Titanium alloy bar surface wear morphology detection method and system

Abstract

The application relates to the technical field of ultrasonic testing, in particular to a titanium alloy bar surface wear morphology detection method and system. The method comprises the following steps: performing non-contact scanning on the titanium alloy bar surface through a preset laser ultrasonic scanning system, obtaining a one-dimensional time domain ultrasonic signal, and decomposing the ultrasonic signal into multiple IMF components; determining the point oscillation persistence factor of each signal point in each IMF component; based on time domain window division, calculating the oscillation discreteness index of each point oscillation persistence factor, and calculating the frequency domain sharpness of each IMF component and the original signal; fusing the oscillation discreteness index and the frequency domain sharpness, generating an adaptive wavelet threshold parameter, and adopting the threshold parameter to denoise each IMF component to obtain an enhanced ultrasonic signal; and generating a titanium alloy bar surface wear morphology diagram. The application realizes significant improvement of the accuracy and robustness of titanium alloy bar surface micro-wear detection in a high-noise environment.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of ultrasonic testing, in particular to a titanium alloy bar surface wear morphology detection method and system. BACKGROUND

[0002] In the scene of titanium alloy bar ultrasonic detection, ultrasonic detection is usually needed, and the internal defects of the bar are identified through ultrasonic images. The common way is to obtain the ultrasonic signal inside the bar through ultrasonic waves, and to process or model the ultrasonic signal to obtain the ultrasonic image. Then, because there is noise in the obtained ultrasonic signal, it is necessary to perform enhancement or denoising processing before converting the ultrasonic signal into an ultrasonic image.

[0003] The prior art obtains the ultrasonic signal inside the bar through ultrasonic waves, and generates an ultrasonic image through processing or modeling to identify defects. However, in actual detection, the ultrasonic signal is easily mixed with broadband thermoelastic noise, which has frequency spectrum overlapping and energy advantage, and is difficult to separate from defect signals such as micro-abrasion and cracks in the frequency domain. Traditional denoising methods such as EMD and wavelet threshold denoising are based on a single dimension of frequency or energy, and are prone to over-denoising (loss of defect signals) or insufficient denoising (noise remaining interference in identification), resulting in poor ultrasonic image quality and low defect identification accuracy, which cannot meet the high-precision wear detection requirements of titanium alloy bars. SUMMARY

[0004] In order to solve the technical problem that over-denoising or insufficient denoising easily occurs when processing defect signals, the purpose of the present application is to provide a titanium alloy bar surface wear morphology detection method and system, and the technical solution adopted is as follows:

[0005] In the first aspect, the present application provides a titanium alloy bar surface wear morphology detection method, which comprises:

[0006] The titanium alloy bar surface is scanned by a pre-set laser ultrasonic scanning system in a non-contact manner, one-dimensional time domain ultrasonic signals are obtained, and the one-dimensional time domain ultrasonic signals are decomposed into multiple IMF components;

[0007] The point oscillation persistence factor of each signal point in each IMF component is determined;

[0008] Based on time domain window division, the oscillation discreteness index of the point oscillation persistence factor in each window is calculated, and the frequency domain sharpness of each IMF component and the original signal is calculated through frequency energy distribution comparison;

[0009] The oscillation discreteness index and the frequency domain sharpness are fused to generate an adaptive wavelet threshold parameter, and the threshold parameter is used to denoise each IMF component to reconstruct an enhanced ultrasonic signal;

[0010] Identify the wear type of the titanium alloy bar and generate a titanium alloy bar surface wear morphology map based on the enhanced ultrasonic signal.

[0011] In some embodiments, the titanium alloy bar surface is scanned by a preset laser ultrasonic scanning system to obtain a one-dimensional time domain ultrasonic signal, including:

[0012] Start the preset laser ultrasonic scanning system, which includes a pulsed laser excitation module for exciting ultrasonic waves and a signal receiving module for receiving ultrasonic echo signals;

[0013] Control the pulsed laser excitation module to emit pulsed laser to the titanium alloy bar surface to excite ultrasonic waves on the titanium alloy bar surface;

[0014] Receive the ultrasonic echo signal reflected from the titanium alloy bar surface through the signal receiving module;

[0015] Sample the ultrasonic echo signal received by the signal receiving module at a preset sampling frequency, and convert the sampled ultrasonic echo signal into a one-dimensional time domain ultrasonic signal.

[0016] In some embodiments, the one-dimensional time domain ultrasonic signal is decomposed into a plurality of IMF components, including:

[0017] Adopting an empirical mode decomposition method to adaptively decompose and process the one-dimensional time domain ultrasonic signal to decompose the one-dimensional time domain ultrasonic signal into a plurality of IMF components;

[0018] Synchronously obtain the residual term corresponding to the one-dimensional time domain ultrasonic signal during the decomposition process.

[0019] In some embodiments, the point oscillation duration factor of each signal point in each IMF component is determined, including:

[0020] Select any one IMF component from the plurality of IMF components obtained by decomposition as a current target IMF component to be calculated;

[0021] Identify all signal points in the current target IMF component that need to calculate the point oscillation duration factor;

[0022] For each signal point in the current target IMF component, respectively obtain the absolute amplitude of the signal point, and the absolute amplitudes of the two adjacent signal points before and after the signal point;

[0023] Calculate the first absolute value of the absolute amplitude change between the current signal point and the front adjacent signal point, and the second absolute value of the absolute amplitude change between the current signal point and the rear adjacent signal point;

[0024] The point oscillation persistence factor of the current signal point is calculated based on the first absolute value and the second absolute value, and the point oscillation persistence factor of all signal points in the current target IMF component is calculated in turn.

[0025] In some embodiments, the step of calculating the oscillation dispersion index of the oscillation duration factor within each window based on time-domain window partitioning includes:

[0026] Obtain the instantaneous frequency of each IMF component, and determine the duration of the time-domain window based on the instantaneous frequency;

[0027] According to the determined time-domain window duration, each IMF component and the one-dimensional time-domain ultrasound signal are divided into windows to obtain several independent windows;

[0028] For each independent window after division and its corresponding IMF component, select any signal point within the independent window and construct a local neighborhood containing the signal point and a preset number of adjacent signal points.

[0029] Calculate the average value of the point oscillation duration factor of all signal points in each local neighborhood, and use it as the local neighborhood factor mean value corresponding to the local neighborhood.

[0030] The number of signal points in each local neighborhood whose oscillation persistence factor is greater than the mean of the corresponding local neighborhood factor is counted, and the weight of the corresponding signal point in the local neighborhood is calculated based on the counted number of signal points.

[0031] Calculate the average value of the point oscillation duration factor of all signal points within each independent window, and combine the weight of each signal point within the independent window with the deviation of the point oscillation duration factor from the average value to obtain the oscillation dispersion of each independent window.

