An ultrasonic flaw detection method, system and device for isostatic pressed graphite finished products

Abstract

The present application relates to the technical field of physical testing, in particular to an ultrasonic flaw detection method, system and equipment for isostatic pressing graphite finished products, the method comprising: determining local fluctuation synchronization features of a target position based on ultrasonic time domain signals of the isostatic pressing graphite finished product at the target position and its spatiotemporal neighborhood positions; determining multi-scale scattering correlation features of the target position by using mutual information between adjacent scale component signal features of the ultrasonic time domain signals; determining scattering events based on local signal-to-noise ratios of the ultrasonic time domain signals, determining event state transition probability matrices based on scattering event features, and obtaining sequence signal statistical evolution features of the target position; determining a comprehensive structure continuity parameter of the target position by combining the above multiple features of the target position, and obtaining a flaw detection result by adjusting separation constraint criteria of ICA based on the comprehensive structure continuity parameter. Through the technical scheme of the present application, substantial improvement in accuracy and reliability of graphite material defect detection is achieved.

Description

Technical Field

[0001] This invention relates to the field of physical testing technology, specifically to an ultrasonic flaw detection method, system, and equipment for isostatically pressed graphite products. Background Technology

[0002] As a key basic material, the accurate detection of internal defects (such as cracks, pores, and inclusions) in isostatically pressed graphite is a core technological bottleneck for ensuring the reliability of high-end equipment. Currently, ultrasonic testing technology is widely used in the industry for internal quality assessment. Existing mainstream solutions mainly rely on classical signal processing methods, such as time-domain thresholding, spectral analysis, and blind source separation techniques represented by ICA (Independent Component Analysis). Among them, ICA is highly anticipated due to its ability to separate potential source signals from mixed observation signals.

[0003] However, when dealing with materials like isostatically pressed graphite, which exhibit significant internal inhomogeneity, existing ICA techniques face a fundamental challenge: their core assumption of "statistical independence of source signals" is severely incompatible with the physical properties of graphite as a continuous medium composed of micropores and density gradients. This results in the inability to effectively separate defect signals from background structural noise when directly applying traditional ICA algorithms, leading to severe "ghosting" and compositional confusion, which greatly interferes with defect identification and localization, reducing the accuracy and reliability of defect detection. Existing attempts to improve ICA have largely focused on adjusting fixed parameters or simple preprocessing, failing to fundamentally resolve the core contradiction of non-independent source signals. Summary of the Invention

[0004] To address the low accuracy and reliability of existing ICA methods for defect detection in isostatically pressed graphite products, this invention aims to provide an ultrasonic flaw detection method, system, and equipment for isostatically pressed graphite products. The specific technical solution adopted is as follows:

[0005] This invention provides an ultrasonic flaw detection method for isostatically pressed graphite products, the method comprising:

[0006] Based on the ultrasonic time-domain signals of isostatically pressed graphite products at the target location and its spatiotemporal neighborhood, the signal fluctuation characteristics differences between the two locations are determined to obtain the local fluctuation synchronization characteristics of the target location.

[0007] By utilizing the mutual information between the adjacent scale component signals of the ultrasonic time-domain signal at the target location, the multi-scale scattering correlation characteristics at the target location are determined;

[0008] The scattering event is determined based on the local signal-to-noise ratio of the ultrasonic time-domain signal at the target location. The state transition probability matrix of the event state evolution process is determined based on the characteristics of the scattering event, and the statistical evolution characteristics of the sequence signal related to the ultrasonic scattering process at the target location are obtained.

[0009] By combining the characteristics of local wave synchronicity, multi-scale scattering correlation, and statistical evolution of sequence signals, the comprehensive structural continuity parameter is determined. Based on the comprehensive structural continuity parameter, the separation constraint criterion of ICA is adjusted to obtain the flaw detection results.

[0010] Furthermore, the determination of the local fluctuation synchronization characteristics of the target location based on the ultrasonic time-domain signals of the isostatically pressed graphite product at the target location and its spatiotemporal neighborhood locations, and the difference in signal fluctuation characteristics between the two locations, includes:

[0011] The signal envelope of the ultrasonic time-domain signal of the isostatically pressed graphite product at the target location is determined, and the envelope curve of the signal energy changing with time is obtained using the signal envelope;

[0012] Perform a first-order difference operation on the envelope curve to obtain a difference sequence characterizing the signal fluctuation mode, and determine the fluctuation mode feature vector composed of the difference sequence.

[0013] By utilizing the wave pattern feature vectors of the target location and its spatiotemporal neighborhood location, the differences in signal wave characteristics between the target location and its spatiotemporal neighborhood location are obtained;

[0014] By utilizing the differences in signal fluctuation characteristics and the set feature parameters corresponding to the spatiotemporal neighborhood location set, the local fluctuation synchronization characteristics of the target location are obtained.

[0015] Furthermore, the step of determining the multi-scale scattering correlation features of the target location by utilizing the mutual information between the signal features of adjacent scale components of the ultrasonic time-domain signal at the target location includes:

[0016] The ultrasonic time-domain signal at the target location is decomposed into multiple scale component signals, and the characteristics of the target scale component signals based on the signal energy and oscillation frequency characteristics are determined.

[0017] The mutual information between the target scale component signal and the scale component signal of its neighboring scale components is determined, and the multi-scale scattering correlation characteristics of the target location are determined by using the mutual information.

[0018] Furthermore, the multi-scale scattering correlation features used to determine the target location using various mutual information include:

[0019] Determine the energy vector formed by the signal energy corresponding to each of the scale component signals, and determine the relative dispersion of the energy vector;

[0020] By combining the mutual information and the relative degree of dispersion, the multi-scale scattering correlation characteristics of the target location are determined.

[0021] Furthermore, the determination of scattering events based on the local signal-to-noise ratio of the ultrasonic time-domain signal at the target location includes:

[0022] Determine the local signal-to-noise ratio between the local peak value and local noise of the ultrasonic time-domain signal at the target location within a preset time window;

[0023] If the local signal-to-noise ratio is greater than a preset dynamic threshold related to local noise, the corresponding local peak is marked as a scattering event.

[0024] Furthermore, the determination of the state transition probability matrix of the event state evolution process based on the scattering event characteristics, and the acquisition of the statistical evolution characteristics of the sequence signal related to the ultrasonic scattering process at the target location, include:

[0025] The scattering events are determined based on the scattering event characteristics consisting of their corresponding arrival time and peak intensity, and the scattering event sequence consisting of each scattering event characteristic at the target location is determined.

[0026] Based on the scattering event sequences at each location, clustering is used to determine the event state sequence of the event state evolution process at the target location;

[0027] Based on the event state sequence, the corresponding state transition probability matrix and steady-state distribution parameters are determined. Using the state transition probability matrix and steady-state distribution parameters, the statistical evolution characteristics of the sequence signal related to the ultrasonic scattering process at the target location are obtained.

