A method for non-inductive operation quality traceability in the glass fiber industry

Abstract

The application discloses a kind of glass fiber industry's non-inductive operation quality traceability method, it is related to glass fiber quality traceability technical field, including, selection PZT sensor is installed, cooperate analog-digital converter to obtain acquisition data, respectively in low frequency interval and high frequency interval calculate impedance value and identify feature, calculate signal-to-noise ratio to confirm identification characteristic frequency point, the change rate of adjacent identification characteristic frequency point is analyzed.The method disclosed in the application achieves high sensitivity detection on the internal structure and defects of the material by fine measurement of the impedance of the glass fiber epoxy plate at low and high frequencies, combined with the calculation and analysis of signal-to-noise ratio, by artificially introducing adhesive defects, the damage that may occur during actual use of the glass fiber epoxy plate can be simulated, and impedance data of the material under different damage states can be collected, by real-time monitoring of the electrical impedance value of the glass fiber epoxy plate at the key frequency point, the internal micro changes of the material can be detected immediately, thereby early damage identification is realized.

Description

Technical Field

[0001] This invention relates to the field of glass fiber quality traceability technology, and in particular to a non-intrusive quality traceability method for the glass fiber industry. Background Technology

[0002] Fiberglass materials are widely used in aerospace, automobile manufacturing and construction engineering due to their excellent mechanical properties, heat resistance and electrical insulation. Among them, fiberglass epoxy board is composed of glass fiber and epoxy resin, and has good adhesion, chemical stability and electrical insulation. This board combines the high strength of glass fiber and the excellent properties of epoxy resin, and is a common fiberglass material.

[0003] During the production of fiberglass materials, minute defects often occur, such as microcracks, bubbles, or poor adhesion. The presence of these defects can have a significant impact on the performance and safety of the final product. Traditional quality control methods, such as visual inspection, ultrasonic inspection, and X-ray inspection, can detect macroscopic defects in materials to a certain extent, but they still have limitations in the detection and real-time monitoring of microscopic defects. It is inconvenient to effectively analyze, monitor, and trace defects inside the material, which not only affects the overall performance of the material but also reduces the efficiency of quality traceability. Summary of the Invention

[0004] In view of the problems existing in the above-mentioned non-contact operation quality traceability methods in the fiberglass industry, this invention is proposed.

[0005] Therefore, the problem that this invention aims to solve is that there are still limitations in the detection and real-time monitoring of micro-defects. It is inconvenient to effectively analyze, monitor and trace defects inside materials, which not only affects the overall performance of materials but also reduces the efficiency of quality traceability.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a non-intrusive operation quality traceability method for the fiberglass industry, comprising,

[0007] PZT sensors are selected for installation, and data is acquired using an analog-to-digital converter. Impedance values ​​are calculated and features are identified in both the low-frequency and high-frequency ranges. The signal-to-noise ratio is calculated to confirm the identified feature frequency points. The rate of change of adjacent identified feature frequency points is analyzed to determine the frequency interval and integrate the scanning frequency range.

[0008] Impedance measurements were performed based on the introduced bonding defects in the sample. The real part of the impedance was calculated based on the measured voltage and current. The specific impact of the introduced defects on the impedance characteristics of the glass fiber epoxy board sample was quantified. The defect threshold of RMSD value was calculated based on historical data and the frequency points were marked.

[0009] Material parameters are obtained and electrical impedance is calculated based on the scanning frequency range. The correlation between RMSD values ​​and electrical impedance is analyzed based on the marked frequency points. The t-distribution is used for testing, and the t-statistic is calculated to screen frequency points with strong linear relationships. Sensitive frequency ranges are integrated based on frequency intervals.

[0010] A random forest model is constructed and trained using K-fold cross-validation. Frequency points within a sensitive frequency range are input into the model, and the prediction threshold is determined through the cumulative distribution function. The model output is then evaluated.

[0011] Based on the model's output, remote early warnings are issued and warning information is sent to multiple terminals. Data is stored in a distributed manner and transmitted with real-time encrypted data. Access permissions are defined and backup log data is backed up using DES encryption.

[0012] As a preferred embodiment of the non-contact operation quality traceability method for the fiberglass industry described in this invention, the step of selecting and installing a PZT sensor, and using an analog-to-digital converter to acquire the collected data, includes:

[0013] Select the PZT sensor for installation, clean the selected bonding location on the fiberglass epoxy board, and use high-performance epoxy resin adhesive to attach the cut PZT sensor to the selected bonding location. Let it stand for 24 hours to allow the adhesive to fully cure and the bonding to be completed.

[0014] Electrically connect the output of the PZT sensor to the input of the analog-to-digital converter, and electrically connect the output of the analog-to-digital converter to the input of the computer via a USB interface.

[0015] Configure the computer driver and read the output data of the analog-to-digital converter.

[0016] As a preferred embodiment of the non-contact operation quality traceability method for the fiberglass industry described in this invention, the steps of calculating impedance values ​​and identifying features in the low-frequency and high-frequency ranges respectively, calculating the signal-to-noise ratio to confirm the identified feature frequency points, analyzing the rate of change of adjacent identified feature frequency points, determining the frequency interval, and integrating the scanning frequency range include:

[0017] Based on the scenario where the PZT sensor is not connected to the glass fiber epoxy board sample, the background noise is measured by using the PZT sensor and analog-to-digital converter, and the noise level in the range of 1 kHz to 1 MHz is recorded.

[0018] Based on the scenario of connecting a PZT sensor to a glass fiber epoxy board sample, measurements were taken using the PZT sensor and an analog-to-digital converter in both low-frequency and high-frequency ranges, recording the impedance value at each frequency point. The impedance calculation in the low-frequency range is expressed as follows:

[0019] ;

[0020] in The impedance represents the low-frequency range at frequency ω, where R represents resistance and C represents capacitance. This represents parallel resistance, and i represents the imaginary unit.

