Multi-parameter performance detection system and method for polyphenyl ether foaming material
By using the feature fusion technology of terahertz time-domain spectroscopy and air-coupled ultrasonic sensors, the problem of multimodal data fusion in the nondestructive testing of polyphenylene ether foam materials was solved, achieving high-precision defect identification and decision-making, and improving the stability and reliability of the testing system.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-02-24
- Publication Date
- 2026-03-24
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Existing technologies cannot effectively achieve adaptive fusion of multimodal sensing data and defect probability confidence for polyphenylene ether foam materials, resulting in low accuracy of nondestructive testing, especially prone to false alarms and conflicts when dealing with complex or ambiguous defects.
A terahertz time-domain spectral sensor and an air-coupled ultrasonic sensor were used to scan polyphenylene ether foam material, extract time-domain and frequency-domain feature parameters, perform feature fusion by calculating sensor confidence, and use a support vector machine model to distinguish defect types and locate their positions, thereby generating disposal decision instructions.
It improves the accuracy of online non-destructive testing of polyphenylene ether foam materials, enhances the comprehensive discrimination ability and decision reliability of complex or ambiguous defects, reduces noise interference and information redundancy, and ensures the stability and accuracy of test results.
Smart Images
Figure CN121720969A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of camera measurement technology, and more specifically, to a multi-parameter performance testing system and method for polyphenylene ether foam materials. Background Technology
[0002] Polyphenylene ether (PPE) foam is a type of high-performance engineering foam made from PPE resin through physical or chemical foaming processes. It features a uniform, closed microporous structure and combines low density, high specific strength, excellent dielectric properties, low moisture absorption, and good heat resistance. It is widely used in aerospace, electronics and communications, and transportation fields as lightweight structural components or functional core materials.
[0003] Existing technologies for nondestructive testing of polyphenylene oxide (PPE) foam materials merely present test data from different sources in a simple parallel manner or make rudimentary logical judgments based on fixed rules, failing to establish an evaluation mechanism that can dynamically and objectively quantify the reliability of the data from each sensor. Due to the lack of precise measurement of the specified performance of sensors and their confidence levels in the current testing environment, the value weight of different data sources cannot be distinguished during information synthesis. Consequently, when dealing with defects with complex shapes or ambiguous features, the accurate indications of superior sensors are easily diluted or even masked by interference signals or false alarms from other sensors. Conflicts may arise between information from different sensing modalities, leading to a dilemma in the system's decision-making process. Ultimately, this increases the uncertainty of defect identification conclusions and reduces the reliability of unified judgments. Therefore, how to achieve adaptive fusion of multimodal sensor data and defect probability confidence levels to improve the judgment accuracy of online nondestructive testing of PPE foam materials has become a challenge for the industry. Summary of the Invention
[0004] This application provides a multi-parameter performance testing system and method for polyphenylene ether foam materials, which can achieve adaptive fusion of multimodal sensing data and defect probability confidence, thereby improving the judgment accuracy of online non-destructive testing of polyphenylene ether foam materials.
[0005] In a first aspect, this application provides a method for testing the multi-parameter performance of polyphenylene ether foam materials, comprising the following steps: When the conveyor carries the polyphenylene ether foam material through the detection station, the displacement pulse signal of the conveyor is used to trigger the terahertz time-domain spectroscopy sensor and the air-coupled ultrasonic sensor to scan the polyphenylene ether foam material, thereby obtaining the terahertz time-domain waveform data and ultrasonic transmission signal of the polyphenylene ether foam material. The terahertz time-domain waveform data is demodulated in the time domain to extract a first feature parameter characterizing near-surface defects of the material, and the ultrasonic transmission signal is identified in the frequency domain to extract a second feature parameter characterizing internal defects and thickness changes of the material. The first feature parameter includes the time-domain reflection peak time and frequency band absorption rate, and the second feature parameter includes the sound wave flight time and signal energy attenuation coefficient. By calculating the probability distribution between the identification results of different defect types of polyphenylene ether foam material by each sensor and the standard results, the sensor confidence of the terahertz time-domain spectroscopy sensor and the air-coupled ultrasonic sensor is determined. Then, the first feature parameter and the second feature parameter are fused using the confidence of each sensor to obtain the defect feature vector of polyphenylene ether foam material. The defect feature vector is used to distinguish the defect type and locate the defect location to obtain the defect judgment result of the polyphenylene ether foam material. The defect judgment result is compared with the pre-stored acceptance standard threshold to generate the disposal decision instruction for the current polyphenylene ether foam material.
[0006] In some embodiments, demodulating the terahertz time-domain waveform data in the time domain and extracting a first feature parameter characterizing near-surface defects in the material specifically includes: Time-domain analysis is performed on the terahertz time-domain waveform data to identify and extract the peak time of its main reflection pulse, which is taken as the time-domain reflection peak time. The terahertz time-domain waveform data is subjected to a fast Fourier transform to obtain its frequency domain spectrum, and the ratio of the energy integral in the specified characteristic frequency band to the reference spectrum is calculated as the frequency band absorption rate. The first characteristic parameter characterizing near-surface defects of the material is determined based on the time-domain reflection peak time and the frequency band absorption rate.
[0007] In some embodiments, frequency domain identification of the ultrasonic transmission signal to extract a second feature parameter characterizing internal defects and thickness variations in the material specifically includes: In the ultrasonic transmission signal, the propagation time of the direct wave from transmission to reception is identified as the sound wave flight time; Calculate the attenuation ratio of the energy amplitude of the ultrasonic transmission signal to the corresponding value of the reference standard signal, and use it as the signal energy attenuation coefficient; A second characteristic parameter characterizing internal defects and thickness variations of a material is determined based on the sound wave flight time and the signal energy attenuation coefficient.