[0028] Furthermore, the determination of the comprehensive structural continuity parameter by combining local wave synchronicity characteristics, multi-scale scattering correlation characteristics, and sequence signal statistical evolution characteristics includes:

[0029] Determine the sensitivity of local wave synchronization characteristics to the correlation changes of multi-scale scattering correlation characteristics, and use the sensitivity of correlation changes to obtain the stability penalty factor of the relationship between local wave synchronization characteristics and multi-scale scattering correlation characteristics;

[0030] By combining the characteristics of local fluctuation synchronization, multi-scale scattering correlation, statistical evolution of sequence signals, and relationship stability penalty factor, a comprehensive structural continuity score for the target location is determined.

[0031] Furthermore, the flaw detection results obtained by adjusting the separation constraint criterion of ICA based on the comprehensive structural continuity parameter include:

[0032] Construct a positive correlation transformation relationship between the comprehensive structural continuity parameter and the ICA correlation coefficient tolerance threshold, and use the positive correlation transformation relationship to obtain the target correlation coefficient tolerance threshold of ICA at the target location;

[0033] The flaw detection results of the isostatically pressed graphite product are obtained by outputting independent components based on the tolerance threshold of the correlation coefficient at each location and the ICA output.

[0034] The present invention also provides an ultrasonic flaw detection system for isostatically pressed graphite products, the system being used to implement the ultrasonic flaw detection method for isostatically pressed graphite products as described in any of the preceding claims; the system includes:

[0035] The signal analysis module is used to determine the difference in signal fluctuation characteristics between the target location and its spatiotemporal neighboring locations based on the ultrasonic time-domain signals of the isostatically pressed graphite finished product at the target location and its spatiotemporal neighboring locations, thereby obtaining the local fluctuation synchronization characteristics of the target location; and to determine the multi-scale scattering correlation characteristics of the target location by utilizing the mutual information between the adjacent scale component signal characteristics of the ultrasonic time-domain signal at the target location.

[0036] The scattering event is determined based on the local signal-to-noise ratio of the ultrasonic time-domain signal at the target location. The state transition probability matrix of the event state evolution process is determined based on the characteristics of the scattering event, and the statistical evolution characteristics of the sequence signal related to the ultrasonic scattering process at the target location are obtained.

[0037] The component analysis module is used to combine local wave synchronicity characteristics, multi-scale scattering correlation characteristics, and sequence signal statistical evolution characteristics to determine the comprehensive structural continuity parameters. Based on the comprehensive structural continuity parameters, the separation constraint criteria of ICA are adjusted to obtain the flaw detection results.

[0038] The present invention also provides an ultrasonic flaw detection device for isostatically pressed graphite products. The device includes a processor, a memory, and an ultrasonic flaw detection program for isostatically pressed graphite products stored in the memory and executable by the processor. When the ultrasonic flaw detection program for isostatically pressed graphite products is executed by the processor, the steps of the ultrasonic flaw detection method for isostatically pressed graphite products as described in any of the preceding claims are implemented.

[0039] The present invention has the following beneficial effects:

[0040] This invention achieves adaptive optimization of Independent Component Analysis (ICA) based on the characterization of the internal structural continuity of graphite materials by constructing a deep analysis system encompassing local wave synchronicity, multi-scale scattering correlation, and scattering dynamics state transition parameters. Compared with existing technologies, this invention can dynamically adjust the separation constraint criteria of ICA according to the local structural continuity state of graphite reflected by ultrasonic signals (quantified by comprehensive structural continuity parameters): in continuous and uniform regions with high comprehensive structural continuity parameter values, the independence constraint is relaxed to suppress "ghosting" noise caused by over-separation, preserving the natural correlation of background scattering; in suspected defect regions with low comprehensive structural continuity parameter values, the independence constraint is strengthened to force abnormal scattering signals to be clearly separated from the strongly correlated background. This intelligent modulation, bound to the physical properties and state of the material, significantly improves the signal-to-noise ratio and purity of the separated signals, enhances the detection sensitivity for small and latent defects, and ensures the fidelity of defect morphology and location information, effectively solving the problem of component confusion and misjudgment caused by the non-independent source signals in graphite-like non-homogeneous materials under traditional fixed-criteria ICA. This method achieves a substantial improvement in the accuracy and reliability of defect detection by establishing an intrinsic relationship between the continuity of material structure and signal separation constraints. Attached Figure Description

[0041] 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.

[0042] Figure 1 A flowchart illustrating the steps of an ultrasonic flaw detection method for isostatically pressed graphite products according to an embodiment of the present invention;

[0043] Figure 2 This is a detailed flowchart of step S1 in an ultrasonic flaw detection method for isostatically pressed graphite products according to an embodiment of the present invention.

[0044] Figure 3 This is a detailed flowchart of step S2 in an ultrasonic flaw detection method for isostatically pressed graphite products according to an embodiment of the present invention.

[0045] Figure 4 This is a detailed flowchart of step S3 in an ultrasonic flaw detection method for isostatically pressed graphite products according to an embodiment of the present invention.

[0046] Figure 5 This is a detailed flowchart of step S4 in an ultrasonic flaw detection method for isostatically pressed graphite products according to an embodiment of the present invention.

[0047] Figure 6 This is a schematic diagram of the hardware operating environment of the ultrasonic flaw detection equipment for isostatically pressed graphite products involved in the embodiments of the present invention.

[0048] Figure 7 This is a schematic diagram of the frame structure of the ultrasonic flaw detection system for isostatically pressed graphite products involved in the embodiments of the present invention. Detailed Implementation

[0049] Before proceeding with the following embodiments of the present invention, it is necessary to explain the main objectives of the present invention in order to facilitate understanding of the overall technical concept of the present invention:

[0050] This invention aims to address the technical challenges of existing ultrasonic flaw detection methods for isostatic graphite based on independent component analysis (ICA), which suffer from non-independent source signals due to neglecting the continuity of the material's internal structure. These challenges result in impure defect separation, severe image ghosting interference, and low reliability of detection results. Specifically, traditional ICA and its improved versions employ fixed statistical independence criteria, failing to adapt to changes in the structural state of graphite from uniform regions to defect edges, leading to a disconnect between algorithm parameters and physical reality. The core objective of this invention is to provide a novel ICA optimization mechanism that quantifies and incorporates prior knowledge of the material's structural continuity. By constructing adaptive parameters dynamically bound to the local structural state, it achieves intelligent adjustment during signal separation, "forcing independence when necessary and allowing correlation when possible." This enables clear and pure extraction of defect signals even under complex background noise, significantly improving the accuracy and robustness of flaw detection.

[0051] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of an ultrasonic flaw detection method for isostatically pressed graphite products proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0052] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0053] The specific scheme of the ultrasonic flaw detection method for isostatically pressed graphite products provided by the present invention will be described in detail below with reference to the accompanying drawings.

[0054] Example 1:

[0055] For the ultrasonic flaw detection method of isostatically pressed graphite products provided by this invention, please refer to [link to relevant documentation]. Figure 1 The diagram illustrates a flowchart of the ultrasonic flaw detection method for isostatically pressed graphite products according to an embodiment of the present invention.