[0032] For each IMF component, the ratio of the standard deviation to the mean of the point oscillation persistence factor within each independent window corresponding to the IMF component is calculated, and the discrete contribution value of each independent window is obtained through the ratio and the oscillation dispersion.

[0033] The discrete contribution values ​​of all independent windows corresponding to each IMF component are summed and then divided by the total number of independent windows corresponding to the IMF component to obtain the oscillation dispersion index of each IMF component.

[0034] In some embodiments, calculating the frequency domain sharpness of each IMF component compared to the original signal through frequency domain energy distribution comparison includes:

[0035] performing Fourier transform on the one-dimensional time domain ultrasonic signal to obtain a frequency spectrum of the one-dimensional time domain ultrasonic signal, and calculating an energy probability density function of the one-dimensional time domain ultrasonic signal based on the frequency spectrum;

[0036] performing Fourier transform on each of the IMF components to obtain a frequency spectrum of each of the IMF components, and calculating an energy probability density function corresponding to each of the IMF components based on the frequency spectrum of each of the IMF components;

[0037] for each of the IMF components, calculating an absolute difference value between the energy probability density function corresponding to the IMF component at each frequency point and the energy probability density function of the one-dimensional time domain ultrasonic signal to form a difference vector value;

[0038] summing all frequency points of the difference vector value to obtain a total difference value of the IMF component and the one-dimensional time domain ultrasonic signal in frequency energy distribution;

[0039] performing normalization processing on the frequency spectrum of the one-dimensional time domain ultrasonic signal to obtain a normalized amplitude spectrum of the one-dimensional time domain ultrasonic signal, and calculating a mean value of the normalized amplitude spectrum;

[0040] calculating an absolute difference value between the normalized amplitude spectrum of the one-dimensional time domain ultrasonic signal and the mean value of the normalized amplitude spectrum at each frequency point, and summing all frequency points of the absolute difference value to obtain a calculation denominator;

[0041] determining the frequency domain sharpness of the IMF component by the total difference value and the calculation denominator.

[0042] In some embodiments, the fusion of the oscillation discreteness indicator and the frequency domain sharpness generates an adaptive wavelet threshold parameter, including:

[0043] obtaining an oscillation discreteness indicator and a frequency domain sharpness corresponding to each of the IMF components;

[0044] performing linear normalization processing on the oscillation discreteness indicator to obtain a reference threshold value corresponding to each of the IMF components;

[0045] performing negative correlation normalization processing on the frequency domain sharpness to obtain a frequency domain correction coefficient corresponding to each of the IMF components;

[0046] fusing and calculating the reference threshold value corresponding to each of the IMF components and the corresponding frequency domain correction coefficient to determine an adaptive wavelet threshold parameter corresponding to each of the IMF components.

[0047] In some embodiments, the denoising of each IMF component by using the threshold parameter to reconstruct an enhanced ultrasonic signal, including:

[0048] call the adaptive wavelet threshold parameter corresponding to each IMF component, and perform wavelet transform on the current IMF component to obtain corresponding wavelet coefficients;

[0049] compare the wavelet coefficients with the adaptive wavelet threshold parameter, eliminate the wavelet coefficients less than the adaptive wavelet threshold parameter, and perform inverse wavelet transform on the retained wavelet coefficients to obtain the current IMF component after denoising;

[0050] call the residual term, participate in signal reconstruction together with all the current IMF components after denoising and the corresponding residual term, and integrate the current IMF components after denoising and the residual term through Hilbert transform to obtain the reconstructed enhanced ultrasonic signal.

[0051] In some embodiments, the identification of the wear type and the generation of the titanium alloy bar surface wear morphology map based on the enhanced ultrasonic signal include:

[0052] performing Hilbert transform on the enhanced ultrasonic signal to construct a complex signal with the enhanced ultrasonic signal as the real part and the transform result as the imaginary part;

[0053] calculating the modulus value of the complex signal to obtain the envelope signal corresponding to the enhanced ultrasonic signal;

[0054] extracting the time domain features from the envelope signal, inputting the extracted time domain features into a preset machine learning classifier, matching with a preset wear feature database, and identifying the wear type and the corresponding position distribution on the surface of the titanium alloy bar through the matching result;

[0055] processing the identified wear type and position distribution by using a three-dimensional imaging algorithm to generate a titanium alloy bar surface wear morphology map.

[0056] In a second aspect, the embodiments of the present application provide a titanium alloy bar surface wear morphology detection system, which includes the following modules:

[0057] The acquisition module is configured to perform non-contact scanning on the surface of the titanium alloy bar by using a preset laser ultrasonic scanning system, acquire a one-dimensional time domain ultrasonic signal, and decompose the ultrasonic signal into multiple IMF components;

[0058] The determination module is configured to determine the point oscillation persistence factor of each signal point in each IMF component;

[0059] The calculation module is configured to calculate the oscillation discreteness index of the point oscillation persistence factor in each window based on time domain window division, and calculate the frequency domain sharpness of each IMF component and the original signal through frequency energy distribution comparison.

[0060] The generation module is used to fuse the oscillation dispersion index and the frequency domain sharpness to generate an adaptive wavelet threshold parameter, and to use the threshold parameter to denoise each IMF component to reconstruct an enhanced ultrasound signal.

[0061] The identification module is used to identify the wear type of the scanned titanium alloy rod based on the enhanced ultrasonic signal and generate a wear morphology map of the titanium alloy rod surface.

[0062] Thirdly, embodiments of the present invention provide an electronic device, including a memory and a processor, wherein the memory stores executable code, and when the processor executes the executable code, it implements the various possible implementations of the first aspect.

[0063] Fourthly, embodiments of the present invention provide a computer program product comprising: computer program code, which, when executed on a computer, causes the computer to perform the method described in the first aspect or any possible implementation thereof.

[0064] Fifthly, embodiments of the present invention provide a computer-readable storage medium having a computer program stored thereon, which, when executed in a computer, causes the computer to perform the various possible implementations of the first aspect.

[0065] The embodiments of the present invention have at least the following beneficial effects:

[0066] This invention achieves superior technical results through a multi-step collaborative process: First, a complete one-dimensional time-domain ultrasonic signal is acquired using non-contact laser ultrasonic scanning, and then decomposed into multiple IMF components by EMD to avoid signal damage and disassemble noise and defect signal carriers; next, the time-domain characteristics of the signal points are quantified by the point oscillation persistence factor, and combined with time-domain windowing, oscillation discreteness index, and frequency-domain sharpness, broadband thermoelastic noise and defect signals are accurately distinguished from each other in both time and frequency domains; then, adaptive wavelet threshold parameters are generated by fusing the two-dimensional indices to specifically denoise and reconstruct the enhanced ultrasonic signal, solving the problems of over- or under-denoising in traditional methods; finally, wear types are identified based on the enhanced signal and a morphology map is generated, significantly improving the accuracy and robustness of micro-wear detection in high-noise environments, and enabling efficient quantitative assessment and localization of wear. Attached Figure Description

[0067] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0068] Figure 1 A method flow chart of a titanium alloy bar surface wear morphology detection method provided by an embodiment of the present application;

[0069] Figure 2 A system block diagram of a titanium alloy bar surface wear morphology detection system provided by an embodiment of the present application;

[0070] Figure 3 A structural schematic diagram of a computer device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0071] In order to further illustrate the technical means and effects adopted by the present application to achieve the predetermined object of the present application, the specific implementation, structure, features and effects of the titanium alloy bar surface wear morphology detection method and system according to the present application are described in detail as follows in combination with the drawings and preferred embodiments.

