Graphite piece performance detection method and system

Abstract

The present application relates to the field of graphite piece detection, more particularly, the present application relates to a kind of graphite piece performance detection method and system, comprising: the surface of the detection area of graphite piece is applied with transient heat flow loading to induce thermal stress;Dynamic acoustic emission signal generated in graphite piece is collected during heat flow loading, and its acoustic emission feature vector representing defect dynamic expansion activity is extracted.The present application utilizes transient heat flow to promote closed defects to open up, and overcomes the limitation of normal temperature nondestructive testing.The method cooperates to collect acoustic emission and ultrasonic dispersion characteristics during loading period, combined with adaptive de-noising and mahalanobis distance fusion algorithm, to eliminate background interference and dimension redundancy to realize multidimensional representation;Finally, based on fatigue life test calibration grading threshold, the scientific mapping of detection index and service reliability is established, and the accuracy of graphite piece performance grade evaluation is improved.

Description

Technical Field

[0001] This invention relates to the field of graphite component testing technology. More specifically, this invention relates to a method and system for testing the performance of graphite components. Background Technology

[0002] Graphite components are widely used in extreme working conditions such as semiconductor crystal growth thermal fields and high-temperature metallurgy due to their excellent electrical and thermal conductivity and high-temperature resistance. Because their service environment involves high temperatures and severe thermal shock, internal microscopic integrity testing before leaving the factory is crucial to prevent premature fracture failure.

[0003] Existing technologies typically perform ultrasonic non-destructive testing under normal temperature and stress-free conditions. Specifically, an acoustic coupling agent is applied to the surface of a graphite part, and a single-frequency ultrasonic wave is emitted using an ultrasonic probe. The transmitted sound velocity and energy attenuation coefficient are calculated and compared with the standard acceptable threshold at room temperature to determine whether there are internal pores or cracks.

[0004] However, the above methods have significant limitations. Graphite components are under high-temperature thermal stress for extended periods during actual service, while at room temperature without stress, the interlayer microcracks and micropores remaining inside the graphite material are often mechanically closed. In this state, the acoustic scattering cross-section of the closed cracks for ultrasonic waves is extremely small, allowing ultrasonic signals to penetrate easily and causing little to no significant reduction in sound velocity or energy attenuation. This prevents the effective excitation and exposure of the dynamic evolution characteristics of micro-defects under thermal expansion. Consequently, even qualified graphite components, once put into high-temperature service, will experience rapid opening and unstable propagation of closed cracks under intense thermal stress, leading to premature failure. Summary of the Invention

[0005] This invention provides a method and system for testing the performance of graphite parts, aiming to overcome the limitation of room temperature testing in identifying closed defects, and to accurately assess the tendency of internal micro-defects in graphite parts under high temperature thermal stress.

[0006] In a first aspect, the present invention provides a method for testing the performance of graphite parts, comprising: S1, applying transient heat flux loading to the surface of the test area of ​​the graphite part to induce thermal stress; S2, acquiring dynamic acoustic emission signals generated inside the graphite part during the heat flux loading, and extracting acoustic emission feature vectors characterizing the dynamic expansion activity of defects; S3, under thermal stress retention state after the heat flux loading stops, emitting broadband ultrasonic signals into the graphite part, acquiring the transmitted broadband ultrasonic signals, and extracting ultrasonic dispersion feature vectors whose phase velocity changes with frequency; S4, combining the acoustic emission feature vectors and the ultrasonic dispersion feature vectors to obtain a joint feature vector to be tested; comparing the joint feature vector to be tested with the benchmark joint feature distribution corresponding to a pre-established standard defect-free sample library to obtain a comprehensive degradation tendency feature value used to characterize the degree of degradation of the graphite part under thermal stress; S5, comparing the comprehensive degradation tendency feature value with a preset grading threshold to determine the comprehensive performance level of the graphite part.

[0007] Furthermore, applying transient heat flux loading includes: calculating the ratio of the square of the target heat conduction distance to the thermal diffusivity of the graphite material to obtain the theoretical heat diffusion time, wherein the target heat conduction distance is half the thickness of the graphite tube wall; defining a time window interval within a preset deviation range based on the theoretical heat diffusion time, and obtaining the preset heating duration within the time window interval.