[0056] The method includes:

[0057] Step S1: Based on the ultrasonic time-domain signals of the isostatically pressed graphite product at the target location and its spatiotemporal neighborhood, determine the signal fluctuation characteristics difference between the two locations to obtain the local fluctuation synchronization characteristics of the target location.

[0058] In this embodiment, a multi-probe array acoustic scanning method can be used. The probe elements are arranged in a predetermined two-dimensional grid and scan the graphite product point by point under the drive of the control system. Each element acts as both an exciter and a receiver at the scanning point, emitting a narrow ultrasonic pulse with a specific center frequency (usually in the range of 0.5MHz to 5MHz, selected according to the graphite thickness and acoustic attenuation characteristics) using the pulse-echo method. The receiving stage records the full waveform time-domain signal of each scanning point (corresponding to a three-dimensional spatial coordinate), that is, the complete scanning signal including longitudinal waves, possible transverse waves, and multiple reflected echoes. The acquisition parameters are optimized, including a sufficient sampling rate (usually not less than 20MHz to ensure time resolution), an appropriate pulse repetition frequency, and a gain adjusted according to the signal dynamic range. Finally, the acquisition system outputs a three-dimensional data cube, with two dimensions corresponding to the scanning plane position and the third dimension being the time series. This data volume completely preserves the dynamic process information of ultrasonic wave propagation and scattering inside the graphite, providing the original data foundation for subsequent multi-level feature analysis.

[0059] Ultrasonic signals are continuous in both space and time. Directly performing a global analysis of the entire detection area of ​​the graphite product will mask local structural features. For example, signal abrupt changes at the edge of a defect will be smoothed out by the stable signal in a uniform region. Therefore, before performing local synchronization calculations, it is necessary to construct reasonable neighborhood relationships and data representation methods.

[0060] The entire detection area is divided into grids according to the spatial scanning path of the ultrasonic probe. Each grid node corresponds to a three-dimensional spatial coordinate position. Any three-dimensional spatial coordinate position is taken as the target position, and the complete time-domain ultrasonic signal acquired at the target position is recorded. Similarly, the time-domain ultrasonic signal of its spatiotemporal neighborhood position is also obtained.

[0061] In this process, a 3×3 grid of points centered on the target location is selected in space, and adjacent sampling points before and after the target time (e.g., one point before and one point after) are selected on the time axis. Together, they form a four-dimensional data subset, which can be called the spatiotemporal neighborhood.

[0062] Specifically, please refer to Figure 2 Step S1 includes:

[0063] Step S11: Determine the signal envelope of the ultrasonic time-domain signal of the isostatically pressed graphite product at the target location, and use the signal envelope to obtain the envelope curve of the signal energy changing with time.

[0064] Step S12: Perform a first-order difference operation on the envelope curve to obtain a difference sequence characterizing the signal fluctuation mode, and determine the fluctuation mode feature vector composed of the difference sequence.

[0065] Step S13: Using the wave pattern feature vectors of the target location and its spatiotemporal neighborhood location respectively, obtain the signal wave feature differences between the target location and the spatiotemporal neighborhood location.

[0066] Step S14: Using the differences in signal fluctuation characteristics and the set feature parameters corresponding to the spatiotemporal neighborhood location set, the local fluctuation synchronization characteristics of the target location are obtained.

[0067] In this embodiment, for each spatial location, taking the target location as an example, a Hilbert transform is performed on its original time-domain signal to extract the signal envelope, and the envelope curve of the signal energy changing with time is obtained, thereby removing the periodic influence of the carrier frequency and highlighting the actual trend of signal energy change.

[0068] Furthermore, a first-order difference operation is performed on the extracted envelope curve to calculate the envelope change rate at each time step, resulting in a difference sequence describing the signal fluctuation pattern. This processing is based on the physical characteristic that ultrasonic signals exhibit gradual energy changes when propagating in a homogeneous medium, but experience dramatic energy abrupt changes at defect boundaries. By capturing the differences in the envelope change rate, the fluctuation pattern is quantified.

[0069] Based on this, a spatiotemporal neighborhood (location) set is defined for the target location. This neighborhood not only includes adjacent grid points in the three-dimensional spatial direction (e.g., using 6-neighbor connections), but also adjacent sampling points in the temporal dimension, i.e., a four-dimensional spatiotemporal neighborhood set. This is because the structural continuity of graphite materials is reflected not only in spatially adjacent positions, but also in the temporal evolution process. The existence of defects will simultaneously disrupt the signal consistency in both space and time.

[0070] Before calculating the local fluctuation synchronization characteristics, for data points located at the scan edge, a mirror filling method is used to process the boundary to ensure the effectiveness of the neighborhood calculation.

[0071] For any grid point located at coordinate P (corresponding to the target location), calculate its local fluctuation synchronization characteristics:

[0072]

[0073] in, Represents the four-dimensional spatiotemporal neighborhood set of the target location P (from which the number of locations in the set is taken). ), which includes all data points that are adjacent in both spatial and temporal dimensions; The first-order difference sequence representing the signal envelope at position P is used to obtain the wave pattern feature vector at that target location. The feature vector representing the fluctuation pattern of Q at other locations within the neighborhood (i.e., any spatiotemporal neighborhood location); Represents the Euclidean distance, used to quantify the degree of difference between two fluctuation modes, i.e., the difference in signal fluctuation characteristics; The kernel bandwidth parameter is typically taken as the average variance of all differential sequences in the neighborhood, and is used to adaptively adjust the scale of the similarity metric. This represents a very small positive number, a safety value set to prevent the denominator from being 0. Specifically, its value can be 0.1, and L is the length of the difference sequence.

[0074] By quantifying the similarity between the target location and the fluctuation patterns of all neighboring locations, and then mapping them to an exponential function, the system can be optimized. The interval is then averaged to obtain the local fluctuation synchronization score. Before the subsequent feature fusion step, the... Perform Min-Max normalization to map its values ​​to The interval ensures that the characteristic values ​​are non-negative and have consistent dimensions.

[0075] In structurally continuous graphite regions, ultrasonic signals at adjacent locations are influenced by similar microstructures, exhibiting highly consistent envelope variation patterns, small distance differences, and exponential function values ​​close to 1. High scores; however, in defect boundaries or structurally anomalous regions, abrupt changes in the signal envelope lead to significant differences in fluctuation patterns, large distance values, and exponential function values ​​approaching 0, thus lowering the overall score. score.

[0076] Step S2: Utilize the mutual information between the adjacent scale component signal features of the ultrasonic time-domain signal at the target location to determine the multi-scale scattering correlation features at the target location;

[0077] Analysis shows that while single-scale synchronicity analysis can initially characterize structural discontinuities, it is difficult to distinguish the essential differences in the impact of uniform micropore distribution and macroscopic defects on acoustic wave scattering. For example, in graphite regions with uniform micropores, local fluctuations may exhibit good synchronicity at a single analytical scale, similar to the characteristics of regions with small defects, which can easily lead to misjudgment.