[0021] The impedance in the high-frequency range is calculated by dividing the voltage at frequency ω by the current at frequency ω.

[0022] Based on the calculated impedance data at different frequencies, the relationship between frequency and impedance is plotted in the graphing software Matlab, and the identification features are determined according to the peak and inflection point of the relationship.

[0023] The signal-to-noise ratio (SNR) is calculated based on the recognition features, and historical experience values ​​are used as the SNR threshold. Recognition features that are greater than or equal to the SNR threshold are selected.

[0024] Based on the selected characteristic frequency points, the rate of change of impedance at adjacent frequency points is calculated. A baseline step size for adjacent frequency steps is set based on historical experience. Using the 10th percentile of the initial rate of change as the threshold, the baseline step size is gradually expanded to adjust the frequency points, as shown below:

[0025] ;

[0026] in Representing frequency point The rate of change of impedance, Representing frequency point impedance;

[0027] Compare the rate of change calculated each time with the threshold. When the rate of change is lower than the set threshold, stop adjusting the frequency point and confirm the frequency interval based on the identified characteristic frequency points.

[0028] Integration is performed based on extended frequency intervals, and the overlapping parts of the frequency intervals are integrated to form a continuous frequency range, which is then set as the scanning frequency range.

[0029] As a preferred embodiment of the non-intrusive operation quality traceability method for the fiberglass industry described in this invention, the method includes: impedance measurement based on introduced bonding defects in the sample; calculation of the real part of the impedance based on the measured voltage and current; quantification of the specific impact of the introduced defects on the impedance characteristics of the fiberglass epoxy board sample; calculation of the defect threshold for RMSD value based on historical data; and marking frequency points.

[0030] Based on the adhesive defects introduced in the fiberglass epoxy board sample, the defects were cut into the fiberglass epoxy board sample using a cutting tool, and the location and size of each defect were accurately marked on the fiberglass epoxy board with a marking pen;

[0031] Impedance measurements were performed using a PZT sensor across a scanning frequency range, providing information on the impedance, location, and magnitude at each measurement frequency point.

[0032] Based on the measurement frequency point k where the adhesion defect is introduced, the voltage is measured using an impedance analyzer. and current The time-domain signal is used to perform Fourier transforms on the voltage and current signals at each frequency point k to obtain the voltage phase. and current phase Through voltage phase and current phase Phase angle calculation by subtraction ;

[0033] At frequency point k, based on the measured voltage and current values, the real part of the impedance is calculated to quantify the specific impact of the introduced defects on the impedance characteristics of the glass fiber epoxy board sample and to calculate... value;

[0034] For each group of data before and after the introduction of adhesive defects, RMSD calculation was applied to calculate the RMSD values ​​at all frequency points. The calculation results were compiled into a table, indicating the RMSD values ​​under different defect states.

[0035] The defect threshold is the sum of the average RMSD value and twice the standard deviation of the historical sample data. Frequency points with RMSD values ​​greater than or equal to the defect threshold are marked as significant damage.

[0036] As a preferred embodiment of the non-contact operation quality traceability method for the fiberglass industry described in this invention, the following steps are included: obtaining material parameters and calculating impedance based on the scanning frequency range; analyzing the correlation between RMSD values ​​and impedance based on marked frequency points; performing a t-distribution test; calculating the t-statistic and screening frequency points with strong linear relationships; and integrating sensitive frequency ranges based on frequency intervals, including...

[0037] Frequency scanning was performed on a fiberglass epoxy board with a PZT sensor attached, using a PZT sensor and an analog-to-digital converter. The scanning range was based on the scanning frequency range. Material parameters, including dielectric constant, dielectric loss factor, piezoelectric coefficient, and elastic modulus, were obtained, and the electrical impedance was calculated, expressed as:

[0038] ;

[0039] ;

[0040] Where p represents the material density, v represents the speed at which the vibrating sound wave propagates through the material, and A represents the cross-sectional area of ​​the material. Indicates admittance. Represents the relative permittivity. This represents the dielectric loss factor, which is related to the energy loss caused by polarization within the material. and These represent the material's inherent mechanical resistance and the added resistance, respectively. Indicates the piezoelectric coefficient. This represents the elastic modulus, where i represents the imaginary unit. Indicates the sensor scaling factor;

[0041] Collect the RMSD values ​​and corresponding admittances of frequency points marked as showing significant damage. The correlation was calculated using the Pearson correlation coefficient, and the t-distribution was used for testing. The t-statistic was calculated, and the corresponding critical value of the t-distribution was found by consulting the t-distribution table. And further calculate the correlation threshold. ;

[0042] If the calculated value of r is greater than or equal to Then judge and There is a strong linear relationship between them, and it is determined that the corresponding frequency point k is more sensitive to damage;

[0043] Based on the frequency intervals between all adjacent frequency points with strong linear relationships, the average value of the frequency intervals is calculated. ,Will The sensitive range of each strongly linear frequency point is determined by the floating value, and overlapping sensitive ranges are merged to form a sensitive frequency range.

[0044] As a preferred embodiment of the non-contact operation quality traceability method for the fiberglass industry described in this invention, the following steps are included: constructing a random forest model and training it using K-fold cross-validation; inputting frequency points within a sensitive frequency range into the model; determining the prediction threshold using the cumulative distribution function; and judging the model output, including...

[0045] Admittance calculated based on the sensitive frequency range The values ​​are used to construct the model dataset, including... The value and the corresponding status label;

[0046] Construct a random forest model and input different frequency points from the model dataset into the model's decision tree. value;

[0047] The model samples the input data multiple times and constructs multiple decision trees. Each decision tree independently classifies the input data. The model then performs a weighted average of the prediction results from all decision trees and selects the maximum value as the final prediction output based on the probability value output by the model.

