Non-inductive operation quality tracing method in glass fiber industry

Abstract

The invention discloses a non-inductive operation quality tracing method in the glass fiber industry, which relates to the technical field of glass fiber quality tracing and comprises the following steps of: selecting a PZT sensor for installation, acquiring acquired data in cooperation with an analog-to-digital converter, respectively calculating impedance values in a low-frequency interval and a high-frequency interval and identifying characteristics, calculating a signal-to-noise ratio to confirm and identify characteristic frequency points, and analyzing the change rate of the adjacent identification characteristic frequency points. According to the method disclosed by the invention, the impedance of the glass fiber epoxy plate is finely measured at low frequency and high frequency, and calculation and analysis of a signal-to-noise ratio are combined, so that high-sensitivity detection of the internal structure and defects of the material is achieved, and possible damage of the glass fiber epoxy plate in the actual use process can be simulated by artificially introducing bonding defects, so that the detection accuracy is improved. And impedance data of the material in different damage states can be collected, and tiny changes in the material can be detected in real time by monitoring the electrical impedance value of the glass fiber epoxy plate at the key frequency point in real time, so that early damage identification is realized.

Description

Technical Field

[0001] The invention relates to the technical field of glass fiber quality tracing, in particular to a non-sensing operation quality tracing method in the glass fiber industry. Background Art

[0002] Glass fiber materials are widely used in aerospace, automobile manufacturing, construction engineering and other fields due to their excellent mechanical properties, heat resistance and electrical insulation. Glass fiber 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 performance of epoxy resin, and is a common glass fiber material.

[0003] However, during the production process of glass fiber materials, subtle defects such as microcracks, bubbles or poor bonding often occur. The existence of these defects may have a significant impact on the performance and safety of the final product. Although traditional quality control methods such as visual inspection, ultrasonic inspection and X-ray inspection can detect macro defects in materials to a certain extent, there are still limitations in the detection and real-time monitoring of micro defects. It is inconvenient to conduct effective analysis, monitoring and traceability of defects inside the material, which not only affects the overall performance of the material, but also reduces the efficiency of quality tracing. Summary of the invention

[0004] In view of the problems existing in the above-mentioned existing non-sensing operation quality tracing method in the glass fiber industry, the present invention is proposed.

[0005] Therefore, the problem to be solved by the present invention is that there are still limitations in the detection and real-time monitoring of microscopic defects, and it is inconvenient to conduct effective analysis, monitoring and tracing of defects inside the material, which not only affects the overall performance of the material, but also reduces the efficiency of quality tracing.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: a non-sensing operation quality tracing method for the glass fiber industry, comprising:

[0007] Select PZT sensor for installation, use analog-to-digital converter to obtain collected data, calculate impedance value and identify features in low-frequency and high-frequency ranges, calculate signal-to-noise ratio to confirm and identify feature frequency points, analyze the change rate of adjacent identified feature frequency points, determine frequency interval and integrate scanning frequency range;

[0008] Impedance measurement is performed based on the introduction of bonding defects into 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. The defect threshold of the RMSD value is calculated based on historical data and the frequency points are marked;

[0009] Obtain material parameters and calculate electrical impedance according to the scanning frequency range, analyze the correlation of RMSD value and electrical impedance according to the marked frequency points, use t distribution for testing, calculate t statistics and screen frequency points with strong linear relationship, and integrate sensitive frequency range based on frequency interval;

[0010] A random forest model was constructed and trained through K-fold cross validation. The frequency points in the sensitive frequency range were input into the model, the prediction threshold was determined through the cumulative distribution function, and the model output was determined.

[0011] Based on the output of the model, remote early warning is carried out and early warning information is sent through multiple terminals and early warning alarms are issued. Data is stored in a distributed manner and encrypted in real time. Access rights are defined and log data is backed up based on DES encryption.