[0078] Specifically, please refer to Figure 3 Step S2 includes:

[0079] Step S21: Scale decomposition is performed on the ultrasonic time-domain signal at the target location to obtain multiple scale component signals, and the target scale component signal characteristics based on signal energy and oscillation frequency characteristics are determined.

[0080] Step S22: Determine the mutual information between the target scale component signal and the scale component signal of its neighboring scale components, and use the mutual information to determine the multi-scale scattering correlation features of the target location.

[0081] More specifically, step S22, which utilizes the mutual information to determine the multi-scale scattering correlation features of the target location, includes:

[0082] Determine the energy vector formed by the signal energy corresponding to each of the scale component signals, and determine the relative dispersion of the energy vector;

[0083] By combining the mutual information and the relative degree of dispersion, the multi-scale scattering correlation characteristics of the target location are determined.

[0084] Based on the above embodiments, in this embodiment, for the target location The original ultrasonic time-domain signal was extracted. Adaptive scaling decomposition was then performed on this time-domain signal to separate the different time-scale components contained within the signal.

[0085] Specifically, an adaptive wavelet packet decomposition method can be used: calculate the information entropy of the signal at different preset decomposition depths, and when the information entropy gain brought by further decomposition is lower than a preset threshold (e.g., setting the relative gain to be less than a certain threshold), the decomposition method can be applied. When the decomposition reaches a certain level, the decomposition stops, and the number of decomposition layers determined at this point is the adaptive scale number of the signal at that location. Adaptive decomposition level It must be at least 2 (i.e., contain at least 2 scale components).

[0086] The decomposition results Feature extraction is performed on each of the scale component signals. For the _th ... For each scale component (as the target scale component), calculate its two core statistical characteristics: first, the normalized energy of that scale component. The first is the proportion of the total energy of the signal component to the total energy of the original signal, used to characterize the intensity contribution of the scale component; the second is the zero-crossing rate of the scale component. This refers to the number of times the signal crosses the zero level per unit time, used to characterize the oscillation frequency characteristics of that scale component. The features of each scale component are combined into a feature vector. That is, the target scale component signal characteristics.

[0087] Then, the mutual information between eigenvectors of adjacent scales is calculated. Mutual information can capture nonlinear statistical dependencies and is better suited than simple correlation coefficients to describe the characteristic correlations generated by complex physical processes such as ultrasonic scattering.

[0088] Specifically, the calculation starts from the first... Scale to the Scale mutual information It measures the reduction in uncertainty about one scale feature when the other scale feature is known. Mutual information It is calculated based on the statistical distribution of the feature vector sample set of the target location P and all grid points in its local spatial neighborhood.

[0089] Construct target location Multiscale scattering correlation features The core formula is as follows:

[0090]

[0091] in, These are weighting coefficients assigned to different scale pairs, using an exponential decay form to emphasize finer scales. The importance of correlations between values ​​that are relatively small. It is a vector composed of energies at various scales, i.e., an energy vector. It is its variance. It is the maximum value among all energy variance values ​​calculated from the full-field scan data; the ratio of these two values ​​represents the relative dispersion or the energy variation coefficient. The product structure in the formula has a clear physical logic: the first part (weighted mutual information sum) reflects the overall coherence of cross-scale feature transmission; a higher value indicates greater self-similarity in scattering behavior at each scale; the second part... As a penalty term, it reflects the balance of energy distribution across scales. When energy is abnormally prominent at a certain scale, this term decreases, thereby reducing the overall energy level. score, This represents a very small positive number. It is a safety value set to prevent the denominator from being 0. Specifically, its value can be 0.1.

[0092] This theory posits that in an ideal graphite region with a continuous and homogeneous structure, acoustic wave scattering should exhibit statistical self-similarity at different scales, meaning strong correlation between scale features (high mutual information) and relatively balanced energy distribution (small energy variance). The value is relatively high. However, in regions with defects, the defects, acting as strong scatterers, disrupt this multi-scale self-similarity, leading to anomalous energy concentration at specific scales (e.g., macroscopic defects primarily affect coarse-scale areas) and the disruption of cross-scale correlation patterns, thus... The value decreased significantly.

[0093] Before the subsequent feature fusion step, the calculated... Perform Min-Max normalization to map its values ​​to The interval ensures that all characteristic values ​​are non-negative and have consistent dimensions.

[0094] Step S3: Determine the scattering event based on the local signal-to-noise ratio of the ultrasonic time-domain signal at the target location, determine the state transition probability matrix of the event state evolution process based on the characteristics of the scattering event, and obtain the statistical evolution characteristics of the sequence signal related to the ultrasonic scattering process at the target location.

[0095] Analysis revealed that while the static correlation characteristics across scales can distinguish between intrinsic homogeneity and macroscopic defects in materials, they still cannot fully characterize the temporal evolution of the ultrasonic scattering signal waveform. For example, a microcrack in the process of initiation and propagation and a stable pore group may exhibit similarities in multi-scale energy distribution, but their scattering events may have fundamentally different occurrence patterns, intensity evolutions, and intervals over time.

[0096] Specifically, step S3, determining the scattering event based on the local signal-to-noise ratio of the ultrasonic time-domain signal at the target location, includes:

[0097] Determine the local signal-to-noise ratio between the local peak value and local noise of the ultrasonic time-domain signal at the target location within a preset time window;

[0098] If the local signal-to-noise ratio is greater than a preset dynamic threshold related to local noise, the corresponding local peak is marked as a scattering event.

[0099] Based on the above embodiments, in this embodiment, for the target location The ultrasonic time-domain signal is analyzed by a sliding window (the preset time window length can be adaptively set according to the signal center frequency). The ratio of the local peak value to the local noise floor (estimated by median) within the window is calculated, i.e., the local signal-to-noise ratio. When this ratio exceeds the dynamic threshold (the threshold can be set to 3 times the local noise standard deviation, and the specific multiple is adjusted according to the global signal-to-noise ratio), the local peak value is marked as a significant scattering event, or simply a scattering event.

[0100] Specifically, please refer to Figure 4 Step S3, based on the characteristics of the scattering event, determines the state transition probability matrix of the event state evolution process, and obtains the statistical evolution characteristics of the sequence signal related to the ultrasonic scattering process at the target location, including:

[0101] Step S31: Determine the scattering event characteristics based on the arrival time and peak intensity of the scattering events, and determine the scattering event sequence of the target location composed of the characteristics of each scattering event;

[0102] Step S32: Based on the scattering event sequence at each location, cluster the event state sequence of the event state evolution process at the target location;

[0103] Step S33: Determine the corresponding state transition probability matrix and steady-state distribution parameters based on the event state sequence, and use the state transition probability matrix and steady-state distribution parameters to obtain the statistical evolution characteristics of the sequence signal related to the ultrasonic scattering process at the target location.

[0104] In this embodiment, the arrival time of each (scattering) event is recorded. And peak intensity (amplitude) , constituting the characteristics of scattering events This constitutes a sequence of scattering events at the target location. The scattering event set represents the target location, where i is used as an index for the event and its state.