[0048] During model training, K-fold cross-validation is used. The dataset is split, with one part used as the validation set and the rest as the training set. This is repeated K times, with different data selected as the validation set each time. After each training session, the model is scored using the validation set, and the model's performance metrics are recorded. The optimal combination of model parameters is selected for final training, and the model parameters are confirmed.

[0049] The model output probabilities are sorted based on the model training process, and the cumulative distribution function value of each probability value is calculated. And determine the prediction threshold;

[0050] In actual monitoring, the system will collect admittance data within the sensitive frequency range in real time. The value is input into the random forest model. The maximum value is selected as the final prediction output based on the probability value output by the model. If the obtained prediction output value is greater than or equal to the prediction threshold, it is judged as quality impairment.

[0051] As a preferred embodiment of the non-contact operation quality traceability method for the fiberglass industry described in this invention, the following steps are taken: Based on the model output, remote early warning is issued and early warning information is sent via multiple terminals. This refers to remote monitoring and early warning based on the judgment results of the random forest model and the Internet of Things (IoT). If a signal of quality damage is detected, an early warning message is sent via SMS gateway and via email to the operator via SMTP server. Specific early warning information is displayed on the monitoring interface, and the integrated PLC industrial controller directly triggers the physical alarm device on-site.

[0052] As a preferred embodiment of the non-contact operation quality traceability method for the fiberglass industry described in this invention, the method includes: distributing data storage, real-time encrypted data transmission, defining access permissions, and backing up log data using DES encryption, including...

[0053] The system employs a distributed storage approach to securely store the collected and computed data in real time. It uses the end-to-end encryption protocol TLS to encrypt the collected and computed data and transmit them to the storage device. Simultaneously, it distributes the data in a multi-node cloud database and performs replication and backup of the multi-node data in the cloud database.

[0054] Role-based access control (RBAC) defines access permissions, clarifies data access permissions in the cloud database, records all access and operation data, records the access and operation data in log files, encrypts the log data using DES, and periodically backs it up to off-site storage.

[0055] A computer device includes: a memory and a processor; the memory stores a computer program, and the processor executes the computer program to implement the steps of the above-described non-contact operation quality traceability method for the fiberglass industry.

[0056] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the aforementioned non-contact operation quality traceability method for the fiberglass industry.

[0057] The beneficial effects of this invention are as follows: By performing refined impedance measurements of fiberglass epoxy boards at low and high frequencies, and combining this with signal-to-noise ratio calculation and analysis, high-sensitivity detection of the internal structure and defects of the material can be achieved. By artificially introducing bonding defects, damage that may occur to the fiberglass epoxy board during actual use can be simulated, and impedance data of the material under different damage states can be collected. By real-time monitoring of the impedance values ​​of the fiberglass epoxy board at key frequency points, minute changes within the material can be detected immediately, thereby achieving early damage identification, realizing high-precision monitoring technology, and improving the overall quality and reliability of the product. Through real-time monitoring of impedance values ​​within the sensitive frequency range, subtle changes in the material can be detected in a timely manner, and the possibility of damage can be predicted. Attached Figure Description

[0058] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0059] Figure 1 This is a flowchart illustrating a non-intrusive operation quality traceability method for the fiberglass industry.

[0060] Figure 2 A schematic diagram illustrating the correlation analysis process for a non-contact operation quality traceability method in the fiberglass industry. Detailed Implementation

[0061] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0062] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0063] Secondly, the term "an embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places throughout this specification does not necessarily refer to the same embodiment, nor is it a single embodiment or an embodiment selectively excluded from other embodiments.

[0064] Example 1, referring to Figure 1 and Figure 2 This is the first embodiment of the present invention, which provides a method for non-contact operation quality traceability in the fiberglass industry. The method includes...

[0065] S1. Select the PZT sensor for installation, use it with an analog-to-digital converter to acquire data, calculate the impedance value and identify features in the low-frequency and high-frequency ranges respectively, calculate the signal-to-noise ratio to confirm the identified feature frequency points, analyze the rate of change of adjacent identified feature frequency points, determine the frequency interval and integrate the scanning frequency range.

[0066] Preferably, a PZT sensor is selected for installation, and a digital-to-analog converter is used to acquire the collected data, including...

[0067] If you choose a PZT sensor, you can select the PZT sensor model 5A4E for installation.

[0068] Clean the selected bonding location on the fiberglass epoxy board, and use high-performance epoxy resin adhesive to attach the cut PZT sensor to the selected bonding location. Let it stand for 24 hours to allow the adhesive to fully cure and the bonding to be completed.

[0069] Electrically connect the output of the PZT sensor to the input of the analog-to-digital converter (ADC), and electrically connect the output of the ADC to the input of the computer via a USB interface. An AD5933 ADC can be used.

[0070] Configure the computer driver and read the output data of the analog-to-digital converter.

[0071] By firmly bonding the cut PZT sensor to the fiberglass epoxy board using high-performance epoxy resin adhesive, a tight bond between the sensor and the material is ensured. The proper use of the sensor allows for precise capture of minute changes in the material's internal electrical impedance, providing a foundation for system data acquisition. Electrically connecting the PZT sensor to the AD5933 analog-to-digital converter and a computer enables efficient real-time data acquisition and processing. Through computer-driven configuration and data reading, the system can automatically collect and analyze data. This measurement system combining the PZT sensor and analog-to-digital converter provides the fiberglass industry with an efficient, accurate, and seamless quality monitoring solution.

[0072] Furthermore, impedance values ​​are calculated and features are identified in both the low-frequency and high-frequency ranges. The signal-to-noise ratio is calculated to confirm the identified feature frequency points. The rate of change of adjacent identified feature frequency points is analyzed to determine the frequency interval and integrate the scanning frequency range, including...

[0073] Based on the scenario where the PZT sensor is not connected to the glass fiber epoxy board sample, the background noise is measured by using the PZT sensor and analog-to-digital converter, and the noise level in the range of 1 kHz to 1 MHz is recorded.