[0072] In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0073] In the description of embodiments of the present application, unless otherwise specified, " / " represents the meaning of or, for example, A / B can represent A or B: "and / or" in the text only describes the association relationship of the associated objects, which means that there can be three relationships, for example, A and / or B, which can represent: A exists alone, A and B exist together, and B exists alone, in addition, in the description of embodiments of the present application, "multiple" means two or more than two.

[0074] Hereinafter, the terms "first", "second" are only used for description purposes, and cannot be understood as implying or suggesting relative importance or implicitly indicating the number of indicated technical features. Therefore, the features defined with "first", "second" can explicitly or implicitly include one or more features.

[0075] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as understood by a person skilled in the art to which the present application belongs.

[0076] The embodiments of the present application are described below in combination with the drawings. Those skilled in the art can know that with the development of technology and the appearance of new scenes, the technical solutions provided by the embodiments of the present application are also applicable to similar technical problems.

[0077] The specific scheme of the titanium alloy bar surface wear morphology detection method and system provided by the present application is described in detail below in combination with the drawings.

[0078] Embodiment one:

[0079] Please refer to Figure 1 , which shows a step flow chart of a titanium alloy bar surface wear morphology detection method provided by an embodiment of the application, the method comprising the following steps:

[0080] S10. The surface of the titanium alloy bar is scanned by a preset laser ultrasonic scanning system to obtain a one-dimensional time domain ultrasonic signal, and the ultrasonic signal is decomposed into multiple IMF components.

[0081] Specifically, first, start the preset laser ultrasonic scanning system. It should be noted that the preset laser ultrasonic scanning system includes two core modules: a pulsed laser excitation module and a signal receiving module. The function of the pulsed laser excitation module is to emit high-energy pulsed laser to the surface of the titanium alloy bar, use the thermal effect of the laser to make the surface of the bar locally and instantaneously expand, and then excite ultrasonic waves. The ultrasonic waves will propagate inside the bar and will be reflected and scattered when encountering wear defects. The signal receiving module usually uses an interferometer or a piezoelectric sensor. The interferometer is suitable for high-precision detection scenarios and can capture ultrasonic echoes with small amplitudes. The piezoelectric sensor is suitable for industrial mass production detection scenarios and has fast response speed and high stability. The pulsed laser excitation module and the signal receiving module work together to receive ultrasonic echo signals reflected from the surface and internal defects of the titanium alloy bar.

[0082] Control the pulsed laser excitation module to emit pulsed laser to the surface of the titanium alloy bar. The energy density and pulse width of the laser need to be adjusted according to the material properties of the titanium alloy bar, such as thermal conductivity and elastic modulus, to ensure that the excited ultrasonic wave energy is neither too low to cause weak echo signals nor too high to damage the surface of the bar.

[0083] The signal receiving module receives the ultrasonic echo signals. When the ultrasonic waves propagate inside the bar and encounter wear defects, defect echoes will be generated. Ultrasonic waves that do not encounter defects will propagate to the bottom of the bar and then reflect, forming bottom echoes. The signal receiving module needs to capture both defect echoes and bottom echoes to provide a data basis for subsequent defect identification.

[0084] The received ultrasonic echo signals are sampled at a preset sampling frequency. It should be noted that the sampling frequency must strictly satisfy the Nyquist theorem, i.e., the sampling frequency must be at least twice the highest frequency of the signal to avoid aliasing distortion during signal sampling. After sampling, the continuous analog ultrasonic echo signals are converted into discrete digital signals. The digital signals are one-dimensional time domain ultrasonic signals that can be used for subsequent signal decomposition and feature extraction.

[0085] Further, an empirical mode decomposition method (EMD) is used for adaptive decomposition of the one-dimensional time-domain ultrasonic signal: the EMD decomposition does not need to preset a base function, and can adaptively divide the frequency components according to the frequency characteristics of the signal itself. First, all local maximum points and minimum points are extracted from the original one-dimensional time-domain ultrasonic signal, and the upper envelope line and the lower envelope line are obtained by cubic spline interpolation fitting, respectively. The mean line of the upper and lower envelope lines is calculated, and the original one-dimensional time-domain ultrasonic signal is subtracted from the mean line to obtain a preliminary intrinsic mode function (IMF) component. If the preliminary component does not meet the judgment condition of the IMF component, in this embodiment, the judgment condition is that the number of extreme points and the number of zero-crossing points differ by more than 1, then the preliminary component is taken as the new "original one-dimensional time-domain ultrasonic signal" to repeat the above screening process until the IMF component that meets the condition is obtained. The IMF component is separated from the original one-dimensional time-domain ultrasonic signal, and the remaining signal is taken as a new signal to be decomposed. The above steps are repeated until the remaining signal is a monotonic signal, i.e., no IMF component can be further decomposed.

[0086] At the same time, the residual term is obtained synchronously during the EMD decomposition process: when the EMD decomposition cannot extract the IMF component from the remaining signal, the remaining signal is the residual term. The residual term is usually a monotonically increasing or monotonically decreasing signal, reflecting the overall trend of the original one-dimensional time-domain ultrasonic signal, such as the baseline slow drift caused by the change of the detection distance. The residual term needs to be retained during subsequent signal reconstruction, otherwise the reconstructed signal will lose the overall trend characteristics.

[0087] The plurality of IMF components obtained by decomposition are recorded separately, and each IMF component needs to be labeled with the corresponding decomposition order (such as IMF1, IMF2, …, IMFn). Generally, the earlier the decomposition order of the IMF component, the higher the frequency, and the more likely it is to be dominated by broadband thermoelastic noise. The later the decomposition order of the IMF component, the lower the frequency, and the more likely it contains defect signals. At the same time, the residual term obtained synchronously is stored in a specified data directory. The storage format needs to be consistent with the IMF component, such as being stored in.txt or.mat format, which facilitates subsequent signal reconstruction.

[0088] S11. Determine the point oscillation persistence factor of each signal point in each of the IMF components.

[0089] The point oscillation persistence factor is a core time-domain feature for distinguishing whether a signal point belongs to noise or a defect signal. Its calculation process needs to accurately correspond to each signal point of each IMF component. Specifically:

[0090] First, select any one of the multiple IMF components obtained by decomposition as the current target IMF component to be calculated, for example, select IMF1 as the current target component first, complete the point oscillation duration factor calculation of all signal points in IMF1, and then select IMF2, IMF3, and so on in turn until all IMF components are calculated.