[0012] As a preferred solution of the non-sensing operation quality tracing method of the glass fiber industry of the present invention, wherein: the PZT sensor is selected for installation and matched with the analog-to-digital converter to obtain the collected data, including:

[0013] Select the PZT sensor for installation, clean the selected pasting position of the glass fiber epoxy board, use high-performance epoxy resin adhesive to paste the cut PZT sensor to the selected pasting position, and let it stand for 24 hours until the adhesive is completely cured and pasting is completed;

[0014] 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;

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

[0016] As a preferred solution of the non-sensing operation quality tracing method for the glass fiber industry of the present invention, wherein: the impedance value is calculated and the characteristics are identified in the low-frequency interval and the high-frequency interval respectively, the signal-to-noise ratio is calculated to confirm the identification of the characteristic frequency point, the change rate of the adjacent identification characteristic frequency points is analyzed, the frequency interval is determined and the scanning frequency range is integrated, including,

[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 the PZT sensor and the analog-to-digital converter, and the noise level in the range of 1kHz to 1MHz is recorded;

[0018] Based on the scenario where the PZT sensor is connected to the glass fiber epoxy board sample, the PZT sensor and the analog-to-digital converter are respectively located in the low-frequency range and the high-frequency range for measurement, and the impedance value corresponding to each frequency point is recorded. The impedance calculation in the low-frequency range is expressed as:

[0019]

[0020] Where Z l (ω) represents the impedance of the low-frequency range frequency ω, R represents resistance, C represents capacitance, R P represents the parallel resistance, 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, a frequency-impedance relationship graph was plotted in the chart software Matlab, and the identification features were determined based on the peak and inflection points of the relationship graph;

[0023] Calculate the signal-to-noise ratio based on the recognition feature, use the historical experience value as the signal-to-noise ratio threshold, and select the recognition feature that is greater than or equal to the signal-to-noise ratio threshold;

[0024] Based on the selected identification characteristic frequency point, the change rate of the corresponding impedance adjacent frequency point is calculated, and the reference step size of the adjacent frequency step size is set based on the history. The 10th percentile of the initial change rate is used as the change threshold, and the reference step size is gradually expanded to adjust the frequency point, which is expressed as:

[0025]

[0026] Where R(ω k ±n) represents the frequency point ω k +n impedance change rate, Z(ω k ±n) represents the frequency point ω k +n impedance;

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

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

[0029] As a preferred solution of the non-sensing operation quality tracing method for the glass fiber industry of the present invention, wherein: the impedance measurement is performed based on the introduction of 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, and the defect threshold of the RMSD value is calculated based on historical data and the frequency point is marked, including,

[0030] Introduce bonding defects based on the glass fiber epoxy board sample, use a cutting tool to cut the defects on the glass fiber epoxy board sample, and use a marking pen to accurately mark the position and size of each defect on the glass fiber epoxy board;

[0031] Use a PZT sensor to measure impedance over a scanning frequency range, and obtain information about the impedance, position, and size of each measurement frequency point;

[0032] The voltage V is measured using an impedance analyzer at the measurement frequency point k where the bonding defect is introduced. k (t) and current I k The voltage and current signals at each frequency point k are Fourier transformed to obtain the voltage phase and current phase By voltage phase and current phase Subtract to calculate the phase angle θ k ;

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

[0034] Apply RMSD calculation to the data before and after each group of experiments with the introduction of bonding defects, calculate the RMSD values ​​of all frequency points, and organize the calculation results into a table, indicating the RMSD values ​​under different defect states;

[0035] The average value and the sum of two times the standard deviation of the RMSD value based on the sample historical data are used as the defect threshold, and the frequency points with RMSD values ​​greater than or equal to the defect threshold are marked as obvious damage.

[0036] As a preferred solution of the non-sensing operation quality tracing method of the glass fiber industry of the present invention, wherein: the material parameters are obtained and the electrical impedance is calculated according to the scanning frequency range, the correlation of the RMSD value and the electrical impedance is analyzed according to the marked frequency points, the t distribution is used for testing, and the t statistic is calculated and the frequency points with strong linear relationships are screened, and the sensitive frequency range is integrated based on the frequency interval, including,

[0037] Based on the PZT sensor and the analog-to-digital converter, the glass fiber epoxy board to which the PZT sensor is attached is subjected to a frequency scan. The scanning range is based on the scanning frequency range. The material parameters including the dielectric constant, dielectric loss factor, piezoelectric coefficient and elastic modulus are obtained, and the electrical impedance is calculated, which is expressed as:

[0038] Z s (ω) = pvA;

[0039]

[0040] Where p represents the material density, v represents the speed at which the vibration sound wave propagates through the material, A represents the cross-sectional area of ​​the material, and Y(ω) represents the electrical impedance. represents the relative dielectric constant, δ represents the dielectric loss factor, and Z represents the energy loss caused by the polarization inside the material. s (ω) and Z α (ω) represent the mechanical impedance of the material itself and the additional impedance, d 3x represents the piezoelectric coefficient, represents the elastic modulus, i represents the imaginary unit, and α represents the sensor scale factor;

[0041] Collect the RMSD values ​​of the frequency points marked as obvious damage and the corresponding electrical impedance Y(ω), calculate the correlation through the Pearson correlation coefficient, use the t distribution for testing, calculate the t statistic, and find the corresponding t distribution critical value t by consulting the t distribution table. c , and further calculate the correlation threshold r c ;

[0042] If the calculated r value is greater than or equal to r c , then judge Y(ω k ) and RMSD k 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 Δω of the frequency intervals is calculated, Δω is used as a floating value to determine the sensitive range of each frequency point with strong linear relationships, and the overlapping sensitive ranges are merged to form a sensitive frequency range.

[0044] As a preferred solution of the non-sensing operation quality tracing method for the glass fiber industry of the present invention, wherein: the random forest model is constructed and trained through K-fold cross validation, the frequency point input model is based on the sensitive frequency range, the prediction threshold is determined by the cumulative distribution function, and the model output is determined, including:

[0045] Based on the electrical impedance Y(ω calculated within the sensitive frequency range k ) values, construct the model data set, including Y(ω k ) and the corresponding status label;

[0046] Construct a random forest model and input Y(ω) at different frequency points in the model dataset into the model's decision tree. k )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 takes a weighted average of the prediction results of all decision trees and selects the maximum value as the final prediction output based on the probability value output by the model.

[0048] K-fold cross validation is used during model training. The data set is split and one part is used as the validation set, and the rest is used as the training set. This is repeated K times, and different data is selected as the validation set each time. After each training, the model is scored with the validation set, the performance indicators of the model are recorded, and the best model parameter combination is selected for final training and the model parameters are confirmed.

[0049] Sort the model output probabilities during model training, calculate the cumulative distribution function value F(P) of each probability value and determine the prediction threshold;

[0050] In actual monitoring, the system will collect the electrical impedance Y(ω k ) is input into the random forest model, and the maximum value is selected as the final prediction output according to 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 damage.

[0051] As a preferred solution of the non-sensing operation quality tracing method for the glass fiber industry described in the present invention, wherein: the remote early warning and sending of early warning information through multiple terminals and early warning alarm based on the output judgment of the model refer to remote monitoring and early warning based on the judgment result of the random forest model and the IoT Internet of Things. If a signal of quality damage is detected, a text message is sent through the SMS gateway, and an email is sent through the SMTP server to the operator for early warning information, and 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.

[0052] As a preferred solution of the non-sensing operation quality tracing method of the glass fiber industry of the present invention, wherein: the distributed storage of data and real-time data encryption transmission, definition of access rights and backup of log data based on DES encryption include:

[0053] The distributed storage method is used to store the collected data and the calculated data in real time and securely. The collected data and the calculated data are encrypted through the end-to-end encryption protocol TLS. When the collected data and the calculated data are transmitted to the storage device, they are distributedly stored in a multi-node cloud database, and the multi-node data in the cloud database is replicated and backed up.

[0054] The role-based access control (RBAC) mechanism defines access rights, clarifies data access rights in the cloud database, records all access and operation record data, and records the access and operation record data in log files. The log data is processed through DES encryption and periodically backed up to off-site storage.

[0055] A computer device comprises: a memory and a processor; the memory stores a computer program, and the processor implements the steps of the non-sensing operation quality tracing method in the glass fiber industry when executing the computer program.

[0056] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the non-sensing operation quality tracing method in the glass fiber industry.