[0074] Based on the scenario of connecting a PZT sensor to a glass fiber epoxy board sample, measurements were taken using the PZT sensor and an analog-to-digital converter in the low-frequency (1kHz-10kHz) and high-frequency (10kHz-1MHz) ranges, respectively. The impedance value corresponding to each frequency point was recorded. The impedance calculation in the low-frequency range is expressed as follows:

[0075] ;

[0076] in The impedance represents the low-frequency range at frequency ω, where R represents resistance and C represents capacitance. This represents parallel resistance, and i represents the imaginary unit.

[0077] The impedance in the high-frequency range is expressed as:

[0078] ;

[0079] in The impedance representing the high-frequency range at frequency ω. Voltage representing frequency ω The current representing the frequency ω;

[0080] At low frequencies, for fiberglass epoxy boards, the capacitive characteristics of the internal structure, such as dielectric constant and dielectric layer thickness, can significantly affect impedance readings. Furthermore, due to the slower current flow, the capacitor's charge-discharge cycle has a greater impact on the measurement results. This is further amplified by the parallel resistor. The composite circuit formed by the capacitor C reflects the non-uniformity and potential defects within the material, therefore The calculation is advantageous for detecting minute physical changes such as cracks or voids;

[0081] At high frequencies, the influence of capacitors decreases as the frequency increases, and resistance becomes the main influencing factor. Furthermore, the current changes rapidly, and the charging and discharging of capacitors has a relatively small impact on impedance, making the influence of resistance more significant. Directly using the ratio of voltage to current to calculate impedance simplifies the calculation process, reduces errors introduced by complex circuit elements, and provides more intuitive information on impedance changes. For fiberglass epoxy boards, observing impedance changes at high frequencies can effectively reflect the uniformity and continuity of the material.

[0082] Based on the calculated impedance data at different frequencies, the relationship between frequency and impedance was plotted in the graphing software Matlab. The peak and inflection point of the relationship were used to identify the identification features, indicating that the material properties change significantly at these frequencies, which is an indication of abrupt changes in the internal structural properties of the material, such as dielectric properties or conductivity.

[0083] The signal-to-noise ratio is calculated based on the identified features and expressed as:

[0084] ;

[0085] SNR represents the calculated signal-to-noise ratio. Represents frequency The impedance when the sample is connected, The impedance representing background noise is calculated. This represents the absolute value of the difference between the sample signal and the noise signal at that frequency. The larger the difference, the more significant the impedance change of the sample at that frequency, and the easier it is to distinguish from the noise. Further dividing by... The calculation differences can be standardized;

[0086] Based on historical experience values ​​as the signal-to-noise ratio (SNR) threshold, select identification features that are greater than or equal to the SNR threshold. 20dB can be selected as the SNR threshold. In impedance measurement and other similar electrical measurements, a SNR of 20dB is sufficient to distinguish meaningful signal variations from random noise. At the same time, in the field of signal processing, especially in applications involving the evaluation of complex material properties, setting 20dB as the lower limit of SNR helps to standardize the test process and make the results comparable across multiple tests and environments.

[0087] Based on the selected characteristic frequency points, the rate of change of impedance at adjacent frequency points is calculated, and a reference step size for adjacent frequency steps is set based on historical experience, expressed as:

[0088] ;

[0089] in This represents the rate of change of the frequency of the k-th identification feature. The impedance represents the k-th identification feature frequency point, and s represents the reference step size, for example, 1kHz;

[0090] Based on the 10th percentile of the initial rate of change as the threshold, the frequency points are adjusted by gradually expanding the baseline step size, as follows:

[0091] ;

[0092] in Representing frequency point The rate of change of impedance, Representing frequency point impedance;

[0093] Compare the rate of change calculated each time with the threshold. When the rate of change is lower than the set threshold, stop adjusting the frequency point and confirm the frequency interval based on the identified characteristic frequency points.

[0094] Integration is performed based on extended frequency intervals, and the overlapping parts of the frequency intervals are integrated to form a continuous frequency range, which is then set as the scanning frequency range.

[0095] By using PZT sensors to perform refined impedance measurements of fiberglass epoxy boards at both low and high frequencies, and combining this with signal-to-noise ratio (SNR) calculation and analysis, high-sensitivity detection of the material's internal structure and defects is achieved. Preliminary impedance calculations at low frequencies reveal minute physical changes in the material and internal non-uniformity reflected by capacitance characteristics and parallel circuits. Furthermore, at high frequencies, the system highlights the influence of resistance and reduces errors in capacitor charging and discharging, thus more accurately reflecting the material's uniformity and continuity and enhancing defect detection accuracy. Through background noise measurement and SNR calculation, the system effectively separates meaningful signals from noise, allowing for the selection of high SNR identification feature points across different frequency ranges. This ensures accurate capture of real changes in material state during monitoring, reduces noise interference, and achieves effective signal-noise separation. By calculating the rate of change of identified characteristic frequency points and setting step size and rate of change thresholds based on historical experience, the system can automatically adjust and expand the frequency scanning range. This ensures that the frequency scanning range covers the most sensitive areas throughout the entire production process. The integrated continuous frequency range not only enhances the detection capability of inherent material defects but also provides a complete and coherent monitoring data chain for subsequent quality traceability. When a material anomaly is detected, the system can quickly trace back to the specific frequency point and its corresponding impedance change, pinpoint the root cause of the problem, and improve the efficiency of problem solving.

[0096] S2, based on the introduced bonding defects in the sample, the impedance is measured, the real part of the impedance is calculated based on the measured voltage and current, and the specific impact of the introduced defects on the impedance characteristics of the glass fiber epoxy board sample is quantified. The defect threshold of RMSD value is calculated based on historical data and the frequency points are marked.

[0097] Preferably, impedance measurements are performed based on introduced bonding defects in the sample. The real part of the impedance is calculated based on the measured voltage and current, and the specific impact of the introduced defects on the impedance characteristics of the glass fiber epoxy board sample is quantified. A defect threshold for the RMSD value is calculated based on historical data, and frequency points are marked.