[0091] Clearly, all signal points in the current target IMF component that need to be calculated for point oscillation duration factor: the current target IMF component is a discrete digital signal, each signal point corresponds to a sampling time, and the signal points at all sampling times need to be calculated for point oscillation duration factor to avoid missing key signal points, such as signal points corresponding to defect echoes.

[0092] Further, for each signal point in the current target IMF component, the absolute amplitude of the signal point is obtained, and it should be noted that the amplitude is the value of the signal at the sampling time, and the absolute amplitude is the absolute value of the amplitude, which is used to eliminate the influence of the positive and negative directions of the amplitude on subsequent calculation, and the absolute amplitudes of the two adjacent signal points before and after the signal point are obtained:

[0093] The time corresponding to the signal point is taken as an example, and the adjacent time points before and after the signal point, i.e. , are obtained. After obtaining the absolute amplitudes of the current time point and the adjacent time points before and after the signal point, the absolute values of the differences between the absolute amplitudes of the current time point and the adjacent time points before and after the signal point are calculated, and the point oscillation duration factor of the current time point is obtained.

[0094]

[0095] Among them, respectively represent the amplitude change of the current time point t and the adjacent time points and ; is the absolute amplitude of the current signal point; is a very small number, which can be selected according to actual needs. The point oscillation duration factor quantifies the total intensity of the amplitude change in the adjacent region of the time point t. Whether the waveform is in the rising state or the falling state, as long as there is amplitude change, the change intensity will be accumulated. For the continuous oscillation of wideband thermal elastic noise, the point oscillation duration factor of each signal point will present a stable and small positive value; for defect signal pulses, the point oscillation duration factor of the pulse edge position will present a very large value. At this time, the greater the value of the point oscillation duration factor , the greater the absolute amplitude The signal point itself can not be large, but the amplitude difference between the signal point and the adjacent signal points before and after the signal point is large, which means that the local environment where the signal point is located can be in a high-frequency vibration state. Therefore, the point oscillation persistence factor can distinguish whether a signal point is in a wideband thermal-elastic noise in stable oscillation or in a special position such as a wave peak or an edge of a defect signal pulse.

[0096] Specifically, if the signal point belongs to a wideband thermal-elastic noise point, the absolute amplitude of the signal point itself may not be large, but the vibration amplitudes of the adjacent signal points before and after the signal point are large, and thus the point oscillation persistence factor obtained is small or in a medium size; if the signal point is in a core position of a defect signal pulse such as a wave peak, the absolute amplitude of the signal point itself may be large, and thus the point oscillation persistence factor obtained is small instead; if the signal point is in an edge position of a defect signal pulse, the absolute amplitude of the signal point itself is extremely small, that is, the denominator in the formula is extremely small, which makes the point oscillation persistence factor obtained extremely large.

[0097] The above operation is repeated for each IMF component, and the point oscillation persistence factor of each signal point corresponding to each time in all IMF components and the original one-dimensional time-domain ultrasonic signal is obtained .

[0098] S12. Based on the time-domain window division, the oscillation discreteness index of the point oscillation persistence factor in each window is calculated, and the frequency-domain sharpness of each IMF component and the original signal is calculated through frequency energy distribution comparison.

[0099] The step of calculating the oscillation discreteness index of the point oscillation persistence factor in each window is specifically as follows:

[0100] First, the instantaneous frequency of each IMF component is obtained, and the time-domain window length is determined according to the instantaneous frequency: each IMF component is processed through Hilbert transform (Hilbert-Huang, HHT), the Hilbert transform is performed on the IMF component to obtain an analytic signal, and then the instantaneous frequency is calculated from the analytic signal. The instantaneous frequency refers to the frequency of the signal at each sampling time, and can reflect the change of the signal frequency with time; the period of the IMF component is calculated by calculating the inverse of the instantaneous frequency, and the period is the inverse of the frequency, reflecting the time required for the signal to complete one complete oscillation; the periods of all IMF components are counted, and the least common multiple of the periods is calculated. The least common multiple is determined as the time-domain window length, which can ensure that each window contains an integer number of IMF component periods, and avoid feature distortion caused by window division segmentation of signal periods.

[0101] According to the determined time domain window length, each IMF component and one-dimensional time domain ultrasonic signal is respectively windowed to obtain a plurality of independent windows. For example, if the total window length is T and the sampling interval is , then the number of signal points contained in each window is (integers are required), and the signal is sequentially intercepted from the start time according to the window length. If the number of signal points in the last window is insufficient, it can be processed by zero padding or truncation.

[0102] For each independent window and the corresponding IMF component after division, any one signal point in the independent window is selected to construct a local neighborhood containing the signal point and a predetermined number of adjacent signal points before and after it. In this embodiment, the predetermined number is usually 4, i.e., the local neighborhood contains the current signal point and 2 signal points before and after it, a total of 5 signal points. This number design can balance the representativeness of local features and computational efficiency. Five signal points can reflect the local oscillation characteristics, and will not cause a sharp increase in computational load due to excessive number. In actual application, the predetermined number can also be adjusted according to actual needs.

[0103] Further, the average value of the point oscillation duration factor of all signal points in each local neighborhood is calculated as the local neighborhood factor mean value corresponding to the local neighborhood. Assuming that a total of windows are divided, the window is taken as an example: for the IMF component in the window, assuming that the local neighborhood of the IMF component contains 5 signal points, and the factor values are w1, w2, w3, w4, and w5, respectively, then the local neighborhood factor mean value is , which can reflect the overall oscillation level of the local neighborhood.

[0104] The number of signal points in each local neighborhood whose point oscillation duration factor is greater than the corresponding local neighborhood factor mean value is counted, and the weight of the corresponding signal point in the local neighborhood is calculated based on the number.

[0105]

[0106] , wherein is the number of signal points in the local neighborhood whose factor value is greater than the mean value; if the number is 3 or more, i.e., it meets the characteristics of "continuous dispersion" of the defect signal, the weight tends to 1, indicating that the signal point has a large contribution to the window oscillation dispersion; if the number is 1 or 5, i.e., it meets the characteristics of "isolated anomaly" of the noise, the weight tends to 0.5, indicating that the contribution of the signal point is small. is an exponential function, which is used to normalize the weight to ensure that the weight value range is within [0.5, 1].

[0107] Further, the oscillation dispersion of the IMF component in the window is calculated :

[0108]

[0109] wherein is the mean value of the point oscillation persistence factor of all signal points in the window , and is a minimum number used to avoid a denominator of 0, The function can map the dispersion to a reasonable range. At this time, the formula solves the problems of pseudo-dispersion caused by noise isolated peaks and pseudo-concentration caused by local flatness of the signal. The greater the oscillation dispersion , the higher the effective dispersion in the current window, that is, the higher the actual signal proportion.