[0008] Limiting the preset heating duration to a time window based on theoretical values ​​ensures that the thermal stress wave forms the maximum temperature gradient in the wall thickness direction to fully expose hidden defects, while also preventing excessive heating time from causing the graphite part to become uniformly heated (thermal stress disappears) or causing localized high-temperature oxidation on the surface.

[0009] Furthermore, the thermal stress holding state includes: after the transient thermal flow loading stops, waiting for a preset delay trigger time to allow the internal thermal stress to reach a preset holding level.

[0010] By setting a preset delay trigger time, the thermoelastic coupling principle is utilized to ensure that the internal thermal stress remains within a relatively stable load-holding range after loading stops. At this time, the microcracks maintain their maximum opening under continuous thermal stress support, which greatly enhances the acoustic scattering cross-section, thereby improving the signal-to-noise ratio and sensitivity of subsequent broadband ultrasonic dispersion detection.

[0011] Furthermore, the ultrasonic dispersion feature vector of phase velocity as a function of frequency is extracted, including: performing time-frequency joint analysis on the transmitted broadband ultrasonic signal to obtain the phase velocities of a preset high-frequency band and a preset low-frequency band, and using the attenuation rate between the phase velocities of the high-frequency band and the low-frequency band as the ultrasonic dispersion feature vector.

[0012] This dispersion eigenvector sensitively reflects the difference in scattering sensitivity of sound waves of different wavelengths when they encounter open defects, and can objectively characterize the spatial size distribution of internal microcracks and micropores.

[0013] Furthermore, it also includes: acquiring background environmental noise signals under normal temperature and stress-free conditions, and setting adaptive bandpass filtering algorithm parameters according to the frequency domain energy envelope of the background environmental noise signals; and using the adaptive bandpass filtering algorithm parameters to filter the acquired dynamic acoustic emission signals.

[0014] Furthermore, the joint feature vector to be tested is compared with the benchmark joint feature distribution corresponding to the pre-established standard defect-free sample library, including: calculating the Mahalanobis distance between the joint feature vector to be tested and the benchmark mean vector of the standard defect-free sample library, as the comprehensive degradation tendency feature value; wherein, the calculation of the Mahalanobis distance is also based on the inverse matrix of the feature covariance matrix of the standard defect-free sample library.

[0015] Mahalanobis distance is used as the comprehensive degradation tendency characteristic value. The inverse matrix of the covariance matrix is ​​introduced, which automatically eliminates the possible collinear interference (correlation) between acoustic emission characteristics and ultrasonic dispersion characteristics, and eliminates the influence of different physical dimensions, making the final single degradation index more objective and statistically reasonable.

[0016] Furthermore, a standard defect-free sample library is established, including: obtaining multiple benchmark graphite parts as test samples, repeating steps S1 to S3 for each test sample to extract the benchmark joint feature vector set; calculating the mean vector of the benchmark joint feature vector set as the benchmark mean vector, and calculating its covariance matrix as the feature covariance matrix.

[0017] By conducting repeated tests on multiple defect-free benchmark samples to construct a standard library, the statistical mean and covariance distribution of the characteristics of defect-free graphite parts were obtained. This allows the system to automatically account for inherent batch differences caused by the natural non-uniformity of graphite materials during subsequent testing, significantly reducing the false alarm rate in the quality judgment process.

[0018] Furthermore, the preset grading threshold is determined by: conducting high-temperature accelerated fatigue life tests on historical batches of graphite parts with different comprehensive deterioration tendency characteristic values, mapping the failure cycle number of graphite parts to the comprehensive deterioration tendency characteristic value, and calibrating the grading threshold.

[0019] In the second aspect, a graphite component performance testing system is also provided, including a processor and a memory, the memory storing a computer program, and the processor executing the computer program to implement any of the above-mentioned graphite component performance testing methods.