[0098] Based on the adhesive defects introduced in the fiberglass epoxy board sample, the defects were cut into the fiberglass epoxy board sample using a cutting tool, and the location and size of each defect were accurately marked on the fiberglass epoxy board with a marking pen;

[0099] Impedance measurements were performed using a PZT sensor across a scanning frequency range, providing information on the impedance, location, and magnitude at each measurement frequency point.

[0100] Based on the measurement frequency point k where the adhesion defect is introduced, the voltage is measured using an impedance analyzer. and current The time-domain signal is used to perform Fourier transforms on the voltage and current signals at each frequency point k to obtain the voltage phase. and current phase Through voltage phase and current phase Phase angle calculation by subtraction ;

[0101] At frequency k, based on the measured voltage and current values, the real part of the impedance is calculated and expressed as:

[0102] ;

[0103] in This represents the real part of the impedance at frequency k. This represents the voltage at frequency point k. The current at frequency k;

[0104] The specific impact of defects introduced by quantization on the impedance characteristics of glass fiber epoxy board samples is expressed as follows:

[0105] ;

[0106] in This represents the real part of the impedance measured at the k-th frequency point after the defect is introduced. Let N represent the real part impedance value of the i-th measurement at the k-th frequency point before the introduction of the defect, and N represent the number of measurements at each frequency point. This represents the root mean square error at the k-th frequency point;

[0107] For each group of data before and after the introduction of adhesive defects, RMSD calculation was applied to calculate the RMSD values ​​at all frequency points. The calculation results were compiled into a table, indicating the RMSD values ​​under different defect states.

[0108] The defect threshold is the sum of the average RMSD value and twice the standard deviation of the historical sample data. Frequency points with RMSD values ​​greater than or equal to the defect threshold are marked as significant damage.

[0109] By artificially introducing bonding defects, the system can simulate the damage that fiberglass epoxy boards may experience during actual use. This allows for the collection of impedance data under different damage states. Through RMSD analysis of the impedance measurements at different frequency points, the system can identify minute damage or defects occurring during production. In particular, by comparing with historical data, RMSD analysis can quantify the extent of these damages and provide precise numerical results, helping to further confirm the severity of the damage. By setting defect thresholds based on historical data, defects can be identified in advance before they significantly affect the overall performance of the material, helping to adjust the production process in a timely manner, thereby preventing the expansion of potential defects and avoiding more serious quality problems, achieving the effect of early prevention. By identifying and marking the frequency points of obvious damage, and comparing with historical data, a reasonable defect threshold can be set to improve the sensitivity to potential quality problems. The flexible use of this precise quantification and marking capability allows for high sensitivity to sample quality changes even under imperceptible operating conditions, achieving accurate quality traceability.

[0110] S3. Obtain material parameters and calculate electrical impedance based on the scanning frequency range. Analyze the correlation between RMSD values ​​and electrical impedance based on the marked frequency points. Use the t-distribution for testing, calculate the t-statistic, and screen frequency points with strong linear relationships. Integrate sensitive frequency ranges based on frequency intervals.

[0111] Preferably, material parameters are obtained and electrical impedance is calculated based on the scanning frequency range. The correlation between RMSD values ​​and electrical impedance is analyzed based on marked frequency points. A t-distribution is used for testing, and the t-statistic is calculated to screen frequency points with strong linear relationships. Furthermore, sensitive frequency ranges are integrated based on frequency intervals, including...

[0112] Frequency scanning was performed on a fiberglass epoxy board with a PZT sensor attached, using a PZT sensor and an analog-to-digital converter. The scanning range was based on the scanning frequency range. Material parameters, including dielectric constant, dielectric loss factor, piezoelectric coefficient, and elastic modulus, were obtained, and the electrical impedance was calculated, expressed as:

[0113] ;

[0114] ;

[0115] Where p represents the material density, v represents the speed at which the vibrating sound wave propagates through the material, and A represents the cross-sectional area of ​​the material. Indicates admittance. Representing the relative permittivity, in glass fiber epoxy boards, the permittivity can indicate the material's electrical insulation properties and the uniformity of its internal structure. This represents the dielectric loss factor, which is related to the energy loss caused by polarization within the material. and These represent the material's inherent mechanical resistance and the added resistance, respectively. This represents the piezoelectric coefficient. In glass fiber epoxy boards, the piezoelectric effect can be used to monitor the material's behavior under load. Represents the elastic modulus, reflecting a material's ability to respond to mechanical stress; 'i' represents the imaginary unit. This represents the sensor's scaling factor, which is determined through sensor calibration experiments to obtain the sensor's sensitivity coefficient.

[0116] Among them, the impedance of the computer This represents the material's ability to impede mechanical vibrations, where the calculator impedance... At that time, the application of v helps to determine the speed at which sound waves propagate in a material, and helps to calculate the corresponding mechanical impedance and additional impedance of the material. Including additional impedance due to measuring device, environmental factors, or sample assembly method, the total impedance of the system can be measured and additional impedance confirmed without connecting the sample, and the piezoelectric coefficient is calculated. The efficiency of the interaction between the electric field and mechanical stress is described, obtained through piezoelectric material property testing, and the elastic modulus in the calculation is also included. This describes a material's ability to resist deformation under external forces, which can be obtained through materials mechanics testing; it is also known as the dielectric loss factor. It can be measured using an LCR meter or an impedance analyzer;

[0117] Collect the RMSD values ​​and corresponding admittances of frequency points marked as showing significant damage. And perform correlation calculations, expressed as:

[0118] ;

[0119] Where M represents the total number of frequency points. This represents the admittance at frequency point k. This represents the mean admittance. This represents the mean of RMSD;

[0120] The test is performed using the t-distribution, and the t-statistic is calculated as follows:

[0121] ;