[0110] Further, the oscillation dispersion index of the current IMF component is calculated :

[0111]

[0112] wherein is the standard deviation of the point oscillation persistence factor of all signal points in the window , and is the mean value of the point oscillation persistence factor of all signal points in the window. In a noise-dominant component, the point oscillation persistence factor of each signal point in the window is uniformly distributed, the standard deviation is small, the mean value is also small, and the oscillation dispersion is relatively low. However, in a signal-dominant component, the point oscillation persistence factor of each signal point in the window contains both maximum and minimum values, so the overall dispersion degree is relatively high, and the oscillation dispersion is relatively high. Further, since the current analysis is mainly used to ensure the actual signal information loss in the subsequent process, the oscillation dispersion obtained above is added as a weight in the positive same logic direction, and it is considered that the window with a higher signal proportion has a higher credibility in participating in the judgment of the signal proportion of the current component signal. The greater the oscillation dispersion index , the higher the overall proportion of defect signals in the IMF component, and the more the signal characteristics of the component need to be preserved during denoising.

[0113] Further, the frequency domain sharpness of each IMF component and the original signal is calculated by comparing the frequency energy distribution as follows:

[0114] Firstly, Fourier transform is performed on the one-dimensional time domain ultrasonic signal to obtain the frequency spectrum of the signal. The Fourier transform can convert the time domain signal into the frequency domain signal. The abscissa of the frequency spectrum is the frequency, and the ordinate is the amplitude or power, which can directly reflect the frequency component distribution of the signal. The energy probability density function of the one-dimensional time domain ultrasonic signal is calculated based on the frequency spectrum The energy probability density function is a normalized processing of the spectrum energy, which reflects the distribution probability of the energy at different frequency points. The greater the value of a certain frequency point is, the higher the energy proportion of the frequency point is.

[0115] The Fourier transform is performed on each IMF component to obtain the frequency spectrum of each IMF component. The frequency spectrum of each IMF component only contains its corresponding frequency component. For example, the frequency spectrum of the high-frequency IMF component is concentrated in the high-frequency band, and the frequency spectrum of the low-frequency IMF component is concentrated in the low-frequency band. Based on the frequency spectrum of each IMF component, the energy probability density function corresponding to each IMF component is calculated The calculation method of the energy probability density function is the same as that of the original signal , so as to ensure the comparability of the two.

[0116] Further, for each IMF component, the absolute difference between the energy probability density function corresponding to the IMF component at each frequency point and the energy probability density function of the original signal is calculated to form a difference vector value , that is Each element of the difference vector value corresponds to the energy distribution difference of a frequency point. The greater the element value is, the greater the energy distribution difference between the IMF component and the original signal at the frequency point is. The difference vector value is summed over all frequency points to obtain the total difference value of the IMF component and the original signal in the frequency energy distribution. The total difference value can quantify the overall frequency domain difference between the IMF component and the original signal. The greater the total difference value is, the more the frequency domain characteristics of the IMF component deviate from the original signal, and the more likely the IMF component contains a defect signal. The frequency energy distribution of the defect signal is usually quite different from that of the original signal.

[0117] Further, the frequency domain sharpness of the IMF component at this time is obtained by integrating the difference vector

[0118] :

[0119]

[0120] At this time, represents the normalized amplitude spectrum (energy distribution) of the original signal, ​​​The mean value of the normalized amplitude spectrum of the original signal is represented. It represents the average level of the energy in the entire frequency domain. The sum value of the difference vector values of all frequency points in the amplitude spectrum is represented, that is, the total difference value. It is a very small constant, which can be 0.01, used to avoid zero denominator. The denominator is represented. At this time, the frequency domain sharpness of the obtained signal is higher, indicating that the component and the original signal in the frequency domain distribution are more different, and it is more likely to contain a defect signal.

[0121] S13. Fuse the oscillation discreteness index and the frequency domain sharpness to generate an adaptive wavelet threshold parameter, and use the threshold parameter to denoise each IMF component to reconstruct an enhanced ultrasonic signal.

[0122] Specifically, first, the oscillation discreteness index and the frequency domain sharpness corresponding to each IMF component are obtained , ensuring that each IMF component corresponds to a set of oscillation discreteness indexes and frequency domain sharpness , without omission or mismatch.

[0123] The oscillation discreteness index is linearly normalized to obtain the reference threshold corresponding to each IMF component: The core of linear normalization is to map to a preset threshold range, for example, [0.1, 0.8], which needs to be determined according to industrial detection experience, to avoid that the threshold is too small to cause insufficient denoising, or the threshold is too large to cause signal loss, and it can also be adjusted according to actual needs;

[0124] Specifically, the normalization formula for linearly normalizing the oscillation discreteness index

[0125]

[0126] , wherein is the minimum value of in all IMF components, is the maximum value of in all IMF components, and are the upper and lower limits of the preset threshold range; the larger the oscillation discreteness index , the larger the reference threshold , indicating that the IMF component has a high proportion of defect signals, and the threshold needs to be increased to retain more signals.

[0127] The frequency domain sharpness The negative correlation normalization processing is performed to obtain a frequency domain correction coefficient corresponding to each IMF component : the negative correlation normalization indicates The larger the correction coefficient is, the smaller the correction coefficient is, because The larger the IMF component is, the larger the frequency domain difference of the IMF component is, the more noise it may contain, and the smaller correction coefficient is required to reduce the final threshold, thereby enhancing the denoising effect. Specifically, the frequency domain sharpness The normalization formula for linear normalization processing is:

[0128]

[0129] Among them is the minimum value of the of all IMF components, is the maximum value of the of all IMF components, and are the preset range of the correction coefficient, for example, [0.2, 0.9].

[0130] Further, the wavelet threshold parameter for wavelet threshold denoising of the current component is calculated:

[0131]

[0132] At this time, for a certain component, the higher the oscillation dispersion index is, the higher the proportion of the actual signal inside is represented, so a higher threshold is used as much as possible. For the added frequency domain sharpness, if the feature is large, it represents that the noise content inside is relatively high, at this time, a lower or standard threshold can be used to actively denoise and effectively suppress noise, otherwise, a higher threshold is used to conservatively denoise and to preserve the signal as much as possible.

[0133] Further, the adaptive wavelet threshold parameter corresponding to each IMF component is called, and the wavelet transform is performed on the current IMF component to obtain corresponding wavelet coefficients: the wavelet transform is a signal processing technology for converting a time domain signal into a "time-scale" domain signal, and a suitable wavelet base function needs to be selected. In industrial ultrasonic testing, db4 wavelet is commonly used because of its good time domain compact support and high frequency domain resolution, and it can also be adjusted according to actual needs; the current IMF component is decomposed into N layers, and in this embodiment, the decomposition layer number is usually 3-5 layers, which needs to be determined according to the signal frequency range. The decomposition layer number of high frequency signal can be appropriately increased, to obtain low frequency approximation coefficients and high frequency detail coefficients of each layer. The low frequency approximation coefficient mainly carries the overall characteristics of the signal, and the high frequency detail coefficient mainly carries the detailed characteristics of the noise and defect signal. The subsequent denoising is mainly aimed at the high frequency detail coefficient.