[0020] Beneficial effects: By applying transient heat flux loading, microscopic defects that are closed at room temperature inside graphite parts are actively opened and slid, effectively solving the problem that traditional room temperature stress-free testing is difficult to excite and identify closed defects. On this basis, the dynamic acoustic emission characteristics during the heat flux loading period and the ultrasonic dispersion characteristics during the thermal stress holding period are collected in synergy. Combined with adaptive environmental noise reduction and a feature-level scientific fusion algorithm based on Mahalanobis distance, the redundant influence of industrial background interference and heterogeneous feature dimensions is reduced, realizing a multi-dimensional fusion characterization of the dynamic evolution and static morphology of defects. Finally, based on the accelerated fatigue life test calibration and grading threshold, a scientific mapping between non-destructive testing indicators and real service reliability is established, improving the accuracy of quality grading testing of graphite parts under extreme conditions. Attached Figure Description

[0021] Figure 1 This is a schematic flowchart illustrating a graphite component performance testing method according to an embodiment of the present invention. Detailed Implementation

[0022] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0023] like Figure 1 As shown, S101: Spatial layout construction of the multimodal sensor network of the detection system and acquisition and preprocessing of baseline state data.

[0024] First, a multimodal sensing network is constructed for the graphite part under test to obtain its three-dimensional shape and size information. Based on this three-dimensional shape and size information, a detection area is planned on the surface of the graphite part under test. At least one non-contact transient thermal excitation source is arranged on one side of the detection area, and several acoustic emission sensor nodes are arranged around the detection area. Broadband ultrasonic transmitting probes and broadband ultrasonic receiving probes are arranged on opposite sides of the detection area, respectively. All the above hardware nodes are connected to the central controller through a synchronous trigger bus to ensure the timing consistency of different modal signal acquisition.

[0025] Subsequently, under normal temperature and stress-free conditions without the transient thermal excitation source activated, the acoustic emission sensor node and broadband ultrasonic receiver probe are started to collect background environmental noise signals for a preset duration. The collected background environmental noise signals are then transmitted to the signal processing module. The signal processing module performs a fast Fourier transform on the background environmental noise signals to extract the frequency domain energy envelope of the noise signals. Based on the maximum amplitude characteristics of the frequency domain energy envelope, a dynamic adaptive filtering threshold is set, and adaptive bandpass filtering algorithm parameters for environmental noise are generated.

[0026] It should be noted that the basis for setting the dynamic adaptive filtering threshold is that there are a large number of mechanical vibrations and electromagnetic interferences in industrial environments. The amplitudes of these interference signals exhibit non-stationary random fluctuations. By extracting the maximum amplitude characteristics of the background environmental noise under stress-free conditions and adding a preset redundant bias factor, a reference noise level can be established. This reference level is used to eliminate spurious signals unrelated to the microscopic defects of the graphite parts in subsequent steps. Furthermore, the propagation delay between the broadband ultrasonic transmitting and receiving probes is calibrated using a preset reference pulse signal to obtain the system hardware delay time constant, which is then used as the calculation benchmark for the actual flight time of the ultrasonic wave.

[0027] When arranging sensor nodes, increasing the number and distribution density of acoustic emission sensor nodes can improve the stability of acoustic emission signal spatial positioning and the comprehensiveness of feature extraction, while reducing the number of nodes can reduce system hardware costs and improve signal processing response speed; adjusting the effective distance of the non-contact transient thermal excitation source can change the initial heat flux density reaching the graphite surface.

[0028] Example Description: In an actual testing workshop, the operator first places the cylindrical graphite part to be tested on the testing platform. After the system reads the outer diameter and height information of the graphite part, it plans the middle part of the side wall as the testing area. Then, the system controls the robotic arm to position a non-contact heat source directly in front of the outer side wall of the graphite part. Multiple acoustic emission and reception probes are attached to the graphite surface around the heat source, and ultrasonic emission and reception probes are attached to the inner and outer walls of the side wall, respectively. All probes are connected to the same multi-channel acquisition controller. At room temperature without heating, the system runs silently for a preset period of time, recording the vibration noise and electromagnetic noise caused by the operation of the machines in the workshop. The system software performs spectrum analysis on this noise and finds that the background noise is mainly concentrated in a specific low-frequency band, and the maximum amplitude is a specific voltage value. The system generates a bandpass filter based on this and sets the threshold to be higher than the maximum amplitude voltage to ensure that any signal acquired afterward that exceeds the threshold is considered to be a genuine response from inside the graphite component, rather than external interference. Finally, the system sends an extremely short electrical pulse to record the internal circuit delay from transmission to reception, which serves as the subtraction benchmark for subsequent sound speed calculations.