[0008] In some embodiments, the sensor confidence of the terahertz time-domain spectroscopy sensor and the air-coupled ultrasonic sensor is determined by calculating the probability distribution between the identification results of each sensor for different defect types of polyphenylene ether foam material and the standard results. Specifically, this includes: Retrieve all near-surface and internal defects of polyphenylene ether foam materials from the historical defect database; The accuracy of the terahertz time-domain spectroscopy sensor in identifying various near-surface defects is statistically analyzed, and the probability distribution of all identification accuracies is constructed to obtain the sensor confidence of the terahertz time-domain spectroscopy sensor. The accuracy rate of air-coupled ultrasonic sensors in identifying various internal defects is statistically analyzed, and the probability distribution of all identification accuracies is constructed to obtain the sensor confidence level of the air-coupled ultrasonic sensors.
[0009] In some embodiments, the defect feature vector of the polyphenylene ether foam material is obtained by feature fusion of the first feature parameter and the second feature parameter using the confidence levels of each sensor, specifically including: The first feature parameter and the second feature parameter are combined to form an initial feature set; The confidence scores of each sensor are used to weight and fuse the feature parameters in the initial feature set to obtain a fused feature vector. Principal component analysis is performed on the fused feature vector, and then the dimension is reduced to form a low-correlation feature vector, thereby obtaining the defect feature vector of polyphenylene ether foam material.
[0010] In some embodiments, using the defect feature vector to distinguish defect types and locate defect positions to obtain defect determination results for polyphenylene ether foam materials specifically includes: The defect feature vector is input into a pre-trained support vector machine multi-classification model. The classification model outputs a predicted label for the defect type, which includes surface cracks, internal pores, inclusions, and no defects, thereby obtaining the defect type. The defect label is mapped onto the three-dimensional digital model of the polyphenylene ether foam material using the displacement pulse signal triggered by the scan, thereby obtaining the defect location; The defect determination result of polyphenylene ether foam material is determined by the defect type and the defect location.
[0011] In some embodiments, comparing the defect determination result with a pre-stored acceptance standard threshold to generate a disposal decision instruction for the current polyphenylene ether foam material specifically includes: The acceptance standard threshold for the defect type in the defect judgment result is preset based on the design standard; The defect size in the defect determination result is compared with the acceptance standard threshold to generate a disposal decision instruction for the current polyphenylene ether foam material.
[0012] Secondly, this application provides a multi-parameter performance testing system for polyphenylene ether foam materials, used to perform a multi-parameter performance testing method for polyphenylene ether foam materials, comprising a defect detection unit, the defect detection unit including: The acquisition module is used to trigger the terahertz time-domain spectroscopy sensor and the air-coupled ultrasonic sensor to scan the polyphenylene ether foam material when the conveyor device drives the polyphenylene ether foam material through the detection station, so as to obtain the terahertz time-domain waveform data and ultrasonic transmission signal of the polyphenylene ether foam material. The processing module is used to demodulate the terahertz time-domain waveform data in the time domain, extract a first feature parameter characterizing near-surface defects of the material, and perform frequency domain identification on the ultrasonic transmission signal to extract a second feature parameter characterizing internal defects and thickness changes of the material. The first feature parameter includes the time-domain reflection peak time and frequency band absorption rate, and the second feature parameter includes the sound wave flight time and signal energy attenuation coefficient. The processing module is also used to determine the sensor confidence of the terahertz time-domain spectroscopy sensor and the air-coupled ultrasonic sensor by calculating the probability distribution between the identification results of different defect types of polyphenylene ether foam material by each sensor and the standard results, and then use the confidence of each sensor to perform feature fusion of the first feature parameter and the second feature parameter to obtain the defect feature vector of polyphenylene ether foam material. The execution module is used to distinguish defect types and locate defect locations using the defect feature vectors, obtain defect judgment results for polyphenylene ether foam materials, compare the defect judgment results with pre-stored acceptance standard thresholds, and generate disposal decision instructions for the current polyphenylene ether foam materials.
[0013] Thirdly, this application provides a computer device including a memory and a processor, the memory storing code, and the processor being configured to acquire the code and execute the above-described multi-parameter performance testing method for polyphenylene ether foam materials.
[0014] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method for multi-parameter performance testing of polyphenylene ether foam materials.
[0015] The technical solutions provided by the embodiments disclosed in this application have the following beneficial effects: This application provides a multi-parameter performance testing system and method for polyphenylene ether (PPE) foam materials. When the conveying device carries the PPE foam material through the testing station, a displacement pulse signal from the conveying device triggers a terahertz time-domain spectroscopy sensor and an air-coupled ultrasonic sensor to scan the PPE foam material, obtaining terahertz time-domain waveform data and ultrasonic transmission signals. The terahertz time-domain waveform data is demodulated in the time domain to extract a first characteristic parameter characterizing near-surface defects in the material. The ultrasonic transmission signals are then identified in the frequency domain to extract a second characteristic parameter characterizing internal defects and thickness variations in the material. The first characteristic parameter includes the time-domain peak reflection time and frequency band absorption rate. The second characteristic parameter includes the sound wave flight time and signal energy attenuation coefficient. By calculating the probability distribution between the identification results of different defect types of polyphenylene ether foam material by each sensor and the standard results, the sensor confidence of the terahertz time-domain spectroscopy sensor and the air-coupled ultrasonic sensor is determined. Then, the first characteristic parameter and the second characteristic parameter are fused using the confidence of each sensor to obtain the defect feature vector of polyphenylene ether foam material. The defect feature vector is used to distinguish the defect type and locate the defect position to obtain the defect judgment result of polyphenylene ether foam material. The defect judgment result is compared with the pre-stored acceptance standard threshold to generate the current disposal decision instruction of polyphenylene ether foam material.