[0105] The statistical evolution of continuous physical quantities (time, intensity) into discrete sequence signals involves calculating two derived features for each event: one is the arrival time interval between the event i and the preceding event {i-1}. (For the first event, a preset baseline interval is used); the second is the normalized peak intensity of the event. This involves normalizing the event's peak intensity by dividing it by the maximum value of all event peak intensities at that location. Then, for the set of scattering events at all locations... K-means clustering is performed in a two-dimensional feature space. The number of clusters is automatically determined using the silhouette coefficient method to ensure the rationality of state partitioning. Each cluster center represents a typical statistical evolution state (such as "high intensity, long interval" or "low intensity, short interval"). Each event is assigned to a specific event state based on its characteristics. .

[0106] Then, based on each event state, an event state transition probability matrix describing the evolution of event states is constructed. For a given position... Its event state sequence is N represents the total number of elements. Next, calculate the transition count matrix for this event state sequence. , the elements Indicates from state Transition to state The number of times (not the index of i). To prevent inaccurate probability estimation due to data sparsity, a smoothing method of adding 1 (i.e., Laplace smoothing) is used, utilizing the transition count matrix. The transition probability matrix is ​​obtained from the values ​​(times) and number of elements in the matrix. The elements therein are represented as Simultaneously, the steady-state distribution parameters of the event state sequence are calculated. That is, by solving (The long-range probability distribution obtained by solving the steady-state distribution of a Markov chain, according to existing definitions)

[0107] Therefore, construct the target location Statistical evolution characteristics of sequence signals The core formula is as follows:

[0108]

[0109] in, Representing state The steady-state probability; Representing state The transition entropy describes the uncertainty of the next state when starting from a given state. This represents the total number of (event) states (which can be considered as N-1). The weighted average transition entropy of the entire state sequence; This represents the maximum possible entropy value, used to normalize the weighted average transfer entropy. The first part of the formula... Ensured The value is positively correlated with the orderliness (low entropy) of the process; that is, the lower the entropy value, the more ordered the process. The higher the value.

[0110] It should be noted that when the number of clusters S≥2, and the number of scattering events detected at the target location is lower than the preset threshold (e.g., 5), the statistical evolution feature D(P) of its sequence signal is directly determined to be the maximum value (e.g., 1), and no matrix calculation is performed.

[0111] Furthermore, the second part of the formula, Defined as the proportion of "long interval" events with time intervals exceeding twice the average interval in the scattering event sequence at the target location, it is used to quantify the non-uniformity of the temporal distribution. It is a scaling parameter, typically taken as the average time interval of all events at that location. (Exponential term) As a penalty factor: when the event time distribution is uniform ( When the interval is small, the factor is close to 1; when there are abnormally long intervals or event clusters ( When the value is large, this factor decreases significantly, thereby reducing... score.

[0112] Here, it is assumed that in the structurally continuous and stable graphite region, micropore scattering should be approximately a stationary random process in time, with relatively deterministic and ordered state transition modes (low normalized transition entropy), and relatively uniform event interval distribution. (small), therefore The value is relatively high. However, in the defect-active region, the scattering process may exhibit a chaotic transfer pattern (increased normalized transfer entropy) or an anomalous time interval distribution (…). (large), leading to The value decreased significantly.

[0113] Before the subsequent feature fusion step, the calculated... Perform Min-Max normalization to map its values ​​to The interval ensures that all characteristic values ​​are non-negative and have consistent dimensions.

[0114] Step S4: Combining the local wave synchronization characteristics, multi-scale scattering correlation characteristics, and sequence signal statistical evolution characteristics, determine the comprehensive structural continuity parameters, and adjust the separation constraint criteria of ICA based on the comprehensive structural continuity parameters to obtain the flaw detection results.

[0115] Specifically, please refer to Figure 5 Step S4, combining local wave synchronicity characteristics, multi-scale scattering correlation characteristics, and sequence signal statistical evolution characteristics, determines the comprehensive structural continuity parameters, including:

[0116] Step S41: Determine the sensitivity of the correlation change between the local wave synchronization feature and the multi-scale scattering correlation feature, and use the sensitivity of the correlation change to obtain the stability penalty factor of the relationship between the local wave synchronization feature and the multi-scale scattering correlation feature.

[0117] Step S42: Combining local wave synchronization characteristics, multi-scale scattering correlation characteristics, sequence signal statistical evolution characteristics, and relationship stability penalty factor, determine the comprehensive structural continuity score of the target location.

[0118] In this embodiment, the three basic features calculated from the above embodiments are... Standardization preprocessing is performed to eliminate fusion bias caused by differences in units and numerical ranges.

[0119] Furthermore, for the target location P, a 3x3 local window is defined centered on it, and calculations are performed within this window. The Pearson correlation coefficients of these three pairs of features were calculated and analyzed. The correlation is strong. Calculations are performed at position P. about The absolute value of the local gradient (or sensitivity) is denoted as the sensitivity to correlation changes (gradient value). ; This represents a very small positive number, a safety value set to prevent the denominator from being 0. Specifically, its value can be 0.1. It should be noted that all very small positive numbers in this embodiment of the invention are safety items set within the system, and their dimensions are consistent with the original denominator. The value of the very small positive number in different calculation processes can also be adjusted in different formulas according to actual detection needs, without limitation or elaboration. The corresponding gradient is estimated within a local window using the central difference method. Its physical meaning is that in a continuous structure, and Normally, with coordinated changes (e.g., regions with high synchronicity also have strong multi-scale correlations), gradient values ​​should remain within a low and stable range; however, when this coordinated relationship is disrupted (e.g., at the edge of a defect, synchronicity drops sharply and multi-scale correlations change asynchronously), gradient values ​​will exhibit abnormal peaks.

[0120] Based on the above analysis, a comprehensive structural continuity parameter for the target location P is constructed. The core formula is as follows:

[0121]

[0122] in, and These are the standardized local fluctuation synchronization characteristics and the statistical evolution characteristics of the sequence signal, respectively. Their product constitutes the molecular basis of the comprehensive structural continuity parameters, emphasizing the core contribution of simultaneously possessing high spatial synchronization and high dynamic order to structural continuity. It is the standardized multi-scale scattering correlation feature, used as the denominator. The logic is: when the multi-scale correlation is abnormally reduced ( (small value), even and The value is not low, which may indicate some kind of hidden structural mismatch (such as the mixing of scatterers of different scales). Therefore, it is necessary to reduce the overall value by using the denominator. score.

[0123] This represents a very small positive number, a safety value set to prevent the denominator from being 0. Specifically, its value can be 0.1. It should be noted that all very small positive numbers in the embodiments of this invention are safety items set within the system, and their dimensions are consistent with the original denominator. The value of very small positive numbers in different calculation processes can also be adjusted in different formulas according to actual detection needs, which will not be limited or elaborated.