[0029] S102: Extraction of dynamic acoustic emission response features of transient thermal stress loading and micro-defects based on preset control parameters.

[0030] After preprocessing, the central controller sends control commands to the non-contact transient thermal excitation source, driving it to apply transient thermal flux loading to the surface of the detection area of ​​the graphite part under test. During the transient thermal flux loading, the surface material of the graphite part undergoes rapid thermal expansion after absorbing thermal energy, while the internal material remains at a relatively low temperature due to the lag in thermal conduction. This induces a significant transient thermal stress gradient in the thickness direction of the internal material of the graphite part. Driven by this transient thermal stress gradient, the microcracks and micropores inside the graphite material, which are in a closed state at room temperature, overcome the interfacial friction and generate microscale slip, friction, and opening and expansion behaviors. This dynamic change in the microstructure is accompanied by the rapid release of local elastic strain energy, thereby radiating high-frequency transient elastic waves, i.e., dynamic acoustic emission signals.

[0031] During this process, the acoustic emission sensor node acquires the aforementioned dynamic acoustic emission signal in real time. The signal processing module uses the adaptive bandpass filter algorithm parameters generated in step S101 to filter and denoise the dynamic acoustic emission signal, removing background environmental noise. Subsequently, an acoustic emission feature vector characterizing the dynamic expansion activity of the defect is extracted from the denoised dynamic acoustic emission signal. This acoustic emission feature vector contains at least one feature parameter that can reflect the energy release intensity, such as the cumulative ring count or the total released energy value.

[0032] It is important to note that a key parameter for controlling the transient thermal excitation source is the preset heating duration. This preset heating duration is calculated in advance by the system based on the rate of heat conduction by the graphite material, ensuring that heating stops just as the heat reaches the center of the crucible wall, thus creating the maximum stress difference. The calculation process for the preset heating duration is as follows: First, half the thickness of the graphite tube wall is taken as the target heat conduction distance. Next, the square of the target heat conduction distance is divided by the thermal diffusivity of the graphite material to calculate the theoretical thermal diffusion time required for heat to reach the center of the tube wall. Finally, using the theoretical thermal diffusion time as a baseline value, a preset deviation percentage (e.g., ±20%) is added or subtracted from this baseline value to define a time window interval. Within this time window interval, the preset heating duration of the external heat source is obtained to create the maximum temperature gradient and thermal stress difference along the tube wall thickness direction.

[0033] Exemplary Explanation: After preparation is complete in step S101, the control system commands the external heat source to rapidly heat the outer wall of the graphite part with high intensity within a short heating time. Due to the extremely short heating time and high intensity, the outer wall of the crucible expands rapidly, but the inside of the crucible has not yet had time to be heated. This creates a force of compression and pulling from the outside in, i.e., thermal stress. The tiny cracks that were originally tightly closed inside the crucible material are forcibly pulled open or rub against each other by this force. In this process, extremely weak high-frequency sound waves are emitted, i.e., acoustic emission signals. The acoustic emission probe attached to the surface keenly captures these sound waves. The acquisition system uses the filter set in the first step to filter out the roaring noise of the workshop machines, retaining only these crisp high-frequency sound waves representing crack activity, and calculates how many such peaks were generated during the entire heating period, as well as the total energy they contain. This data is combined into a data set representing the crack activity level.

[0034] S103: Broadband ultrasonic guided wave excitation and spatial radio frequency dispersion feature extraction under thermal stress load conditions.

[0035] After the transient thermal excitation stops, the temperature distribution of the material inside the graphite part is in a relaxation process of rebalancing. During this relaxation stage, the thermal stress has not completely dissipated, and the internal microcracks are supported by the thermal stress to maintain the geometry of opening or expanding. This stage is defined as the thermal stress holding state. After the transient thermal excitation stops, the central controller sends an excitation command to the broadband ultrasonic transmitting probe after a preset delay trigger time. The broadband ultrasonic transmitting probe emits a broadband ultrasonic signal containing continuous spectrum components into the graphite part.

[0036] Continuing from the above, the calibration method for the preset delayed trigger time is as follows: based on the thermal diffusion penetration time calculated above, combined with the thermoelastic coupling coefficient of the graphite material, calculate the time required for the internal thermal stress to decay from the peak value to the preset maintenance level. This delayed trigger time serves as the benchmark trigger condition to ensure that the ultrasonic detection process occurs precisely in the state range where the internal crack opening displacement is at its maximum opening degree.