[0122] Based on general significance level Given a value of 0.05 and degrees of freedom M-2, the corresponding critical value of the t-distribution is found by consulting the t-distribution table. Furthermore, the correlation threshold is calculated and expressed as:

[0123] ;

[0124] Where t represents the t-statistic. Indicates the correlation threshold;

[0125] If the calculated value of r is greater than or equal to Then judge and There is a strong linear relationship between them, and it is determined that the corresponding frequency point k is more sensitive to damage;

[0126] If the calculated value of r is less than Then judge and There is no strong linear relationship between them;

[0127] Based on the frequency intervals between all adjacent frequency points with strong linear relationships, and calculating the average value of the frequency intervals, it is expressed as:

[0128] ;

[0129] in Let B represent the average frequency interval, and let B represent the total number of frequency points k with a strong linear relationship. This represents the frequency value of the (i+1)th frequency point k. This represents the frequency value of the i-th frequency point k;

[0130] Will The sensitive range of each strongly linear frequency point is determined by the floating value, and overlapping sensitive ranges are merged to form a sensitive frequency range.

[0131] By monitoring the impedance values ​​of fiberglass epoxy boards at key frequency points in real time, minute changes within the material can be detected instantly, enabling early damage identification. This high-precision monitoring technology improves the overall quality and reliability of the product. Historical data comparison tracks the material's health history, providing strong data support for dynamic traceability and making the production process more transparent and controllable. By calculating and applying correlation thresholds, the system can automatically identify potential quality problems, alerting operators or automatically adjusting production processes, thus improving the reliability and safety of the production process. This helps companies proactively prevent potential problems during production, rather than reacting afterward. Through correlation analysis, the system can automatically screen the most damage-sensitive frequency points and calculate their sensitivity ranges, using these ranges for subsequent monitoring to ensure the system can capture even the most subtle changes in the material's state. By combining impedance measurement, RMSD analysis, and the determination of sensitive frequency ranges, the system provides a complete quality traceability chain. When quality problems occur, the system can trace back to specific frequency points and impedance changes, quickly locating the root cause and improving problem-solving efficiency.

[0132] S4. Construct a random forest model and train it using K-fold cross-validation. Input the model with frequency points based on the sensitive frequency range, determine the prediction threshold through the cumulative distribution function, and then determine the model output.

[0133] Preferably, a random forest model is constructed and trained using K-fold cross-validation. Frequency points within a sensitive frequency range are input into the model, a prediction threshold is determined using the cumulative distribution function, and the model output is then evaluated, including...

[0134] Admittance calculated based on the sensitive frequency range The values ​​are used to construct the model dataset, including... The value and the corresponding status label;

[0135] Construct a random forest model and input different frequency points from the model dataset into the model's decision tree. value;

[0136] The model samples the input data multiple times and constructs multiple decision trees. Each decision tree independently classifies the input data. The model then performs a weighted average of the prediction results from all decision trees and selects the maximum value as the final prediction output based on the probability value output by the model.

[0137] During model training, K-fold cross-validation is used. The dataset is split, with one part used as the validation set and the rest as the training set. This is repeated K times, with different data selected as the validation set each time. After each training session, the model is scored using the validation set, and the model's performance metrics are recorded. The optimal combination of model parameters is selected for final training, and the model parameters are confirmed.

[0138] The model output probabilities are sorted based on the model training process, and the cumulative distribution function value of each probability value is calculated. Through general significance level The prediction threshold is calculated using a value of 0.05, and is expressed as follows:

[0139] ;

[0140] ;

[0141] Where X represents the total number of samples. Let represent the probability value of the i-th sample. It is an indicator function that takes the value 1 if the condition is met, and 0 otherwise. Indicates the prediction threshold;

[0142] In actual monitoring, the system will collect admittance data within the sensitive frequency range in real time. The value is input into the random forest model. The maximum value is selected as the final prediction output based on the probability value output by the model. If the obtained prediction output value is greater than or equal to the prediction threshold, it is judged as quality impairment.

[0143] By monitoring the impedance values ​​within a sensitive frequency range in real time, the model can promptly detect subtle changes in the material and predict the likelihood of damage. When the model output probability value exceeds a predetermined threshold, the system automatically triggers an alarm, alerting operators in advance or automatically adjusting the production process to prevent damage from escalating or developing into more serious problems. This enables damage prediction for fiberglass epoxy board materials. Through the model's analysis and prediction of fiberglass epoxy board materials, the system can also provide timely information on material performance and lifespan. This allows for timely understanding of the actual performance differences between different material ratios and even different batches during the production process, leading to a better understanding of material behavior and lifespan, and optimization of material use and maintenance strategies. Quality prediction based on seamless operation allows for timely detection of material defects and quality traceability, improving product quality and safety, ensuring efficient production processes and high-quality products. Real-time impedance monitoring within a sensitive frequency range enables the model to accurately track every change in the material's state, achieving continuous quality control during seamless operation. When the model output probability exceeds a set threshold, the system immediately issues an alarm, allowing for immediate response to prevent the spread of quality defects, ensuring the quality and consistency of the production process, and achieving full-process quality control and traceability from raw materials to finished products.

[0144] S5 determines the remote early warning based on the model's output, sends early warning information and alarms through multiple terminals, stores data in a distributed manner, transmits encrypted data in real time, defines access permissions, and backs up log data using DES encryption.

[0145] Preferably, based on the model's output, remote early warning is issued and early warning information is sent through multiple terminals. The judgment results of the random forest model are used to conduct remote monitoring and early warning based on the Internet of Things. If a signal of quality damage is detected, an early warning message is sent to the operator via SMS gateway and email via SMTP server. The specific early warning information is displayed on the monitoring interface, and the physical alarm device on site is directly triggered through the integrated PLC industrial controller.