[0016] Therefore, in this application, the defect feature vector is used to distinguish defect types and locate defect positions to obtain the defect judgment result of polyphenylene ether foam material. The defect judgment result is compared with the pre-stored acceptance standard threshold to generate the current disposal decision instruction for polyphenylene ether foam material. First, determining the first and second feature parameters can achieve multi-dimensional and complementary quantitative characterization of the material's physical state. The time-domain reflection peak time and frequency band absorption rate extracted from the terahertz time-domain waveform quantify the dielectric anomalies and spectral absorption changes of the material near the surface from both time and frequency domain dimensions, respectively, and are sensitive to defects such as surface cracks, delamination, and density inhomogeneity. The sound wave flight time and signal energy attenuation coefficient extracted from the ultrasonic transmission signal accurately reflect the time delay change and energy scattering absorption degree of the ultrasonic wave propagation path, respectively, and are sensitive to internal pores, inclusions, and thickness changes. This process transforms raw signals from different physical domains (photoelectric and acoustic) into a set of quantitative indicators strongly correlated with specific defect mechanisms, constructing a complementary and orthogonal feature space. This ensures that the input to subsequent fusion modules is physically verified high-value information, improving the effectiveness and specificity of the information from the source and avoiding the noise interference and information redundancy problems commonly encountered when directly fusing raw signals. Then, by determining the defect feature vector, adaptive information fusion and optimization based on dynamic weighting of sensor historical performance can be achieved, significantly improving the system's comprehensive discrimination capability and decision reliability for complex or ambiguous defects. This process introduces prior knowledge of sensor confidence, dynamically transforming the historical recognition accuracy of terahertz and ultrasonic sensors for different types of defects into their weighting coefficients in the current fusion decision, achieving differentiated weighting based on data source reliability. Specifically, when judging near-surface defects, the feature parameters provided by the terahertz sensor are given higher weight; while when evaluating internal defects, the feature parameters of the ultrasonic sensor gain greater influence. The subsequent dimensionality reduction based on principal component analysis further eliminated potential collinearity among the weighted features, generating a refined low-dimensional feature vector with strong independence in each dimension. This feature vector integrates the advantages of multi-source information and suppresses interference from relatively unreliable data, providing the subsequent classifier with an input space that has higher information quality, more significant feature discrimination, and clearer decision boundaries. This enables the system to make more stable and accurate comprehensive judgments when facing complex defects with low signal-to-noise ratios or indistinct features, ultimately improving the overall judgment accuracy. In summary, based on the above scheme, adaptive fusion of multimodal sensing data and defect probability confidence can be achieved, thereby improving the judgment accuracy of online non-destructive testing of polyphenylene ether foam materials. Attached Figure Description
[0017] Figure 1 This is an exemplary flowchart of a multi-parameter performance testing method for polyphenylene ether foam materials according to some embodiments of this application; Figure 2 This is an exemplary flowchart illustrating the determination of defect feature vectors according to some embodiments of this application; Figure 3 This is a schematic diagram of the structure of a defect detection unit according to some embodiments of this application; Figure 4 This is a schematic diagram of the structure of a computer device for implementing a multi-parameter performance testing method for polyphenylene ether foam materials, according to some embodiments of this application. Detailed Implementation
[0018] To better understand the technical solution of this application, the technical solution of this application will be described in detail below with reference to the accompanying drawings and specific embodiments.
[0019] refer to Figure 1 This figure is an exemplary flowchart of a multi-parameter performance testing method for polyphenylene ether foam materials according to some embodiments of this application. The figure mainly includes the following steps: In step 101, when the conveying device carries the polyphenylene ether foam material through the detection station, the displacement pulse signal of the conveying device is used to trigger the terahertz time-domain spectroscopy sensor and the air-coupled ultrasonic sensor to scan the polyphenylene ether foam material, thereby obtaining the terahertz time-domain waveform data and ultrasonic transmission signal of the polyphenylene ether foam material.
[0020] It should be noted that in this application, the displacement pulse signal is a timing signal used to synchronously control the scanning trigger and the spatial position coordinates of the material; the terahertz time-domain spectral sensor is a photoelectric detection device used to emit broadband terahertz pulses and receive the time-domain electric field signal generated after reflection or transmission in the material, so as to obtain the internal physical information of the material; the air-coupled ultrasonic sensor is a high-frequency acoustic detection device used to excite ultrasonic waves in the air in a non-contact manner and receive the ultrasonic wave signal after penetrating the material; the terahertz time-domain waveform data is a time-domain electric field intensity sequence characterizing the dielectric constant distribution, delamination, and near-surface defect information inside the polyphenylene ether foam material; the ultrasonic transmission signal is a sequence of original sound pressure signals characterizing the amplitude attenuation and time delay caused by changes in internal structure, defects, and thickness during the penetration of the polyphenylene ether foam material.