[0124] It is a balancing factor used to control the strength of the gradient penalty term; its value is typically determined through cross-validation. (Exponential term) The key relationship stability penalty factor: when the local gradient... When smaller, it means and When the relationship is stable, this factor is close to 1 and does not affect the score; when A larger value indicates an abnormal relationship between two key features, and this factor will decrease significantly, thereby reducing the effectiveness of the factor. Scoring helps to capture early or latent defects where individual feature values ​​appear normal but the relationships between features are unbalanced.

[0125] Here, it is believed that a graphite region with excellent structural continuity should satisfy: spatial fluctuations are highly synchronized ( High), the scattering dynamics process is stable and ordered ( (High), and this synchronicity maintains a stable synergistic relationship with multi-scale correlation. Small). At the same time, its multi-scale scattering behavior should also maintain self-similarity ( (However, it's not low). The formula cleverly enforces this comprehensive condition through a combination of multiplication and division structures and an exponential penalty term: only regions that perform well in both numerators and denominators, have no abnormalities in the denominator, and exhibit stable relationships between features can obtain the highest comprehensive structure continuity parameter value. In defective regions, the decay of the numerator, the abnormal increase of the denominator, or the triggering of the gradient penalty term will all lead to a significant decrease in the comprehensive structure continuity parameter value (score).

[0126] Specifically, step S4, which adjusts the separation constraint criterion of ICA based on the comprehensive structural continuity parameter to obtain the flaw detection result, includes:

[0127] Construct a positive correlation transformation relationship between the comprehensive structural continuity parameter and the ICA correlation coefficient tolerance threshold, and use the positive correlation transformation relationship to obtain the target correlation coefficient tolerance threshold of ICA at the target location;

[0128] The flaw detection results of the isostatically pressed graphite product are obtained by outputting independent components based on the tolerance threshold of the correlation coefficient at each location and the ICA output.

[0129] In this embodiment, based on the above embodiments, during the iterative solution of the separation matrix using the ICA algorithm, the results are calculated based on each local region (i.e., each location). The value dynamically relaxes or tightens the constraint on the correlation of the source signal. Value converted to a value between arrive Correlation coefficient tolerance threshold Its specific mapping relationship (positive correlation transformation relationship) can be defined as: ,in This is the maximum tolerance threshold (e.g., set to 1). The slope coefficient is set according to actual needs (e.g., 0.6). yes The threshold (which can be the median or mean).

[0130] This yields the tolerance thresholds for the correlation coefficients adjusted at each position, satisfying the following conditions: Regions with high values ​​(structural continuity) are assigned higher values. The value allows for a certain correlation between the separated signal components, avoiding the forced decomposition of signals that are essentially from the same source but modulated by a continuous structure, thereby suppressing "ghosting"; for Regions with low values ​​(potentially defects) are assigned extremely low values. The value almost mandates signal independence to ensure that the abnormal scattering signals generated by defects can be clearly separated and not confused with background noise.

[0131] Furthermore, this adaptively optimized ICA algorithm is used to process the full ultrasonic matrix data of the entire graphite specimen. The algorithm outputs a set of separated signal components (independent components) and their corresponding spatial distribution maps. After the ICA outputs the independent components, a screening step can be added (or other existing screening methods can be used): calculating the spatial energy distribution map of each independent component and... Negative correlation (or spatial overlap) of the maps. Selecting low... Independent components with high energy concentration in the high-value region are treated as defect signal components and removed from the high-value region. Background noise components with high energy concentration in the defect area are identified. Based on the prior knowledge that "defect signals should have spatial sparsity and local high energy characteristics," one or more signal components most likely corresponding to the actual defect are automatically selected from these components. Finally, the spatial distribution maps of these defect components are fused and binarized to directly generate a clear graphite flaw detection result map that indicates the location, approximate shape, and relative severity of the internal defects, providing an intuitive and reliable basis for subsequent quality assessment and decision-making.

[0132] This invention achieves adaptive optimization of Independent Component Analysis (ICA) based on the characterization of the internal structural continuity of graphite materials by constructing a deep analysis system encompassing local wave synchronicity, multi-scale scattering correlation, and scattering dynamics state transition parameters. Compared with existing technologies, this invention can dynamically adjust the separation constraint criteria of ICA according to the local structural continuity state of graphite reflected by ultrasonic signals (quantified by comprehensive structural continuity parameters): in continuous and uniform regions with high comprehensive structural continuity parameter values, the independence constraint is relaxed to suppress "ghosting" noise caused by over-separation, preserving the natural correlation of background scattering; in suspected defect regions with low comprehensive structural continuity parameter values, the independence constraint is strengthened to force abnormal scattering signals to be clearly separated from the strongly correlated background. This intelligent modulation, bound to the physical properties and state of the material, significantly improves the signal-to-noise ratio and purity of the separated signals, enhances the detection sensitivity for small and latent defects, and ensures the fidelity of defect morphology and location information, effectively solving the problem of component confusion and misjudgment caused by the non-independent source signals in graphite-like non-homogeneous materials under traditional fixed-criteria ICA. This method achieves a substantial improvement in the accuracy and reliability of defect detection by establishing an intrinsic relationship between the continuity of material structure and signal separation constraints.

[0133] Example 2:

[0134] This invention also proposes an ultrasonic flaw detection device for isostatically pressed graphite products. The ultrasonic flaw detection device for isostatically pressed graphite products can be an ultrasonic flaw detector, a computer, or other data processing equipment, or a combination of these devices. Figure 6 As shown, Figure 6 This is a schematic diagram of the hardware operating environment of the ultrasonic flaw detection equipment for isostatically pressed graphite products involved in the embodiments of the present invention.

[0135] like Figure 6 As shown, the ultrasonic flaw detection equipment for isostatically pressed graphite products may include: a processor 1001, such as a CPU; a network interface 1004; a user interface 1003; a memory 1005; and a communication bus 1002. The communication bus 1002 is used to enable communication between these components. The user interface 1003 may include a display or an input unit such as a control panel; optionally, the user interface 1003 may also include a standard wired interface or a wireless interface. The network interface 1004 may optionally include a standard wired interface or a wireless interface (such as a WIFI interface). The memory 1005 may be a high-speed RAM memory or a stable, non-volatile memory, such as a disk storage device. Optionally, the memory 1005 may also be a storage system independent of the aforementioned processor 1001. The memory 1005, as a computer storage medium, may include an ultrasonic flaw detection program for isostatically pressed graphite products (hereinafter referred to as the "ultrasonic flaw detection program").

[0136] Those skilled in the art will understand that Figure 6 The hardware structure shown does not constitute a limitation on the device and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0137] Continue to refer to Figure 6 , Figure 6 The memory 1005, which is a computer-readable storage medium, may include an operating system, a user interface module, a network communication module, and an ultrasonic flaw detection program for isostatically pressed graphite products.

[0138] exist Figure 6 In this embodiment, the network communication module is mainly used to connect to the server and can communicate with the server for data; while the processor 1001 can call the ultrasonic flaw detection program of the isostatic graphite finished product stored in the memory 1005 and execute the steps in the above embodiments.

[0139] Based on the hardware structure of the ultrasonic flaw detection equipment for isostatically pressed graphite products described above, various embodiments of the ultrasonic flaw detection method for isostatically pressed graphite products of the present invention are implemented.