[0037] When broadband ultrasonic signals penetrate the interior of a graphite component under thermal stress, the opening microcracks and expanding micropores constitute a strong acoustic scatterer. According to the laws of acoustic scattering physics, ultrasonic waves of different frequencies exhibit significantly different scattering sensitivities to defects of different sizes. High-frequency ultrasonic waves, with their shorter wavelengths, produce strong Rayleigh scattering and mode conversion when encountering the opening microcrack interface, resulting in significant dissipation of wavefront energy and a marked decrease in phase velocity. In contrast, low-frequency ultrasonic waves, with their longer wavelengths, can produce diffraction effects and are less affected by the opening of microcracks, maintaining a relatively stable phase velocity. Therefore, when broadband ultrasonic signals penetrate graphite materials with opening defects, they exhibit a strong dispersion phenomenon where the phase velocity varies with frequency.

[0038] A broadband ultrasonic receiving probe receives transmitted broadband ultrasonic signals. The signal processing module performs time-frequency joint analysis on the received broadband ultrasonic signals, extracting the low-frequency phase velocity of a preset low-frequency band and the high-frequency phase velocity of a preset high-frequency band from the transmitted signal. Subsequently, the attenuation rate between the high-frequency and low-frequency phase velocities is calculated and used as the ultrasonic dispersion feature vector characterizing the static spatial size distribution of the defect. Specifically, the time-frequency joint analysis can employ either a short-time Fourier transform (STFT) algorithm or a continuous wavelet transform (CWT) algorithm. When extracting dispersion features, the center frequency of the preset low-frequency band can be defined as 0.5 MHz, and the center frequency of the preset high-frequency band can be defined as 3.5 MHz.

[0039] Example Explanation: The instant the external heat source stops heating, the temperature inside the graphite is not yet balanced, and the tiny cracks that were just stretched open remain open. The system waits precisely for a short time to ensure the cracks are at their maximum opening. At this moment, the control system commands the ultrasonic transmitting probe on the inner wall to emit a mixed ultrasonic beam containing both low and high frequencies. This ultrasonic beam encounters the opened cracks as it passes through the graphite tube wall. The low-frequency ultrasonic waves, like long waves, easily bypass the cracks, and their speed is not significantly affected. However, the high-frequency ultrasonic waves, with their short wavelengths, experience severe reflection and scattering upon impact with the open cracks, causing the high-frequency waves to penetrate at a slower speed. The receiving probe on the outer wall records the penetrating mixed ultrasonic waves. The system software breaks them down, measuring the speeds of the low-frequency and high-frequency ultrasonic waves separately, and calculating the percentage by which the high-frequency waves are slower than the low-frequency waves. This percentage decrease in speed reflects the spatial distribution of the cracks stretched open by thermal stress inside the graphite.

[0040] Step S104: Evaluate the fusion of acoustic emission and ultrasonic multi-source heterogeneous features and the overall degradation tendency based on scientific statistical algorithms.

[0041] After obtaining the acoustic emission feature vector reflecting the dynamic expansion activity of the defect and the ultrasonic dispersion feature vector reflecting the static spatial size distribution of the defect, this step abandons the practice of simply adding these two types of data with different physical dimensions and heterogeneous physical sources, and introduces the Mahalanobis Distance algorithm from multivariate statistics for feature-level fusion.

[0042] First, a standard defect-free sample library is pre-established. The process involves selecting several benchmark graphite parts, verified by destructive physical testing to be free of macroscopic and microscopic defects, as test samples. Under the same environmental and parameter conditions as steps S101 to S103, multiple tests are performed. Acoustic emission feature vectors and ultrasonic dispersion feature vectors are extracted from each test set, and these feature vectors are concatenated into a benchmark joint feature vector set. Subsequently, the benchmark mean vector of this benchmark joint feature vector set and the feature covariance matrix, which reflects the intrinsic correlation between different modal characteristics, are calculated. This feature covariance matrix reflects the statistical distribution law of acoustic emission and ultrasonic signal characteristic fluctuations caused by fundamental physical phenomena such as microscopic lattice thermal vibration within qualified materials.