[0021] In practice, firstly, as the conveyor belt continuously carries the polyphenylene ether foam material through the detection station, a photoelectric encoder mounted on the drive shaft of the conveyor belt generates displacement pulse signals in real time, proportional to the distance traveled by the conveyor belt. Whenever the accumulated number of displacement pulse signals reaches the number of pulses corresponding to a preset scanning interval, this pulse signal is sent as a synchronous trigger command to the control units of the terahertz time-domain spectroscopy sensor and the air-coupled ultrasonic sensor. Secondly, upon receiving the trigger command, the terahertz time-domain spectroscopy sensor emits a laser pulse from its femtosecond laser. This pulse is then generated and emitted as a broadband terahertz pulse onto the surface of the polyphenylene ether foam material via a photoconductive antenna. After propagating and reflecting within the material, the pulse is received by the sensor's detection unit and recorded using optical delayed line scanning technology. The complete waveform of the returned terahertz electric field intensity changing over time is used as the terahertz time-domain waveform data for that scanning point. Then, after receiving the same trigger command, the air-coupled ultrasonic sensor emits an ultrasonic pulse with a center frequency of a specific value (e.g., 400 kHz) towards the surface of the polyphenylene ether foam material. After penetrating the material, the ultrasonic pulse is captured by the receiving probe located on the other side of the material. The receiving probe converts the sound pressure signal into an electrical signal. After passing through a preamplifier, the complete voltage amplitude of the electrical signal is recorded as a sequence of changes over time. This voltage amplitude time sequence is used as the ultrasonic transmission signal for that scanning point. Finally, the terahertz time-domain waveform data and ultrasonic transmission signal corresponding to each trigger point are associated and stored to form a pair of original sensing data that correspond one-to-one with the spatial location of the material.
[0022] In step 102, the terahertz time-domain waveform data is demodulated in the time domain to extract a first feature parameter characterizing near-surface defects of the material, and the ultrasonic transmission signal is identified in the frequency domain to extract a second feature parameter characterizing internal defects and thickness changes of the material. The first feature parameter includes the time-domain reflection peak time and frequency band absorption rate, and the second feature parameter includes the sound wave flight time and signal energy attenuation coefficient.
[0023] In some embodiments, the extraction of a first characteristic parameter characterizing near-surface defects in a material by temporal demodulation of the terahertz time-domain waveform data can be achieved using the following steps: Time-domain analysis is performed on the terahertz time-domain waveform data to identify and extract the peak time of its main reflection pulse, which is taken as the time-domain reflection peak time. The terahertz time-domain waveform data is subjected to a fast Fourier transform to obtain its frequency domain spectrum, and the ratio of the energy integral in the specified characteristic frequency band to the reference spectrum is calculated as the frequency band absorption rate. The first characteristic parameter characterizing near-surface defects of the material is determined based on the time-domain reflection peak time and the frequency band absorption rate.
[0024] It should be noted that, in this application, the time-domain reflection peak time is a parameter that quantifies the time it takes for a terahertz pulse to be reflected near the surface of the material and return to the sensor; the frequency band absorptivity is a proportional coefficient that quantifies the degree of change in the energy absorption characteristics of polyphenylene ether foam within a specified frequency range in the terahertz band; and the first characteristic parameter is a set of quantitative indicators that characterize whether there are defects such as cracks, delamination, or density anomalies in the near-surface region of the polyphenylene ether foam.
[0025] In practice, firstly, the pulse peak with the largest absolute amplitude is searched in the waveform data sequence to determine the time corresponding to the peak point. The time interval obtained by subtracting the reference time of the transmitted pulse from this time is the time taken for the main reflection pulse to travel from transmission to return to the sensor. This time interval can be used as the time-domain reflection peak time. Then, a fast Fourier transform is performed on the terahertz time-domain waveform data to convert it from the time domain to the frequency domain, thereby obtaining the frequency domain spectrum of the measurement point. Within a characteristic frequency band selected in advance based on the material properties (e.g., 0.2 to 0.8 terahertz), the square of the amplitude of the frequency domain spectrum is integrated to obtain the spectral energy value of the current measurement point. At the same time, the spectral energy values of the same batch of defect-free standard samples in the same characteristic frequency band are obtained from a pre-established database as a reference. Calculate the ratio of the spectral energy value of the current measuring point to the spectral energy value of the reference spectral point, and use this ratio as the frequency band absorptivity. For example, if the calculated ratio is 1.5, it means that the energy absorption intensity of the current measuring point is 1.5 times that of the defect-free standard sample. Finally, the combination of the time domain reflection peak time and the frequency band absorptivity is used as the first characteristic parameter characterizing near-surface defects of the material.
[0026] In some embodiments, frequency domain identification of the ultrasonic transmission signal to extract a second feature parameter characterizing internal defects and thickness variations of the material can be achieved through the following steps: In the ultrasonic transmission signal, the propagation time of the direct wave from transmission to reception is identified as the sound wave flight time; Calculate the attenuation ratio of the energy amplitude of the ultrasonic transmission signal to the corresponding value of the reference standard signal, and use it as the signal energy attenuation coefficient; A second characteristic parameter characterizing internal defects and thickness variations of a material is determined based on the sound wave flight time and the signal energy attenuation coefficient.
[0027] It should be noted that, in this application, the time of flight of an acoustic wave is a physical quantity that quantifies the time interval required for an ultrasonic wave to propagate from the transmitting sensor to the receiving sensor in the polyphenylene ether foam material under test; the signal energy attenuation coefficient is a scaling parameter that quantifies the proportion of signal energy lost when the ultrasonic wave penetrates the polyphenylene ether foam material under test relative to the energy loss when propagating in a defect-free standard material; and the second characteristic parameter is a set of quantitative indicators that characterize the presence of structural features such as pores, inclusions, or uniform thickness inside the polyphenylene ether foam material.