[0140] In addition, the present invention also provides an ultrasonic flaw detection system for isostatically pressed graphite products (hereinafter referred to as "ultrasonic flaw detection system"), please refer to Figure 7 The ultrasonic flaw detection system for the isostatically pressed graphite finished product includes:

[0141] The signal analysis module A10 is used to determine the signal fluctuation characteristics between the target location and its spatiotemporal neighborhood locations based on the ultrasonic time-domain signals of the isostatically pressed graphite product at the target location and its spatiotemporal neighborhood locations, thereby obtaining the local fluctuation synchronization characteristics of the target location; it uses the mutual information between the adjacent scale component signal characteristics of the ultrasonic time-domain signal at the target location to determine the multi-scale scattering correlation characteristics of the target location; it determines the scattering events based on the local signal-to-noise ratio of the ultrasonic time-domain signal at the target location, and determines the state transition probability matrix of the event state evolution process based on the scattering event characteristics, thereby obtaining the sequence signal statistical evolution characteristics of the target location related to the ultrasonic scattering process;

[0142] The component analysis module A20 is used to combine local wave synchronicity characteristics, multi-scale scattering correlation characteristics, and sequence signal statistical evolution characteristics to determine the comprehensive structural continuity parameters. Based on the comprehensive structural continuity parameters, the separation constraint criteria of ICA are adjusted to obtain the flaw detection results.

[0143] Furthermore, the signal analysis module A10 is also used for:

[0144] The signal envelope of the ultrasonic time-domain signal of the isostatically pressed graphite product at the target location is determined, and the envelope curve of the signal energy changing with time is obtained using the signal envelope;

[0145] Perform a first-order difference operation on the envelope curve to obtain a difference sequence characterizing the signal fluctuation mode, and determine the fluctuation mode feature vector composed of the difference sequence.

[0146] By utilizing the wave pattern feature vectors of the target location and its spatiotemporal neighborhood location, the differences in signal wave characteristics between the target location and its spatiotemporal neighborhood location are obtained;

[0147] By utilizing the differences in signal fluctuation characteristics and the set feature parameters corresponding to the spatiotemporal neighborhood location set, the local fluctuation synchronization characteristics of the target location are obtained.

[0148] Furthermore, the signal analysis module A10 is also used for:

[0149] The ultrasonic time-domain signal at the target location is decomposed into multiple scale component signals, and the characteristics of the target scale component signals based on the signal energy and oscillation frequency characteristics are determined.

[0150] The mutual information between the target scale component signal and the scale component signal of its neighboring scale components is determined, and the multi-scale scattering correlation characteristics of the target location are determined by using the mutual information.

[0151] Furthermore, the signal analysis module A10 is also used for:

[0152] Determine the energy vector formed by the signal energy corresponding to each of the scale component signals, and determine the relative dispersion of the energy vector;

[0153] By combining the mutual information and the relative degree of dispersion, the multi-scale scattering correlation characteristics of the target location are determined.

[0154] Furthermore, the signal analysis module A10 is also used for:

[0155] Determine the local signal-to-noise ratio between the local peak value and local noise of the ultrasonic time-domain signal at the target location within a preset time window;

[0156] If the local signal-to-noise ratio is greater than a preset dynamic threshold related to local noise, the corresponding local peak is marked as a scattering event.

[0157] Furthermore, the signal analysis module A10 is also used for:

[0158] The scattering events are determined based on the scattering event characteristics consisting of their corresponding arrival time and peak intensity, and the scattering event sequence consisting of each scattering event characteristic at the target location is determined.

[0159] Based on the scattering event sequences at each location, clustering is used to determine the event state sequence of the event state evolution process at the target location;

[0160] Based on the event state sequence, the corresponding state transition probability matrix and steady-state distribution parameters are determined. Using the state transition probability matrix and steady-state distribution parameters, the statistical evolution characteristics of the sequence signal related to the ultrasonic scattering process at the target location are obtained.

[0161] Furthermore, the component analysis module A20 is also used for:

[0162] Determine the sensitivity of local wave synchronization characteristics to the correlation changes of multi-scale scattering correlation characteristics, and use the sensitivity of correlation changes to obtain the stability penalty factor of the relationship between local wave synchronization characteristics and multi-scale scattering correlation characteristics;

[0163] By combining the characteristics of local fluctuation synchronization, multi-scale scattering correlation, statistical evolution of sequence signals, and relationship stability penalty factor, a comprehensive structural continuity score for the target location is determined.

[0164] Furthermore, the component analysis module A20 is also used for:

[0165] Construct a positive correlation transformation relationship between the comprehensive structural continuity parameter and the ICA correlation coefficient tolerance threshold, and use the positive correlation transformation relationship to obtain the target correlation coefficient tolerance threshold of ICA at the target location;

[0166] The flaw detection results of the isostatically pressed graphite product are obtained by outputting independent components based on the tolerance threshold of the correlation coefficient at each location and the ICA output.

[0167] The specific implementation of the ultrasonic flaw detection system for isostatically pressed graphite products of the present invention is basically the same as the embodiments of the ultrasonic flaw detection method for isostatically pressed graphite products described above, and will not be repeated here.

[0168] Furthermore, the present invention also provides a computer-readable storage medium. The computer-readable storage medium stores an ultrasonic testing program for isostatically pressed graphite products, wherein when the ultrasonic testing program for isostatically pressed graphite products is executed by a processor, the steps of the ultrasonic testing method for isostatically pressed graphite products as described above are implemented.

[0169] The method implemented when the ultrasonic flaw detection procedure for isostatically pressed graphite products is executed can be referred to in various embodiments of the ultrasonic flaw detection method for isostatically pressed graphite products of the present invention, and will not be repeated here.

[0170] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0171] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0172] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0173] The above description is only a preferred embodiment of the present invention and does not limit the scope of protection of the present invention. All equivalent structural / method transformations made under the inventive concept of the present invention using the contents of the present invention specification and drawings, or direct / indirect applications in other related technical fields, are included within the scope of protection of the present invention.