[0134] The wavelet coefficient is compared with the adaptive wavelet threshold parameter Perform comparison, eliminate wavelet coefficients less than , and perform inverse wavelet transform on the retained wavelet coefficients to obtain the current IMF component after denoising: The comparison object is mainly the high-frequency detail coefficient, and the coefficient less than is determined as a noise coefficient and is directly eliminated; the coefficient greater than or equal to is determined as a signal coefficient and is retained; after retention, the processed wavelet coefficients are converted back to a time domain signal through inverse wavelet transform, which is the inverse process corresponding to wavelet decomposition, and the time domain signal is the IMF component after denoising; repeat the step to complete the denoising of all IMF components.

[0135] Retrieve the previously stored residual term, and all denoised IMF components participate in signal reconstruction together with the residual term: The core of signal reconstruction is to superimpose the denoised IMF components and the residual term in the reverse order of the original decomposition, that is:

[0136] Y

[0137] wherein Y represents the original signal, and Z represents the residual term; the time dimension of each component needs to be ensured to be consistent without misplacement or loss during reconstruction; the denoised IMF components and the residual term are integrated through Hilbert transform, the Hilbert transform is performed on each denoised IMF component and the residual term respectively to obtain the respective analytic signals, and then the real part is obtained after superimposing all the analytic signals to obtain the reconstructed enhanced ultrasonic signal; the noise of the enhanced ultrasonic signal is significantly reduced, and the characteristics of the defect signal are more prominent, which can be directly used for subsequent wear identification.

[0138] S14. Identify the wear type of the scanned titanium alloy bar and generate a titanium alloy bar surface wear morphology map based on the enhanced ultrasonic signal.

[0139] Specifically, first, Hilbert transform is performed on the enhanced ultrasonic signal to construct a complex signal with the enhanced ultrasonic signal as the real part and the transform result as the imaginary part; the Hilbert transform can perform phase shift processing on the enhanced ultrasonic signal to obtain a transform result orthogonal to the original signal; the expression of the complex signal is:

[0140]

[0141] wherein is the enhanced ultrasonic signal (real part), is the Hilbert transform result (imaginary part), is the imaginary unit; the complex signal can reflect the amplitude and phase information of the signal at the same time, and lays a foundation for subsequent envelope extraction.

[0142] Calculate the modulus value of the complex signal to obtain the envelope signal corresponding to the enhanced ultrasonic signal, and the modulus value of the complex signal is​ The calculation formula is:

[0143]

[0144] This modulus is the envelope signal. The envelope signal can smooth and enhance the high-frequency fluctuations of the ultrasound signal and highlight the overall trend of the signal. The defect echo appears as a distinct peak in the envelope signal, which facilitates feature extraction.

[0145] Temporal features are extracted from the envelope signal and input into a pre-defined machine learning classifier for matching with a pre-defined wear feature database. The extracted temporal features include: peak value (the maximum amplitude value corresponding to the defect echo in the envelope signal, reflecting the size of the defect; a larger peak value indicates a larger defect); rise time (the time from the envelope signal baseline to the peak value, reflecting the shape of the defect; a shorter rise time indicates a sharper defect edge); and duration (the total time from the defect echo rising to the baseline to falling back to the baseline, reflecting the extent of the defect; a longer duration indicates a wider defect extent). The pre-defined machine learning classifier is typically a Support Vector Machine (SVM) or a Convolutional Neural Network (CNN). Network (CNN); The wear feature database is a pre-built standard database containing different wear types and corresponding time-domain feature samples, such as abrasive wear, fatigue spalling, adhesive wear, etc. Each sample is clearly labeled with wear type and parameters; The extracted time-domain features are input into a classifier, which outputs the wear type and corresponding position distribution on the surface of the titanium alloy bar by matching similarity with the database samples. It should be noted that the position distribution is converted through the time dimension of the signal. Based on the ultrasonic propagation speed and signal propagation time, the distance between the defect and the detection probe is calculated, thereby determining the position coordinates of the wear on the surface of the bar.

[0146] Finally, a 3D imaging algorithm is used to process the identified wear types and location distributions to generate a wear morphology map of the titanium alloy bar surface. The 3D imaging algorithm is typically a volume rendering algorithm or a surface rendering algorithm. Volume rendering algorithms can directly render the 3D data of the wear area to generate an intuitive 3D image; surface rendering algorithms require first constructing a surface model of the wear area (such as a triangular facet model), and then performing texture mapping and lighting processing on the model. The generated wear morphology map must include the 3D coordinates of the wear area (X, Y, and Z axes correspond to the length, circumference, and depth directions of the bar, respectively), wear depth (the maximum difference in the Z-axis direction, reflecting the severity of wear), and wear area (the projected area of ​​the wear area on the XY plane, reflecting the coverage of wear). This morphology map can intuitively display the wear state of the titanium alloy bar surface, providing data support for subsequent wear assessment and repair plan development.

[0147] Embodiment two:

[0148] Please refer to Figure 2 It shows a titanium alloy bar surface wear morphology detection system provided by an embodiment of the application, and the system comprises:

[0149] The acquisition module 20 is configured to acquire one-dimensional time-domain ultrasonic signals by performing non-contact scanning on the titanium alloy bar surface through a preset laser ultrasonic scanning system, and decompose the ultrasonic signals into a plurality of IMF components.

[0150] The determination module 21 is configured to determine a point oscillation persistence factor of each signal point in each of the IMF components.

[0151] The calculation module 22 is configured to calculate an oscillation discreteness index of the point oscillation persistence factor in each window based on time-domain window division, and calculate a frequency-domain sharpness of each IMF component and the original signal through frequency energy distribution comparison.

[0152] The generation module 23 is configured to fuse the oscillation discreteness index and the frequency-domain sharpness, generate an adaptive wavelet threshold parameter, and denoise each IMF component by using the threshold parameter to reconstruct an enhanced ultrasonic signal.

[0153] The identification module 24 is configured to identify a wear type of the scanned titanium alloy bar and generate a titanium alloy bar surface wear morphology map based on the enhanced ultrasonic signal.

[0154] Optionally, the transmission medium can be a wired link, such as but not limited to a coaxial cable, an optical fiber, a digital subscriber line, etc., or a wireless link, such as but not limited to Wireless Fidelity (WIFI), Bluetooth, and a mobile device network, etc.

[0155] It should be noted that the apparatus provided in the above embodiments is only exemplified by the division of the above functional modules, and in actual applications, the above functions can be completed by different functional modules according to needs, that is, the internal structure of the computer device is divided into different functional modules to complete all or part of the functions described above.