[0028] In practice, firstly, for each ultrasonic signal acquired at a scanning point, starting from the known laser triggering time, the first significant pulse peak exceeding a preset amplitude threshold is searched in the time series. The moment this peak appears is considered the moment the ultrasonic wave arrives at the receiving sensor. The difference between this arrival time and the laser triggering time is calculated, and this time difference is used as the sound wave flight time at that scanning point. Then, for the complete ultrasonic transmission signal acquired at the current scanning point, within a fixed time window, the root mean square (RMS) value of the signal waveform is calculated. This RMS value characterizes the energy amplitude of the current signal. Simultaneously, from a pre-established database, the RMS value of a reference standard signal obtained by detecting a known defect-free standard sample with standard thickness under the same detection parameters and geometric conditions is obtained. Then, the ratio of the current signal energy amplitude to the reference standard signal energy amplitude is calculated, and this ratio is used as the signal energy attenuation coefficient. For example, if the calculated ratio is 0.7, it means that the energy of the current signal is only 70% of that of the standard reference signal, that is, 30% energy attenuation has occurred. Finally, the set of sound wave flight time and signal energy attenuation coefficient is used as the second characteristic parameter characterizing the internal defects and thickness changes of the material.
[0029] In step 103, by calculating the probability distribution between the identification results of each sensor for different defect types of polyphenylene ether foam material and the standard results, the sensor confidence of the terahertz time-domain spectroscopy sensor and the air-coupled ultrasonic sensor is determined. Then, the first feature parameter and the second feature parameter are fused using the confidence of each sensor to obtain the defect feature vector of polyphenylene ether foam material.
[0030] In some embodiments, the sensor confidence of the terahertz time-domain spectroscopy sensor and the air-coupled ultrasonic sensor can be determined by calculating the probability distribution between the identification results of each sensor for different defect types of polyphenylene ether foam material and the standard results using the following steps: Retrieve all near-surface and internal defects of polyphenylene ether foam materials from the historical defect database; The accuracy of the terahertz time-domain spectroscopy sensor in identifying various near-surface defects is statistically analyzed, and the probability distribution of all identification accuracies is constructed to obtain the sensor confidence of the terahertz time-domain spectroscopy sensor. The accuracy rate of air-coupled ultrasonic sensors in identifying various internal defects is statistically analyzed, and the probability distribution of all identification accuracies is constructed to obtain the sensor confidence level of the air-coupled ultrasonic sensors.
[0031] It should be noted that, in this application, sensor confidence is a weighted coefficient that measures the reliability of the feature information provided by each sensor; the historical defect database is a reference dataset used to store information on polyphenylene ether foam material samples with confirmed defect types, locations, and sizes; near-surface defects are a spatial location classification of defects used to calibrate the detection capability of terahertz time-domain spectroscopy sensors; internal defects are a spatial location classification of defects used to calibrate the detection capability of air-coupled ultrasonic sensors; recognition accuracy is a statistical indicator used to quantify the ability of a single type of sensor to correctly detect a specified type of defect; and probability distribution is a statistical law used to describe the dispersion of recognition accuracy of each sensor for different types of defects.
[0032] In practice, firstly, all defect records of polyphenylene ether foam materials verified by destructive experiments or high-precision computed tomography scans are retrieved from the historical defect database. Based on the burial depth of the defects in the material, they are classified into near-surface defects and internal defects. Then, the ratio of the number of correct identifications of all samples classified as near-surface defects by the terahertz time-domain spectroscopy sensor to the total number of detections in the historical defect database is calculated to obtain the identification accuracy for each near-surface defect. A probability distribution of all identification accuracies is then constructed using kernel density estimation, and all probability distributions are arranged in ascending order of their corresponding identification accuracy. The probability distribution curve of the terahertz time-domain spectroscopy sensor is obtained by connecting the curves. The normalized area under the probability distribution curve of the terahertz time-domain spectroscopy sensor is used as the sensor confidence level. Finally, the recognition accuracy of the air-coupled ultrasonic sensor for all internal defect samples is statistically analyzed in the historical defect database. The probability distribution of all recognition accuracy is constructed by kernel density estimation. All probability distributions are connected in ascending order of the corresponding recognition accuracy to obtain the probability distribution curve of the air-coupled ultrasonic sensor. The normalized area under the probability distribution curve of the air-coupled ultrasonic sensor is used as the sensor confidence level.
[0033] In some embodiments, the confidence levels of each sensor are used to fuse the first feature parameter and the second feature parameter to obtain a defect feature vector of the polyphenylene ether foam material, with reference to... Figure 2 The diagram is a flowchart illustrating the process of determining defect feature vectors in some embodiments of this application. In this embodiment, the determination of defect feature vectors can be achieved using the following steps: In step 1031, the first feature parameter and the second feature parameter are combined to form an initial feature set; In step 1032, the confidence scores of each sensor are used to perform weighted fusion of each feature parameter in the initial feature set to obtain a fused feature vector; In step 1033, principal component analysis is performed on the fused feature vector, and then a low-correlation feature vector is formed after dimensionality reduction, thereby obtaining the defect feature vector of polyphenylene ether foam material.
[0034] It should be noted that in this application, the defect feature vector is an optimized feature set used to input the classification model for accurate defect identification and localization; the initial feature set is a multivariate data set used to gather raw feature information from different sensors; the fusion feature vector is an intermediate feature representation used to reflect the comprehensive information from multiple sensors after reliability weighting; and the low-correlation feature vector is a feature set used to eliminate information redundancy and highlight significant defect patterns.