[0043] When evaluating the actual graphite part to be tested, the acoustic emission feature vector extracted in step S102 is concatenated with the ultrasonic dispersion feature vector extracted in step S103 to form a joint feature vector. Subsequently, the Mahalanobis distance between the joint feature vector and the baseline defect-free distribution state is calculated using the following scientific model. This distance serves as a comprehensive degradation tendency characteristic value for determining the degree of degradation of the graphite part under thermal stress. The calculation formula is as follows: ;in, This represents the calculated characteristic value of the overall degradation tendency. Represents the joint feature vector to be tested. This represents the baseline mean vector corresponding to the standard defect-free sample library. The inverse matrix of the characteristic covariance matrix is ​​represented by the symbol. This represents the matrix transpose operation.

[0044] This fusion method follows real engineering principles. By introducing a covariance matrix, it effectively eliminates the difference between the dimensions of acoustic emission energy accumulation and ultrasonic velocity attenuation, and automatically eliminates redundant covariance information between multi-source heterogeneous features. The calculated comprehensive degradation tendency characteristic value is essentially a dimensionless statistical distance. The larger the value, the more serious the high-temperature degradation tendency is, indicating that the dynamic crack activity and static defect scattering size of the graphite part under thermal stress excitation deviate significantly from the physical fluctuation range of normal qualified products.

[0045] Example Explanation: After obtaining acoustic emission data representing crack activity and ultrasonic data representing crack size, the challenge lies in combining these two data sets with completely different units; simply adding them together is as meaningless as adding temperature and length. Therefore, the system utilizes a built-in standard database, repeatedly testing and statistically analyzing the normal fluctuation range and correlation matrix. The system then uses the Mahalanobis distance formula, substituting the combined data of the graphite part under test into the calculation. During this calculation, the covariance matrix in the formula automatically smooths out the unit differences between crack activity and crack size, and measures the relationship between them. Finally, the formula calculates a single value, which is the Mahalanobis distance. If this value is small, it indicates that the performance of the graphite part under test is very close to that of a perfect graphite part; if this value is large, it conclusively proves that once the graphite part is heated, the internal crack activity is abnormally intense, making it extremely prone to damage.

[0046] Step S105: Determine the overall performance level of the graphite component.

[0047] After obtaining the comprehensive degradation tendency characteristic value, the central controller executes the final performance judgment logic. Several preset graded threshold ranges are pre-programmed in the system memory. The physical source and acquisition method of these graded thresholds are as follows: For graphite parts with different Mahalanobis distance characteristic values ​​in historical batches, accelerated fatigue life tests are carried out in an ultra-high temperature test furnace simulating actual industrial applications. The number of thermal cycle cycles for each sample until the resistivity increases sharply or macroscopic cracking failure occurs is recorded. According to the statistical distribution of failure cycles, the life range is mapped to the Mahalanobis distance characteristic value axis, and at least one critical threshold for distinguishing between qualified and unqualified is calibrated, as well as thresholds for classifying different levels such as excellent, medium, and potential problems. For example, the qualified critical threshold is 2.5 and the unqualified critical threshold is 6.0.

[0048] The comprehensive degradation tendency characteristic value calculated in step S104 (represented by the dimensionless Mahalanobis distance characteristic value in this embodiment) is compared with the qualified critical threshold and the unqualified critical threshold. The specific performance judgment logic is as follows:

[0049] First-level judgment: If the comprehensive deterioration tendency characteristic value is less than or equal to the qualified critical threshold, the graphite part is judged to have a dense internal structure and a very low thermal stress deterioration tendency, and is given the highest level qualified label, such as the superior product label.

[0050] Second interval judgment: If the comprehensive deterioration tendency characteristic value is greater than the qualified critical threshold and less than or equal to the unqualified critical threshold, the graphite part is judged to have a slight or moderate thermal stress deterioration tendency and is assigned a downgraded use label, such as a load-limited / cooling use label.

[0051] Third interval judgment: If the comprehensive deterioration tendency characteristic value is greater than the non-conforming critical threshold, it is determined that the graphite part is prone to unstable propagation of internal microcracks under thermal stress. The central controller immediately triggers an alarm command, determines it as a non-conforming product, and controls the system to generate a comprehensive evaluation report containing detailed time-domain waveforms, feature extraction parameters, and multi-modal fusion distance values.

