormal operation of fan blades, characterized in that: include: Acquire the audio to be analyzed generated by the target fan blade during operation, and generate a spectrogram to be analyzed corresponding to the audio to be analyzed; Determine the segmentation point sequence coordinates of the sweeping and sweeping intervals of the target fan blade during operation according to the spectrogram to be analyzed, and segment the spectrogram to be analyzed according to the segmentation point sequence coordinates to obtain a spectrogram segmentation result; Performing transparency superposition on the spectrogram segmentation result to generate an average spectrogram to be analyzed corresponding to the spectrogram to be analyzed; Segmenting the average spectrogram to be analyzed according to preset time intervals to obtain a sequence of sub-average spectrograms to be analyzed; The similarities of adjacent sub-average spectrograms to be analyzed in the sub-average spectrogram sequence to be analyzed are calculated to obtain a similarity calculation result sequence, and the anomaly detection result of the audio to be analyzed is determined according to the similarity calculation result sequence and a preset anomaly detection threshold.

2. The method according to claim 1, characterized in that The step of generating a spectrogram to be analyzed corresponding to the audio to be analyzed includes: Performing frame processing on the audio to be analyzed according to preset frame parameters to obtain a frame processing result; Multiplying each frame spectrum in the frame processing result by a window function to obtain a windowing processing result; Performing Fourier transform on the windowing processing result to obtain a Fourier transform result; The energy value of each frequency component in the Fourier transform result is calculated, and the energy value is mapped to a color value to obtain a spectrogram to be analyzed corresponding to the audio to be analyzed.

3. The method according to claim 2, characterized in that The step of determining the segmentation point sequence coordinates of the sweeping and sweeping intervals of the target fan blade during operation according to the to-be-analyzed spectrogram includes: Determine the spectrogram matrix to be analyzed corresponding to the spectrogram to be analyzed according to the energy of each frequency component; Performing cluster analysis on the feature vectors in the spectrogram matrix to be analyzed to obtain a cluster analysis result; Determine the dimension that affects the clustering effect according to the cluster analysis result, and perform dimensionality reduction processing on the spectrogram matrix to be analyzed according to the dimension that affects the clustering effect to obtain a reduced-dimensional spectrogram matrix; Performing cluster analysis on the dimension-reduced spectrogram matrix to obtain a preliminary label sequence of the sweeping and sweeping intervals of the target fan blades during operation; The preliminary label sequence is subjected to outlier filtering to obtain the sweeping and sweeping interval segmentation point sequence coordinates of the target fan blade during operation.

4. The method according to claim 1, characterized in that: The calculating the similarity of adjacent sub-average spectrograms to be analyzed in the sub-average spectrogram sequence to be analyzed comprises: Dividing adjacent sub-average spectrograms to be analyzed in the sub-average spectrogram sequence to be analyzed into a preset first number of block images of the same size; Calculate the mean, variance and covariance of each of the block images by Gaussian weighting; Calculate the structural similarity of the corresponding block image groups in the adjacent sub-average spectrograms to be analyzed according to the mean, the variance and the covariance; The average value of the structural similarities of all the block image groups is calculated, and the average value is used as the similarity of the adjacent sub-average spectrograms to be analyzed.

5. The method according to claim 1, characterized in that The preset abnormality detection threshold includes a normal operation threshold and an abnormal operation threshold of at least one abnormal type, and the abnormality detection result of the audio to be analyzed is determined according to the similarity calculation result sequence and the preset abnormality detection threshold, including: According to the preset abnormality detection threshold and the preset evaluation function, a target evaluation function corresponding to each of the preset abnormality detection thresholds is obtained, wherein there are at least two target evaluation functions; Substituting each similarity calculation result in the similarity calculation result sequence into at least two of the target evaluation functions to obtain at least two evaluation values ​​for each of the similarity calculation results; Calculating a probability value corresponding to the evaluation value according to the evaluation value, and determining a maximum probability value among the probability values; When there are a continuous preset second number of the maximum probability values ​​corresponding to the same target evaluation function, determining the preset anomaly detection threshold corresponding to the type of the target evaluation function as the target anomaly detection threshold; When the target abnormality detection threshold is the normal operation threshold, determining that the abnormality detection result of the audio to be analyzed is no abnormality; In the case where the target abnormality detection threshold is the abnormal operation threshold, it is determined that the abnormality detection result of the audio to be analyzed is an abnormality of an abnormal type corresponding to the abnormal operation threshold.

6. The method according to claim 1, characterized in that Before generating the spectrogram to be analyzed corresponding to the audio to be analyzed, the method further includes: Removing wind noise from the audio to be analyzed by using a Butterworth high-pass filter to obtain the audio to be analyzed with the wind noise removed; Determine the signal incident direction of the audio to be analyzed after removing wind noise by a sparse Bayesian learning direction finding algorithm; The signal in the signal incident direction is enhanced by a conventional beamforming algorithm, and interference in the non-signal incident direction is weakened, so as to obtain the audio to be analyzed with enhanced signal.

7. A fan blade operation abnormality detection device, characterized in that: include: A spectrogram generation module, used to obtain the audio to be analyzed generated by the target wind turbine blade during operation, and generate a spectrogram to be analyzed corresponding to the audio to be analyzed; a spectrogram segmentation module, for determining the segmentation point sequence coordinates of the sweeping and sweeping intervals of the target fan blade during operation according to the spectrogram to be analyzed, and segmenting the spectrogram to be analyzed according to the segmentation point sequence coordinates to obtain a spectrogram segmentation result; An average spectrogram generation module, used for performing transparency superposition on the spectrogram segmentation result to generate an average spectrogram to be analyzed corresponding to the spectrogram to be analyzed; A sub-average spectrogram generation module, used for segmenting the average spectrogram to be analyzed according to preset time intervals to obtain a sequence of sub-average spectrograms to be analyzed; The abnormality detection result determination module is used to calculate the similarity of the adjacent sub-average spectrograms to be analyzed in the sub-average spectrogram sequence to be analyzed, obtain a similarity calculation result sequence, and determine the abnormality detection result of the audio to be analyzed based on the similarity calculation result sequence and a preset abnormality detection threshold.

8. A computer device, characterized in that: The computer device comprises: one or more processors; A memory for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the method for detecting abnormal operation of wind blades as described in any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method for detecting abnormal operation of a wind turbine blade as described in any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the computer program implements the method for detecting abnormal operation of a wind turbine blade as described in any one of claims 1 to 6.

Images