[0035] In practice, firstly, the first and second feature parameters are concatenated along their feature dimensions to form an initial feature set. Then, based on a predetermined confidence level of the terahertz time-domain spectral sensor, scalar multiplication is performed on all feature components belonging to the first feature parameter in the initial feature set. Simultaneously, based on the confidence level of the air-coupled ultrasonic sensor, scalar multiplication is performed on all feature components belonging to the second feature parameter, resulting in a weighted fused feature vector. Finally, principal component analysis is performed on this fused feature vector. This process calculates the covariance matrix of the fused feature vector, its eigenvalues, and eigenvectors. The eigenvectors corresponding to the largest eigenvalues are selected as projection directions, and the original fused feature vector is projected onto these principal component directions, achieving dimensionality reduction. The low-correlation feature vectors formed after dimensionality reduction through principal component analysis are then used as the defect feature vectors of the polyphenylene ether foam material.
[0036] In step 104, the defect feature vector is used to distinguish the defect type and locate the defect location to obtain the defect judgment result of the polyphenylene ether foam material. The defect judgment result is compared with the pre-stored acceptance standard threshold to generate the disposal decision instruction for the current polyphenylene ether foam material.
[0037] In some embodiments, the defect feature vector is used to distinguish defect types and locate defect positions to obtain the defect determination result of polyphenylene ether foam material, which can be achieved by the following steps: The defect feature vector is input into a pre-trained support vector machine multi-classification model. The classification model outputs a predicted label for the defect type, which includes surface cracks, internal pores, inclusions, and no defects, thereby obtaining the defect type. The defect label is mapped onto the three-dimensional digital model of the polyphenylene ether foam material using the displacement pulse signal triggered by the scan, thereby obtaining the defect location; The defect determination result of polyphenylene ether foam material is determined by the defect type and the defect location.
[0038] It should be noted that, in this application, the defect determination result is a qualitative and quantitative conclusion reflecting the current quality status of the tested polyphenylene ether foam material; the multi-classification model is a machine learning model used to classify and distinguish the health status of the polyphenylene ether foam material; the defect type prediction label is a classification identifier used to qualitatively describe the types of defects or defect-free states present in the polyphenylene ether foam material; the defect type is a qualitative judgment conclusion used to describe the specific types of discontinuities in the polyphenylene ether foam material; the three-dimensional digital model is a computer-aided design model characterizing the geometric shape and spatial structure of the tested polyphenylene ether foam material; and the defect location is the location information that quantitatively describes the coordinates of the identified defects in the three-dimensional space of the polyphenylene ether foam material.
[0039] In practice, firstly, the defect feature vector is input into a pre-trained support vector machine multi-classification model using historical defect sample data. This model finds a decision hyperplane in the feature space that optimally separates samples of different categories, and calculates and discriminates the input defect feature vector accordingly, outputting a predicted label representing the specific defect type. This predicted label includes surface cracks, internal pores, inclusions, and no defects, and is thus used as the defect type. Next, the position mapping module records the cumulative displacement of the transmission device corresponding to each scan trigger signal. This displacement defines the coordinates of the detection point along the material's length. Combining the fixed arrangement coordinates of the sensor array along the material's width and the material cross-sectional reference defined in the 3D digital model, the defect label is associated with a 2D scan coordinate. Through coordinate transformation, this 2D scan coordinate is precisely mapped to the 3D spatial coordinates in the 3D digital model of the polyphenylene ether foam material, thus obtaining the defect location. Finally, the set of defect types and defect locations is used as the defect determination result for the polyphenylene ether foam material.
[0040] In some embodiments, comparing the defect determination result with a pre-stored acceptance standard threshold to generate a disposal decision instruction for the current polyphenylene ether foam material can be achieved through the following steps: The acceptance standard threshold for the defect type in the defect judgment result is preset based on the design standard; The defect size in the defect determination result is compared with the acceptance standard threshold to generate a disposal decision instruction for the current polyphenylene ether foam material.
[0041] It should be noted that in this application, the disposal decision instruction is an operational command used to guide the production process to perform subsequent processing on the currently inspected polyphenylene ether foam material; the acceptance standard threshold is a critical size value used to determine whether a specified type of defect in the polyphenylene ether foam material is allowed to exist; and the defect size is a geometric quantity used to quantify the severity of the identified defects in the polyphenylene ether foam material.
[0042] In practice, firstly, based on the design standards and process specifications of polyphenylene ether foam materials, a clear acceptance standard threshold is preset for each type of defect in the defect judgment results. For example, the maximum allowable length of surface cracks and the maximum allowable equivalent diameter of internal pores are set. Then, the quantitative defect size of the identified defects is extracted from the defect judgment results, and the defect size of each defect is compared with the acceptance standard threshold of its corresponding type one by one. According to the comparison results, if the defect size of all defects is lower than its corresponding acceptance standard threshold, a "qualified" instruction is generated. If the defect size of any defect exceeds its corresponding acceptance standard threshold but is within the repairable range, a "rework" instruction is generated. If the defect size of any defect exceeds its corresponding acceptance standard threshold and is irreparable, a "scrap" instruction is generated. The conclusion drawn from this comparison and logical judgment is used as the disposal decision instruction for the current polyphenylene ether foam material.