[0156] Figure 3 is a structural schematic diagram of a computer device provided by an embodiment of the application. As shown in the example of Figure 3 The computer device 30 comprises a memory 31, a processor 32, and a computer program 33 stored in the memory 31 and running on the processor 32, wherein the processor 32 executes the computer program 33, so that the computer device can execute any one of the titanium alloy bar surface wear morphology detection methods introduced above.

[0157] In addition, the embodiment of the present application also protects a device, which can include a memory and a processor, wherein the memory stores executable program code, and the processor is configured to invoke and execute the executable program code to perform the titanium alloy bar surface wear morphology detection method provided by the embodiment of the present application.

[0158] The embodiment of the present application can divide the device into functional modules according to the above method examples, for example, each functional module can be corresponding, or two or more functions can be integrated in one processing module, and the integrated module can be realized in the form of hardware. It should be noted that the division of the modules in the present embodiment is illustrative, and is only a logical function division, and another division mode can be used in actual implementation.

[0159] It should be understood that the device provided by the embodiment of the present application is used to perform the above titanium alloy bar surface wear morphology detection method, and thus the same effect as the above implementation method can be achieved.

[0160] In the case of using integrated units, the device can include a processing module and a storage module. When the device is applied to equipment, the processing module can be used to control and manage the actions of the equipment. The storage module can be used to support the equipment to execute mutual program codes and the like. The processing module can be a processor or a controller, which can realize or execute various exemplary logical blocks, modules and circuits described in combination with the disclosure. The processor can also be a combination of computing functions, such as one or more microprocessor combinations, a combination of digital signal processing (Digital Signal Processing, DSP) and microprocessors, and the like, and the storage module can be a memory.

[0161] In addition, the device provided by the embodiment of the present application can be a chip, an assembly or a module, the chip can include a connected processor and a memory; wherein the memory is used to store instructions, when the processor invokes and executes the instructions, the chip can execute the titanium alloy bar surface wear morphology detection method provided by the above embodiment.

[0162] The embodiment of the present application also provides a computer readable storage medium, which stores computer program code, when the computer program code runs on the computer, the computer executes the above related method steps to realize the titanium alloy bar surface wear morphology detection method provided by the above embodiment.

[0163] The embodiment of the present application also provides a computer program product, when the computer program product runs on the computer, the computer executes the above related steps to realize the titanium alloy bar surface wear morphology detection method provided by the above embodiment.

[0164] Among them, the device, computer readable storage medium, computer program product or chip provided by the embodiment of the application are used to execute the corresponding method provided above, so the beneficial effects that can be achieved are referable to the beneficial effects of the corresponding method provided above, which will not be repeated here. Through the description of the above implementation, those skilled in the art can understand that, for the convenience and brevity of description, only the above-mentioned division of functional modules is taken as an example, and in actual application, the above-mentioned functions can be completed by different functional modules according to needs, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In the embodiments provided by the application, it should be understood that the disclosed device and method can be implemented in other ways.

[0165] The device embodiment described above is only schematic, for example, the division of modules or units is only a logical function division, and actual implementation can have another division manner, for example, a plurality of units or components can be combined or integrated into another device, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the shown or discussed each other can be indirect coupling or communication connection through some interface, device or unit, which can be electrical, mechanical or other forms.

[0166] It should be further understood that the terms "comprise", "comprise", or any other variant thereof are intended to cover non-exclusive inclusion, so that processes, methods, articles or terminal devices including a series of elements not only include those elements, but also include other elements not explicitly listed or inherent to such processes, methods, articles or terminal devices. Without more limitations, the element defined by the statement "comprises a" does not exclude the presence of additional identical elements in the process, method, article or terminal device comprising the element.

[0167] It should be noted that: the above-mentioned sequence of the embodiments of the application is only for description, not representing the advantages and disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are also possible or can be advantageous.

[0168] Each embodiment in the specification adopts a progressive description manner, and the same or similar parts between each embodiment can be referred to each other, and each embodiment mainly describes the difference from other embodiments.

[0169] The above merely illustrates the specific embodiments of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of the changes or replacements within the technical range disclosed by the present application, which should be covered in the protection scope of the present application.

Claims

1. A method for detecting surface wear topography of a titanium alloy bar, the method comprising: The method includes the following steps: ​ The surface of the titanium alloy rod is scanned non-contactly using a preset laser ultrasonic scanning system to obtain a one-dimensional time-domain ultrasonic signal, which is then decomposed into multiple IMF components. Determine the point oscillation duration factor for each signal point in each of the IMF components; Based on time-domain window division, the oscillation dispersion index of the oscillation duration factor within each window is calculated, and the frequency domain sharpness of each IMF component compared with the original signal is calculated by comparing the frequency domain energy distribution. By integrating the oscillation dispersion index and the frequency domain sharpness, an adaptive wavelet threshold parameter is generated, and the threshold parameter is used to denoise each IMF component to reconstruct an enhanced ultrasound signal. Based on the enhanced ultrasonic signal, the wear type of the scanned titanium alloy rod is identified and a wear morphology map of the titanium alloy rod surface is generated; The process of generating adaptive wavelet threshold parameters by fusing the oscillation dispersion index and frequency domain sharpness includes: obtaining the oscillation dispersion index and frequency domain sharpness corresponding to each IMF component; performing linear normalization on the oscillation dispersion index to obtain the baseline threshold corresponding to each IMF component; performing negative correlation normalization on the frequency domain sharpness to obtain the frequency domain correction coefficient corresponding to each IMF component; and fusing the baseline threshold and the corresponding frequency domain correction coefficient for each IMF component to determine the adaptive wavelet threshold parameter corresponding to each IMF component.

2. The method of claim 1, wherein the method is characterized by: The step of performing a non-contact scan of the surface of the titanium alloy rod using a preset laser ultrasonic scanning system to obtain a one-dimensional time-domain ultrasonic signal includes: Start the preset laser ultrasound scanning system, which includes a pulsed laser excitation module for exciting ultrasound waves and a signal receiving module for receiving ultrasound echo signals; The pulsed laser excitation module is controlled to emit pulsed laser light onto the surface of the titanium alloy rod to generate ultrasonic waves on the surface of the titanium alloy rod. The signal receiving module receives the ultrasonic echo signal reflected from the surface of the titanium alloy rod. The ultrasonic echo signal received by the signal receiving module is sampled and processed at a preset sampling frequency, and the sampled ultrasonic echo signal is converted into a one-dimensional time-domain ultrasonic signal.

3. The method of claim 1, wherein the method is characterized by: The process of decomposing the one-dimensional time-domain ultrasound signal into multiple IMF components includes: The one-dimensional time-domain ultrasound signal is adaptively decomposed into multiple IMF components using the empirical mode decomposition method. During the decomposition process, the residual terms corresponding to the one-dimensional time-domain ultrasonic signal are acquired simultaneously.