[0146] By integrating early warning functions, the system can notify relevant personnel through multiple channels the instant a potential risk occurs, ensuring the timeliness and accuracy of information transmission. These channels include SMS and email, on-site early warning reminders, and real-time warning information. The use of multiple early warning methods and the synchronous triggering of on-site alarm devices enable operators to take swift action to prevent accidents or further material damage. This system not only improves the safety of operating procedures but also provides strong support for achieving quality traceability under seamless operation, ensuring the stability and reliability of industrial production processes.

[0147] Furthermore, data is stored in a distributed manner, and real-time encrypted data transmission is implemented. Access permissions are defined, and backup log data is backed up using DES encryption, including...

[0148] The system employs a distributed storage approach to securely store the collected and computed data in real time. It uses the end-to-end encryption protocol TLS to encrypt the collected and computed data and transmit them to the storage device. Simultaneously, it distributes the data in a multi-node cloud database and performs replication and backup of the multi-node data in the cloud database.

[0149] Role-based access control (RBAC) defines access permissions, clarifies data access permissions in the cloud database, records all access and operation data, records the access and operation data in log files, encrypts the log data using DES, and periodically backs it up to off-site storage.

[0150] By adopting distributed storage, the system effectively improves data security and reliability. The combination of encrypted transmission and distributed storage ensures that data remains intact and secure even during transmission or when storage nodes are attacked. Role-based access control and encrypted log mechanisms ensure the security and transparency of data access, providing a reliable basis for subsequent auditing and traceability. Furthermore, the off-site storage of encrypted logs further enhances the security of personnel access control. Off-site storage of log files improves security, facilitates access to and traceability of personnel operation records, and enhances traceability capabilities.

[0151] Example 2

[0152] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0153] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0154] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0155] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0156] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method of non-inductive operation quality traceability in the glass fiber industry, characterized by: comprising, selecting a PZT sensor for installation, cooperating with an analog-to-digital converter to obtain collected data, calculating impedance values in a low-frequency interval and a high-frequency interval respectively and identifying features, calculating a signal-to-noise ratio to confirm the identified feature frequency points, analyzing the change rate of adjacent identified feature frequency points, determining the frequency interval and integrating the scanning frequency range, comprising, based on the scenario that the PZT sensor is not connected to the glass epoxy plate sample, measuring the background noise through the PZT sensor and the analog-to-digital converter, and recording the noise level in the range of 1kHz to 1MHz; based on the scenario that the PZT sensor is connected to the glass epoxy plate sample, measuring through the PZT sensor and the analog-to-digital converter located in the low-frequency interval and the high-frequency interval respectively, recording the impedance value corresponding to each frequency point, the impedance calculation in the low-frequency interval is expressed as: ； wherein Z(ω) represents the impedance of the low frequency range frequency ω, R represents the resistance, C represents the capacitance, Z(ω) represents the impedance of the low frequency range frequency ω, R represents the resistance, C represents the capacitance, the impedance in the high-frequency interval is calculated by the voltage of the frequency ω divided by the current of the frequency ω; based on the calculated impedance data at different frequencies, drawing a graph of frequency and impedance in the chart software Matlab, and determining the identified features according to the peak value and the inflection point of the graph; based on the signal-to-noise ratio calculated based on the identified features, the historical experience value is used as the signal-to-noise ratio threshold, and the identified features greater than or equal to the signal-to-noise ratio threshold are selected; based on the selected identified feature frequency points, the change rate of the impedance of the adjacent frequency points corresponding to the impedance is calculated, and the reference step of the adjacent frequency step is set based on the historical experience, the 10th percentile of the initial change rate is used as the change threshold, and the frequency points are gradually expanded by adjusting the reference step, expressed as: ； in Representing frequency point The rate of change of impedance, Representing frequency point impedance, Representing frequency point impedance; compare the change rate calculated each time with the threshold value, when the change rate is lower than the set change threshold, stop adjusting the frequency point, and confirm the frequency interval according to the identified feature frequency point; based on the expanded frequency interval, integrate the frequency interval to form a continuous frequency range, and set it as the scanning frequency range; based on the sample introducing adhesive defects for impedance measurement, calculating the real part of the impedance based on the measured voltage and current, and quantifying the specific influence of the introduced defects on the impedance characteristics of the glass epoxy plate sample, calculating the defect threshold value based on historical data RMSD value and marking the frequency point; according to the scanning frequency range, obtain the material parameters and calculate the electrical impedance, analyze the correlation of the RMSD value of the electrical impedance according to the marked frequency point, use t distribution for testing, calculate t statistics and screen the frequency points with strong linear relationship, and integrate the sensitive frequency range based on the frequency interval, comprising, based on the PZT sensor and the analog-to-digital converter for frequency scanning of the glass epoxy plate pasted on the PZT sensor, the scanning range is based on the scanning frequency range, the material parameters including dielectric constant, dielectric loss factor, piezoelectric coefficient and elastic modulus are obtained, and the electrical impedance is calculated, expressed as: ； ； where p represents the material density, v represents the speed at which a sound wave propagates through the material, and A represents the cross-sectional area of the material, represents the admittance, represents the relative permittivity, represents the dielectric loss factor, and respectively represent the mechanical impedance of the material itself and the additional impedance, represents the piezoelectric coefficient, represents the elastic modulus, and i represents the imaginary unit, represents the sensor scaling factor; Collect the RMSD values of the frequency points marked as obvious damage and the corresponding admittance , and the correlation is calculated by Pearson correlation coefficient, the t distribution is used for test, and the t statistic is calculated, and the corresponding t distribution critical value is found by consulting the t distribution table , and further calculate the correlation threshold ; If the calculated r value is greater than or equal to , it is determined that and have a strong linear relationship, and the corresponding frequency point k is more sensitive to damage, wherein is the admittance of the frequency point , and is the RMSD value of the frequency point . based on the frequency intervals between the adjacent frequency points of all strong linear relationships, and calculating the average value of the frequency intervals , the sensitive range of each strong linear relationship frequency point is determined as a floating value, and the overlapping sensitive ranges are merged to form a sensitive frequency range construct a random forest model and train it through K-fold cross-validation, input the frequency points in the sensitive frequency range into the model, determine the prediction threshold through the cumulative distribution function, and judge the model output; according to the output judgment of the model, carry out remote early warning, send warning information through multiple terminals, and carry out warning alarm, store the data in a distributed manner, and transmit the real-time data in an encrypted manner, define access permissions and backup log data based on DES encryption.