Claims

1. An ultrasonic flaw detection method for isostatically pressed graphite products, characterized in that, The method includes: Based on the ultrasonic time-domain signals of isostatically pressed graphite products at the target location and its spatiotemporal neighborhood, the signal fluctuation characteristics differences between the two locations are determined to obtain the local fluctuation synchronization characteristics of the target location. By utilizing the mutual information between the adjacent scale component signals of the ultrasonic time-domain signal at the target location, the multi-scale scattering correlation characteristics at the target location are determined; The scattering event is determined based on the local signal-to-noise ratio of the ultrasonic time-domain signal at the target location. The state transition probability matrix of the event state evolution process is determined based on the characteristics of the scattering event, and the statistical evolution characteristics of the sequence signal related to the ultrasonic scattering process at the target location are obtained. By combining the characteristics of local wave synchronicity, multi-scale scattering correlation, and statistical evolution of sequence signals, the comprehensive structural continuity parameter is determined. Based on the comprehensive structural continuity parameter, the separation constraint criterion of ICA is adjusted to obtain the flaw detection results. Methods for determining the synchronicity characteristics of local fluctuations include: The signal envelope of the ultrasonic time-domain signal of the isostatically pressed graphite product at the target location is determined, and the envelope curve of the signal energy changing with time is obtained using the signal envelope; Perform a first-order difference operation on the envelope curve to obtain a difference sequence characterizing the signal fluctuation mode, and determine the fluctuation mode feature vector composed of the difference sequence. By utilizing the wave pattern feature vectors of the target location and its spatiotemporal neighborhood location, the differences in signal wave characteristics between the target location and its spatiotemporal neighborhood location are obtained; By utilizing the differences in signal fluctuation characteristics and the set feature parameters corresponding to the spatiotemporal neighborhood location sets, the local fluctuation synchronization characteristics of the target location are obtained. The corresponding calculation formula is: ;in, Represents the four-dimensional spatiotemporal neighborhood set of the target location P; The first-order difference sequence representing the signal envelope at position P is used to obtain the wave pattern feature vector at that target location. This represents the feature vector of the fluctuation pattern at other locations Q within the neighborhood; Represents Euclidean distance; For kernel bandwidth parameters, represents a very small positive number, and L is the length of the difference sequence; Methods for determining multi-scale scattering correlation features of target locations include: The ultrasonic time-domain signal at the target location is decomposed into multiple scale component signals, and the characteristics of the target scale component signals based on the signal energy and oscillation frequency characteristics are determined. Determine the mutual information between the target scale component signal and the scale component signal of its neighboring scale component signals, and use the mutual information to determine the multi-scale scattering correlation characteristics of the target location. The multi-scale scattering correlation features used to determine the target location using mutual information include: Determine the energy vector formed by the signal energy corresponding to each of the scale component signals, and determine the relative dispersion of the energy vector; By combining the mutual information and the relative degree of dispersion, the multi-scale scattering correlation characteristics of the target location are determined. The corresponding calculation formula is: ;in, These are the weighting coefficients assigned to different scale pairs; It is a vector composed of energies at various scales, i.e., an energy vector. It is its variance. It is the maximum value among all energy variance values ​​calculated from the full-field scan data; As a penalty item, Represents a very small positive number. For the first Scale to the Mutual information at scales; Indicates the first eigenvectors of each scale component Indicates the first eigenvectors of each scale component Indicates the target location The number of adaptive scales; Methods for determining the statistical evolution characteristics of sequence signals include: The scattering events are determined based on the scattering event characteristics consisting of their corresponding arrival time and peak intensity, and the scattering event sequence consisting of each scattering event characteristic at the target location is determined. Based on the scattering event sequences at each location, clustering is used to determine the event state sequence of the event state evolution process at the target location; Based on the event state sequence, the corresponding state transition probability matrix and steady-state distribution parameters are determined. Using the state transition probability matrix and steady-state distribution parameters, the statistical evolution characteristics of the sequence signal related to the ultrasonic scattering process at the target location are obtained. The corresponding calculation formula is: ;in, Representing state The steady-state probability; Representing state The transfer entropy, of which This represents the total number of states; The weighted average transition entropy of the entire state sequence; This represents the maximum possible entropy value. The exponential term represents the proportion of events in the scattering event sequence at this target location whose time interval exceeds twice the average interval, out of the total number of events. As a punitive factor; Represents the transition probability matrix Elements in; This represents the scale parameter, which is typically the average time interval of all events at that location. Methods for determining the continuity parameters of a composite structure include: Determine the sensitivity of local wave synchronization characteristics to the correlation changes of multi-scale scattering correlation characteristics, and use the sensitivity of correlation changes to obtain the stability penalty factor of the relationship between local wave synchronization characteristics and multi-scale scattering correlation characteristics; By combining the characteristics of local wave synchronicity, multi-scale scattering correlation, statistical evolution of sequence signals, and relationship stability penalty factor, the comprehensive structural continuity score of the target location is determined. Structural continuity parameters The calculation formula is: ;in, and These are the standardized local fluctuation synchronization characteristics and the statistical evolution characteristics of the sequence signal, respectively. It is a standardized multi-scale scattering correlation feature. Represents a very small positive number. It is a balancing factor. Indicates the sensitivity to changes in the association.

2. The ultrasonic flaw detection method for isostatically pressed graphite products according to claim 1, characterized in that, The determination of scattering events based on the local signal-to-noise ratio of the ultrasonic time-domain signal at the target location includes: Determine the local signal-to-noise ratio between the local peak value and local noise of the ultrasonic time-domain signal at the target location within a preset time window; If the local signal-to-noise ratio is greater than a preset dynamic threshold related to local noise, the corresponding local peak is marked as a scattering event.

3. The ultrasonic flaw detection method for isostatically pressed graphite products according to claim 1, characterized in that, The flaw detection results obtained by adjusting the separation constraint criterion based on the comprehensive structural continuity parameter to adjust the ICA include: Construct a positive correlation transformation relationship between the comprehensive structural continuity parameter and the ICA correlation coefficient tolerance threshold, and use the positive correlation transformation relationship to obtain the target correlation coefficient tolerance threshold of ICA at the target location; The flaw detection results of the isostatically pressed graphite product are obtained by outputting independent components based on the tolerance threshold of the correlation coefficient at each location and the ICA output.

4. An ultrasonic flaw detection system for isostatically pressed graphite products, characterized in that, The system is used to implement the ultrasonic flaw detection method for isostatically pressed graphite products as described in any one of claims 1 to 3; the system includes: The signal analysis module is used to determine the difference in signal fluctuation characteristics between the target location and its spatiotemporal neighboring locations based on the ultrasonic time-domain signals of the isostatically pressed graphite finished product at the target location and its spatiotemporal neighboring locations, thereby obtaining the local fluctuation synchronization characteristics of the target location; and to determine the multi-scale scattering correlation characteristics of the target location by utilizing the mutual information between the adjacent scale component signal characteristics of the ultrasonic time-domain signal at the target location. The scattering event is determined based on the local signal-to-noise ratio of the ultrasonic time-domain signal at the target location. The state transition probability matrix of the event state evolution process is determined based on the characteristics of the scattering event, and the statistical evolution characteristics of the sequence signal related to the ultrasonic scattering process at the target location are obtained. The component analysis module is used to combine local wave synchronicity characteristics, multi-scale scattering correlation characteristics, and sequence signal statistical evolution characteristics to determine the comprehensive structural continuity parameters. Based on the comprehensive structural continuity parameters, the separation constraint criteria of ICA are adjusted to obtain the flaw detection results.

5. An ultrasonic flaw detection device for isostatically pressed graphite products, characterized in that, The device includes a processor, a memory, and an ultrasonic testing program for isostatically pressed graphite products stored in the memory and executable by the processor, wherein when the ultrasonic testing program for isostatically pressed graphite products is executed by the processor, the steps of the ultrasonic testing method for isostatically pressed graphite products as described in any one of claims 1 to 3 are implemented.

Images