[0052] The present invention also provides a graphite component performance testing system, the system comprising a processor and a memory, the memory storing computer program instructions, which, when executed by the processor, implement a graphite component performance testing method according to the first aspect of the present invention.

Claims

1. A method for testing the performance of graphite components, characterized in that, include: S1. Apply transient heat flow loading to the surface of the detection area of ​​the graphite part to induce thermal stress; S2. Acquire dynamic acoustic emission signals generated inside the graphite part during hot flow loading, and extract acoustic emission feature vectors that characterize the dynamic expansion activity of defects. S3. Under the thermal stress retention state after the heat flow loading stops, a broadband ultrasonic signal is emitted into the graphite part, the transmitted broadband ultrasonic signal is collected, and the ultrasonic dispersion feature vector of its phase velocity as a function of frequency is extracted. S4. Combine the acoustic emission feature vector with the ultrasonic dispersion feature vector to obtain the joint feature vector to be measured; compare the joint feature vector to be measured with the benchmark joint feature distribution corresponding to the pre-established standard defect-free sample library to obtain the comprehensive degradation tendency feature value used to characterize the degree of degradation of graphite parts under thermal stress. S5. Compare the comprehensive deterioration tendency characteristic value with the preset grading threshold to determine the comprehensive performance level of the graphite part.

2. The method for testing the performance of graphite parts according to claim 1, characterized in that, Applying transient heat flux loading includes: The theoretical thermal diffusion time is obtained by calculating the ratio of the square of the target thermal conduction distance to the thermal diffusivity of the graphite material, where the target thermal conduction distance is half the thickness of the graphite tube wall. A time window is defined within a preset deviation range based on the theoretical thermal diffusion time, and the preset heating duration is obtained within the time window.

3. The method for testing the performance of graphite parts according to claim 1, characterized in that, Thermal stress holding conditions include: After the transient heat flow loading stops, a preset delay trigger time is waited for the internal thermal stress to reach the preset maintenance level.

4. The method for testing the performance of graphite parts according to claim 1, characterized in that, Extract the ultrasonic dispersion feature vector of its phase velocity as a function of frequency, including: A time-frequency joint analysis is performed on the transmitted broadband ultrasonic signal to obtain the phase velocities of a preset high-frequency band and a preset low-frequency band, and the attenuation rate between the phase velocities of the high-frequency band and the low-frequency band is used as the ultrasonic dispersion feature vector.

5. The method for testing the performance of graphite parts according to claim 1, characterized in that, Also includes: Background environmental noise signals were collected under normal temperature and stress-free conditions, and the parameters of the adaptive bandpass filtering algorithm were set according to the frequency domain energy envelope of the background environmental noise signals. The acquired dynamic acoustic emission signals are filtered using an adaptive bandpass filtering algorithm.

6. The method for testing the performance of graphite parts according to claim 1, characterized in that, The joint feature vector to be tested is compared with the benchmark joint feature distribution corresponding to a pre-established standard defect-free sample library, including: The Mahalanobis distance between the joint feature vector to be tested and the baseline mean vector of the standard defect-free sample library is calculated as the comprehensive degradation tendency feature value; the Mahalanobis distance is also based on the inverse matrix of the feature covariance matrix of the standard defect-free sample library.

7. The method for testing the performance of graphite parts according to claim 1, characterized in that, Establish a standard, defect-free sample library, including: Multiple reference graphite parts are obtained as test samples. Steps S1 to S3 are repeated for each test sample to extract the reference joint feature vector set. The mean vector of the reference joint feature vector set is calculated as the reference mean vector, and its covariance matrix is ​​calculated as the feature covariance matrix.

8. The method for testing the performance of graphite parts according to claim 1, characterized in that, Determine the preset grading thresholds, including: High-temperature accelerated fatigue life tests were conducted on historical batches of graphite parts with different comprehensive deterioration tendency characteristic values. The failure cycle number of the graphite parts was mapped to the comprehensive deterioration tendency characteristic value to calibrate the grading threshold.

9. A graphite component performance testing system, comprising a processor and a memory, characterized in that, The memory stores a computer program, and the processor executes the computer program to implement the graphite component performance testing method as described in any one of claims 1-8.

Images