[0043] In another aspect, in some embodiments, this application provides a multi-parameter performance testing system for polyphenylene ether foam materials. This system includes a defect detection unit, referencing... Figure 3 The figure is a schematic diagram of the structure of a defect detection unit according to some embodiments of this application. The defect detection unit includes: a data acquisition module 201, a processing module 202, and an execution module 203, which are described below: The acquisition module 201 in this application is mainly used to use the displacement pulse signal of the conveying device to trigger the terahertz time-domain spectroscopy sensor and the air-coupled ultrasonic sensor to scan the polyphenylene ether foam material when the conveying device drives the polyphenylene ether foam material through the detection station, so as to obtain the terahertz time-domain waveform data and ultrasonic transmission signal of the polyphenylene ether foam material. Processing module 202, in this application, is used to demodulate the terahertz time-domain waveform data in the time domain, extract a first feature parameter characterizing near-surface defects of the material, and perform frequency domain identification on the ultrasonic transmission signal to extract a second feature parameter characterizing internal defects and thickness changes of the material. The first feature parameter includes the time-domain reflection peak time and frequency band absorption rate, and the second feature parameter includes the sound wave flight time and signal energy attenuation coefficient. It should be noted that the processing module 202 is also used to determine the sensor confidence of the terahertz time-domain spectroscopy sensor and the air-coupled ultrasonic sensor by calculating the probability distribution between the identification results of different defect types of polyphenylene ether foam material by each sensor and the standard results, and then use the confidence of each sensor to perform feature fusion of the first feature parameter and the second feature parameter to obtain the defect feature vector of polyphenylene ether foam material. The execution module 203 in this application is mainly used to distinguish the defect type and locate the defect location using the defect feature vector, obtain the defect judgment result of the polyphenylene ether foam material, compare the defect judgment result with the pre-stored acceptance standard threshold, and generate the disposal decision instruction of the current polyphenylene ether foam material.
[0044] The foregoing has detailed examples of a multi-parameter performance testing system and method for polyphenylene ether foam materials provided in the embodiments of this application. It is understood that the corresponding apparatus includes hardware structures and / or software modules for performing each function in order to achieve the aforementioned functions. Those skilled in the art should readily recognize that, based on the units and algorithm steps of the examples described in conjunction with the embodiments disclosed herein, this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed in hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.
[0045] In some embodiments, this application also provides a computer device, the computer device including a memory and a processor, the memory for storing a computer program, and the processor for calling and running the computer program from the memory, so that the computer device performs the above-described multi-parameter performance testing method for polyphenylene ether foam materials.
[0046] In some embodiments, reference Figure 4 The dashed lines in the figure indicate that the unit or module is optional. This figure is a schematic diagram of the structure of a computer device for implementing a multi-parameter performance testing method for polyphenylene ether foam materials according to an embodiment of this application. The multi-parameter performance testing method for polyphenylene ether foam materials described in the above embodiments can be achieved through… Figure 4 The computer device shown is used to implement this, and the computer device includes at least one processor 301, a memory 302 and at least one communication unit 305. The computer device may be a terminal device, a server or a chip.
[0047] Processor 301 can be a general-purpose processor or a special-purpose processor. For example, processor 301 can be a central processing unit (CPU), which can be used to control computer devices, execute software programs, and process data from software programs. The computer device may also include a communication unit 305 for inputting (receiving) and outputting (transmitting) signals.
[0048] For example, the computer device may be a chip, and the communication unit 305 may be the input and / or output circuit of the chip, or the communication unit 305 may be the communication interface of the chip, which may be a component of a terminal device, network device or other device.
[0049] For example, the computer device may be a terminal device or a server, and the communication unit 305 may be a transceiver of the terminal device or the server, or the communication unit 305 may be a transceiver circuit of the terminal device or the server.
[0050] The computer device may include one or more memories 302 storing a program 304. The program 304 can be executed by a processor 301 to generate instructions 303, causing the processor 301 to execute the method described in the above method embodiments according to the instructions 303. Optionally, the memory 302 may also store data (such as a target audit model). Optionally, the processor 301 may also read data stored in the memory 302, which may be stored at the same storage address as the program 304, or it may be stored at a different storage address than the program 304.
[0051] The processor 301 and memory 302 can be configured separately or integrated together, for example, integrated on the system on chip (SOC) of the terminal device.
[0052] It should be understood that each step of the above method embodiment can be completed by hardware logic circuits or software instructions in the processor 301. The processor 301 can be a CPU, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, such as discrete gates, transistor logic devices, or discrete hardware components.
[0053] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0054] For example, in some embodiments, this application also provides a computer-readable storage medium storing instructions or code that, when executed on a computer, cause the computer to implement the above-described multi-parameter performance testing method for polyphenylene ether foam materials.
[0055] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.
[0056] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.
Claims
1. A method for testing the multi-parameter performance of polyphenylene ether foam materials, characterized in that, Includes the following steps: When the conveyor carries the polyphenylene ether foam material through the detection station, the displacement pulse signal of the conveyor is used to trigger the terahertz time-domain spectroscopy sensor and the air-coupled ultrasonic sensor to scan the polyphenylene ether foam material, thereby obtaining the terahertz time-domain waveform data and ultrasonic transmission signal of the polyphenylene ether foam material. The terahertz time-domain waveform data is demodulated in the time domain to extract a first feature parameter characterizing near-surface defects of the material, and the ultrasonic transmission signal is identified in the frequency domain to extract a second feature parameter characterizing internal defects and thickness changes of the material. The first feature parameter includes the time-domain reflection peak time and frequency band absorption rate, and the second feature parameter includes the sound wave flight time and signal energy attenuation coefficient. By calculating the probability distribution between the identification results of different defect types of polyphenylene ether foam material by each sensor and the standard results, the sensor confidence of the terahertz time-domain spectroscopy sensor and the air-coupled ultrasonic sensor is determined. Then, the first feature parameter and the second feature parameter are fused using the confidence of each sensor to obtain the defect feature vector of polyphenylene ether foam material. The defect feature vector is used to distinguish the defect type and locate the defect location to obtain the defect judgment result of the polyphenylene ether foam material. The defect judgment result is compared with the pre-stored acceptance standard threshold to generate the disposal decision instruction for the current polyphenylene ether foam material.