2. A method of non-inductive handling quality traceability in the fiberglass industry as claimed in claim 1, characterized in that: The PZT sensor is selected and installed, and an analog-to-digital converter is used to collect data, including, The PZT sensor is selected and installed, the selected bonding position of the glass fiber epoxy plate is cleaned, the cut PZT sensor is bonded to the selected bonding position using a high-performance epoxy adhesive, and the adhesive is allowed to fully cure and complete bonding for 24 hours; The output end of the PZT sensor is electrically connected to the input end of the analog-to-digital converter, and the output end of the analog-to-digital converter is electrically connected to the input end of the computer through a USB interface; The computer driver is configured and the output data of the analog-to-digital converter is read.

3. A method of non-inductive handling of quality traceability in the fiberglass industry as claimed in claim 2, characterized in that: The impedance measurement is based on the introduction of adhesive defects in the sample, the real part of the impedance is calculated based on the measured voltage and current, and the specific effect of the introduced defects on the impedance characteristics of the glass fiber epoxy plate sample is quantified. The defect threshold value is calculated based on the historical data RMSD value and the frequency point is marked, including, Based on the introduction of adhesive defects in the glass fiber epoxy plate sample, a cutting tool is used to cut defects on the glass fiber epoxy plate sample, and a marker pen is used to accurately mark the position and size of each defect on the glass fiber epoxy plate; The PZT sensor is used to perform impedance measurement according to the scanning frequency range, and the impedance, position and size information for each measurement frequency point are obtained; Based on the measurement frequency point k where the adhesion defect is introduced, the voltage is measured using an impedance analyzer. and current The time-domain signal is used to perform Fourier transforms on the voltage and current signals at each frequency point k to obtain the voltage phase. and current phase Through voltage phase and current phase Phase angle calculation by subtraction ; At the frequency point k, according to the measured voltage and current values, the real part of the impedance is calculated, the specific influence of the introduced defects on the impedance characteristics of the glass fiber epoxy plate sample is quantified and the values are calculated; RMSD calculation is applied to the data before and after each group of introduced adhesive defects, and the RMSD values of all frequency points are calculated. The calculation results are arranged in a table, and the RMSD values under different defect states are marked. The sum of the average value and twice the standard deviation of the RMSD values based on the historical data of the sample is used as the defect threshold value. The frequency points with RMSD values greater than or equal to the defect threshold value are marked as obvious damage.

4. A method of non-inductive handling of quality traceability in the fiberglass industry as claimed in claim 3, characterized in that: The random forest model is constructed and trained through K-fold cross-validation, the frequency points in the sensitive frequency range are input into the model, the prediction threshold is determined through the cumulative distribution function, and the model output is determined, including, Model data sets are constructed based on the values of admittance calculated in the sensitive frequency range , including the values of admittance and the corresponding status labels. constructing a random forest model and inputting the values of different frequency points in the model data set into the decision tree of the model values; The model inputs data multiple times and constructs multiple decision trees, each decision tree independently classifies the input data, the model performs weighted average on the prediction results of all decision trees, and the maximum value of the probability value output by the model is selected as the final prediction output; During model training, K-fold cross-validation is used for training, the data set is divided, and 1 part is used as the validation set and the remaining part is used as the training set. Repeat K times, and each time select different data as the validation set. After each training, use the validation set to score the model, record the performance indicators of the model, select the best model parameter combination for final training, and confirm the model parameters; ranking model output probabilities during a model training process, computing a cumulative distribution function value for each probability value and determining a prediction threshold; In actual monitoring, the system will collect the value of admittance in the sensitive frequency range in real time Input into the random forest model, select the maximum value as the final prediction output according to the probability value output by the model, and if the prediction output value is greater than or equal to the prediction threshold, it is judged as quality damage.

5. A method of non-inductive handling of quality traceability in the fiberglass industry as claimed in claim 4, characterized in that: According to the output of the model, remote early warning is performed, and warning information is sent through multiple terminals and warning alarm is performed. Through the determination result of the random forest model, remote monitoring and early warning are performed based on IoT. If a quality damage signal is detected, an SMS message is sent through an SMS gateway, an email is sent to an operator through an SMTP server, and specific early warning information is displayed on a monitoring interface. The integrated PLC industrial controller directly triggers the physical alarm device on site.

6. A method of non-inductive handling of quality traceability in the fiberglass industry as claimed in claim 5, characterized in that: The data is stored in a distributed manner, real-time data encryption transmission is performed, access permissions are defined, and DES encryption is used to backup log data, including, The distributed storage method is adopted to store the collected data and the calculated data in real time, the collected data and the calculated data are encrypted through an end-to-end encryption protocol TLS, and are transmitted to the storage while being stored in the cloud database of multiple nodes, the data of the multiple nodes in the cloud database are copied and backed up; The access permission is defined based on a role-based access control (RBAC) mechanism, the data access permission in the cloud database is explicitly defined, all access and operation record data are recorded, the access and operation record data are recorded in a log file, the log data are processed through DES encryption, and are periodically backed up to a remote storage.

7. A computer device comprising: A memory and a processor; The memory stores a computer program, and the processor executes the computer program to implement the steps of the glass fiber industry non-sensing operation quality traceability method in any one of claims 1 to 6.

8. A computer readable storage medium having stored thereon a computer program, characterized in that: The computer program is executed by the processor to implement the steps of the glass fiber industry non-sensing operation quality traceability method in any one of claims 1 to 6.

Images