4. The method of claim 1, wherein the method is characterized by: The determination of the point oscillation duration factor for each signal point in each of the IMF components includes: Select any one of the multiple IMF components obtained from the decomposition as the current target IMF component to be calculated; Identify all signal points in the current target IMF component that require calculation of the oscillation persistence factor; For each signal point in the current target IMF component, the absolute amplitude of the signal point and the absolute amplitude of each of the two adjacent signal points before and after the signal point are obtained respectively. calculating a first absolute value of an absolute amplitude change between the current signal point and a previous adjacent signal point, and simultaneously calculating a second absolute value of an absolute amplitude change between the current signal point and a subsequent adjacent signal point; calculating a point oscillation persistence factor of the current signal point according to the first absolute value and the second absolute value, and sequentially calculating point oscillation persistence factors of all signal points in the current target IMF component.

5. The method of claim 1, wherein the method is characterized by: The oscillation discreteness index of the point oscillation persistence factor in each window is calculated based on the time domain window division, including: obtaining the instantaneous frequency of each IMF component, and determining the time domain window length according to the instantaneous frequency; performing window division on each IMF component and the one-dimensional time domain ultrasonic signal respectively according to the determined time domain window length, to obtain a plurality of independent windows; for each independent window and the corresponding IMF component after division, selecting an arbitrary signal point in the independent window, and constructing a local neighborhood comprising the signal point and a preset number of adjacent signal points before and after the signal point; calculating the average value of the point oscillation persistence factors of all signal points in each local neighborhood as the average value of the local neighborhood factor corresponding to the local neighborhood; counting the number of signal points in each local neighborhood whose point oscillation persistence factor is greater than the average value of the local neighborhood factor corresponding to the local neighborhood, and calculating the weight of the corresponding signal point in the local neighborhood based on the counted number of signal points; calculating the average value of the point oscillation persistence factors of all signal points in each independent window, and combining the weight of each signal point in the independent window, and the deviation degree of the point oscillation persistence factor and the average value, to obtain the oscillation discreteness of each independent window; for each IMF component, calculating the ratio of the standard deviation to the average value of the point oscillation persistence factor in each independent window corresponding to the IMF component, and obtaining the discrete contribution value of each independent window through the ratio and the oscillation discreteness; summing the discrete contribution values of all independent windows corresponding to each IMF component, and then dividing by the total number of independent windows corresponding to the IMF component, to obtain the oscillation discreteness index of each IMF component.

6. The method of claim 1, wherein the method is characterized by: The frequency domain sharpness of each IMF component and the original signal is calculated through the comparison of the frequency energy distribution, including: performing Fourier transform on the one-dimensional time domain ultrasonic signal to obtain the frequency spectrum of the one-dimensional time domain ultrasonic signal, and calculating the energy probability density function of the one-dimensional time domain ultrasonic signal based on the frequency spectrum; performing Fourier transform on each IMF component to obtain the frequency spectrum of each IMF component, and calculating the energy probability density function corresponding to each IMF component based on the frequency spectrum of each IMF component; for each IMF component, calculating the absolute difference value between the energy probability density function corresponding to the IMF component and the energy probability density function of the one-dimensional time domain ultrasonic signal at each frequency point to form a difference vector value; summing the difference vector values at all frequency points to obtain the total difference value in the frequency energy distribution between the IMF component and the one-dimensional time domain ultrasonic signal. The spectrum of the one-dimensional time domain ultrasonic signal is normalized to obtain a normalized amplitude spectrum of the one-dimensional time domain ultrasonic signal, and a mean value of the normalized amplitude spectrum is calculated; The absolute difference value between the normalized amplitude spectrum of the one-dimensional time domain ultrasonic signal and the mean value of the normalized amplitude spectrum at each frequency point is calculated, and the absolute difference value is summed by traversing all frequency points to obtain a calculation denominator; The frequency domain sharpness of the IMF component is determined by the total difference value and the calculation denominator.

7. The method of claim 3, wherein the method is characterized by: The threshold parameter is used to denoise each IMF component to reconstruct an enhanced ultrasonic signal, including: The adaptive wavelet threshold parameter corresponding to each IMF component is called, and wavelet transform is performed on the current IMF component to obtain corresponding wavelet coefficients; The wavelet coefficients are compared with the adaptive wavelet threshold parameter, and the wavelet coefficients smaller than the adaptive wavelet threshold parameter are removed, and the retained wavelet coefficients are inversely wavelet transformed to obtain the denoised current IMF component; The residual term is called, and all denoised current IMF components and the corresponding residual term jointly participate in signal reconstruction, and the denoised current IMF components and the residual term are integrated by Hilbert transform to obtain the reconstructed enhanced ultrasonic signal.

8. The method of claim 1, wherein the method is characterized by: The enhanced ultrasonic signal is used to identify the wear type and generate a titanium alloy bar surface wear morphology map, including: The Hilbert transform is performed on the enhanced ultrasonic signal to construct a complex signal with the enhanced ultrasonic signal as the real part and the transform result as the imaginary part; The modulus of the complex signal is calculated to obtain an envelope signal corresponding to the enhanced ultrasonic signal; Time domain features are extracted from the envelope signal, and the extracted time domain features are input into a preset machine learning classifier and matched with a preset wear feature database, and the matching result is used to identify the wear type and corresponding position distribution on the titanium alloy bar surface; A three-dimensional imaging algorithm is used to process the identified wear type and position distribution to generate a titanium alloy bar surface wear morphology map.

9. A titanium alloy bar surface wear topography detection system, characterized by, The system includes the following modules: An acquisition module is configured to acquire a one-dimensional time domain ultrasonic signal by performing non-contact scanning on a titanium alloy bar surface using a preset laser ultrasonic scanning system, and decompose the ultrasonic signal into multiple IMF components; A determination module is configured to determine a point oscillation persistence factor of each signal point in each IMF component; A calculation module is configured to calculate an oscillation discreteness index of the point oscillation persistence factor in each window based on time domain window division, and calculate a frequency domain sharpness of each IMF component and an original signal by comparing frequency energy distributions; A generation module is configured to fuse the oscillation discreteness index and the frequency domain sharpness to generate an adaptive wavelet threshold parameter, and use the threshold parameter to denoise each IMF component to reconstruct an enhanced ultrasonic signal. The fusion of the oscillation discreteness index and the frequency domain sharpness generates an adaptive wavelet threshold parameter, including: obtaining the oscillation discreteness index and the frequency domain sharpness corresponding to each IMF component; performing linear normalization processing on the oscillation discreteness index to obtain a reference threshold value corresponding to each IMF component; performing negative correlation normalization processing on the frequency domain sharpness to obtain a frequency domain correction coefficient corresponding to each IMF component; and fusing the reference threshold value corresponding to each IMF component with the corresponding frequency domain correction coefficient to determine an adaptive wavelet threshold parameter corresponding to each IMF component; The identification module is configured to identify the wear type of the titanium alloy bar based on the enhanced ultrasonic signal and generate a titanium alloy bar surface wear morphology diagram.

Images