2. The method as described in claim 1, characterized in that, Demodulating the terahertz time-domain waveform data in the time domain and extracting the first characteristic parameter characterizing near-surface defects in the material specifically includes: Time-domain analysis is performed on the terahertz time-domain waveform data to identify and extract the peak time of the main reflection pulse as the time-domain reflection peak time; The terahertz time-domain waveform data is subjected to a fast Fourier transform to obtain a frequency-domain spectrum, and the ratio of the energy integral in a specified characteristic frequency band to the reference spectrum is calculated as the frequency band absorption rate. The first characteristic parameter characterizing near-surface defects of the material is determined based on the time-domain reflection peak time and the frequency band absorption rate.
3. The method as described in claim 1, characterized in that, The frequency domain identification of the ultrasonic transmission signal, and the extraction of the second feature parameter characterizing the internal defects and thickness changes of the material, specifically includes: In the ultrasonic transmission signal, the propagation time of the direct wave from transmission to reception is identified as the sound wave flight time; The attenuation ratio of the energy amplitude of the ultrasonic transmission signal to the corresponding value of the reference standard signal is calculated as the signal energy attenuation coefficient. A second characteristic parameter characterizing internal defects and thickness variations of a material is determined based on the sound wave flight time and the signal energy attenuation coefficient.
4. The method as described in claim 1, characterized in that, By calculating the probability distribution between the identification results of each sensor for different defect types of polyphenylene ether foam material and the standard results, the sensor confidence levels of the terahertz time-domain spectroscopy sensor and the air-coupled ultrasonic sensor are determined, specifically including: Retrieve all near-surface and internal defects of polyphenylene ether foam material from the historical defect database; The accuracy of the terahertz time-domain spectroscopy sensor in identifying various near-surface defects is statistically analyzed, and the probability distribution of all identification accuracies is constructed to obtain the sensor confidence of the terahertz time-domain spectroscopy sensor. The accuracy rate of air-coupled ultrasonic sensors in identifying various internal defects is statistically analyzed, and the probability distribution of all identification accuracies is constructed to obtain the sensor confidence level of the air-coupled ultrasonic sensors.
5. The method as described in claim 1, characterized in that, The first feature parameter and the second feature parameter are fused using the confidence scores of each sensor to obtain the defect feature vector of the polyphenylene ether foam material, specifically including: The first feature parameter and the second feature parameter are combined to form an initial feature set; The confidence scores of each sensor are used to weight and fuse the feature parameters in the initial feature set to obtain a fused feature vector. Principal component analysis is performed on the fused feature vector, and then the dimension is reduced to form a low-correlation feature vector, thereby obtaining the defect feature vector of polyphenylene ether foam material.
6. The method as described in claim 1, characterized in that, Using the defect feature vector to distinguish defect types and locate defect positions, the specific defect determination results for polyphenylene ether foam materials include: The defect feature vector is input into a pre-trained support vector machine multi-classification model. The classification model outputs a predicted label for the defect type, which includes surface cracks, internal pores, inclusions, and no defects, thereby obtaining the defect type. The defect label is mapped onto the three-dimensional digital model of the polyphenylene ether foam material using the displacement pulse signal triggered by the scan, thereby obtaining the defect location; The defect determination result of polyphenylene ether foam material is determined by the defect type and the defect location.
7. The method as described in claim 1, characterized in that, The defect determination result is compared with the pre-stored acceptance standard threshold to generate the current disposal decision instruction for the polyphenylene ether foam material. Specifically, this includes: The acceptance standard threshold for the defect type in the defect judgment result is preset based on the design standard; The defect size in the defect determination result is compared with the acceptance standard threshold to generate a disposal decision instruction for the current polyphenylene ether foam material.
8. A multi-parameter performance testing system for polyphenylene ether foam materials, used to perform the multi-parameter performance testing method for polyphenylene ether foam materials as described in any one of claims 1 to 7, wherein the multi-parameter performance testing system for polyphenylene ether foam materials includes a defect detection unit, characterized in that, The defect detection unit includes: The acquisition module is used to trigger the terahertz time-domain spectroscopy sensor and the air-coupled ultrasonic sensor to scan the polyphenylene ether foam material when the conveyor device drives the polyphenylene ether foam material through the detection station, so as to obtain the terahertz time-domain waveform data and ultrasonic transmission signal of the polyphenylene ether foam material. The processing module is used to demodulate the terahertz time-domain waveform data in the time domain, extract a first feature parameter characterizing near-surface defects of the material, and perform frequency domain identification on the ultrasonic transmission signal to extract a second feature parameter characterizing internal defects and thickness changes of the material. The first feature parameter includes the time-domain reflection peak time and frequency band absorption rate, and the second feature parameter includes the sound wave flight time and signal energy attenuation coefficient. The processing module is also used to determine the sensor confidence of the terahertz time-domain spectroscopy sensor and the air-coupled ultrasonic sensor by calculating the probability distribution between the identification results of different defect types of polyphenylene ether foam material by each sensor and the standard results, and then use the confidence of each sensor to perform feature fusion of the first feature parameter and the second feature parameter to obtain the defect feature vector of polyphenylene ether foam material. The execution module is used to distinguish defect types and locate defect locations using the defect feature vectors, obtain defect judgment results for polyphenylene ether foam materials, compare the defect judgment results with pre-stored acceptance standard thresholds, and generate disposal decision instructions for the current polyphenylene ether foam materials.
9. A computer device, characterized in that, The computer device includes a memory and a processor, the memory storing code, and the processor being configured to retrieve the code and execute the multi-parameter performance testing method for polyphenylene ether foam materials as described in any one of claims 1 to 7.
10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the multi-parameter performance testing method for polyphenylene ether foam materials as described in any one of claims 1 to 7.