[0057] The beneficial effects of the present invention are as follows: by finely measuring the impedance of the glass fiber epoxy board at low frequency and high frequency, and combining the calculation and analysis of the signal-to-noise ratio, high-sensitivity detection of the internal structure and defects of the material can be achieved; by artificially introducing bonding defects, the damage that may occur to the glass fiber epoxy board during actual use can be simulated, and the impedance data of the material in different damage states can be collected; by real-time monitoring of the electrical impedance value of the glass fiber epoxy board at key frequency points, the slight changes in the material can be detected immediately, thereby realizing early damage identification, realizing high-precision monitoring technology and improving the overall quality and reliability of the product; by real-time monitoring of the electrical impedance value within the sensitive frequency range, the slight changes in the material can be discovered in time, and the possibility of damage can be predicted. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0059] Figure 1 The figure is a flow chart of the non-contact quality traceability method in the glass fiber industry.

[0060] Figure 2 Schematic diagram of the analytical correlation flow of the non-contact operation quality traceability method in the glass fiber industry. DETAILED DESCRIPTION

[0061] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below in conjunction with the accompanying drawings.

[0062] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0063] Secondly, the term "one 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 term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor is it a separate or selective embodiment that is mutually exclusive with other embodiments.

[0064] Example 1, reference Figure 1 and Figure 2 , which is the first embodiment of the present invention, and which provides a non-sensing operation quality tracing method for the glass fiber industry. The non-sensing operation quality tracing method for the glass fiber industry includes:

[0065] S1, select the PZT sensor for installation, cooperate with the analog-to-digital converter to obtain the collected data, calculate the impedance value and identify the characteristics in the low-frequency range and high-frequency range respectively, calculate the signal-to-noise ratio to confirm the identification of the characteristic frequency point, analyze the change rate of adjacent identification characteristic frequency points, determine the frequency interval and integrate the scanning frequency range;

[0066] Preferably, a PZT sensor is selected for installation and matched with an analog-to-digital converter to obtain collected data, including:

[0067] Choose PZT sensor, you can choose model 5A4E PZT sensor for installation;

[0068] Clean the selected pasting position of the glass fiber epoxy board, use high-performance epoxy resin adhesive to paste the cut PZT sensor on the selected pasting position, and let it stand for 24 hours until the adhesive is completely cured and pasting is completed;

[0069] 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 AD5933 analog-to-digital converter can be used;

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

[0071] By firmly adhering the cut PZT sensor to the glass fiber epoxy board with high-performance epoxy resin adhesive, the close combination between the sensor and the material is ensured. The rational use of the sensor can accurately capture the tiny electrical impedance changes inside the material, providing a basis for the system's data acquisition. By electrically connecting the PZT sensor to the AD5933 analog-to-digital converter and the computer, efficient real-time data acquisition and processing can be achieved. Through computer-driven configuration and data reading, the system can automatically collect and analyze data. The measurement system combining PZT sensors with analog-to-digital converters can provide an efficient, accurate, and non-sensitive quality monitoring solution for the glass fiber industry.

[0072] Furthermore, the impedance values ​​are calculated and the features are identified in the low-frequency interval and the high-frequency interval, the signal-to-noise ratio is calculated to confirm the identification of the characteristic frequency points, the change rate of the adjacent identification characteristic frequency points is analyzed, the frequency interval is determined and the scanning frequency range is integrated, 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 the PZT sensor and the analog-to-digital converter, and the noise level in the range of 1kHz to 1MHz is recorded;

[0074] Based on the scenario where the PZT sensor is connected to the glass fiber epoxy board sample, the PZT sensor and the analog-to-digital converter are respectively located in the low frequency range (1kHz-10kHz) and the high frequency range (10kHz-1MHz) for measurement, and the impedance value corresponding to each frequency point is recorded. The impedance calculation in the low frequency range is expressed as:

[0075]

[0076] Where Z l (ω) represents the impedance of the low-frequency range frequency ω, R represents resistance, C represents capacitance, R P represents the parallel resistance, i represents the imaginary unit;

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

[0078]

[0079] Where Z h (ω) represents the impedance at the high frequency interval frequency ω, V(ω) represents the voltage at the frequency ω, and I(ω) represents the current at the frequency ω;

[0080] At low frequencies, for glass fiber epoxy boards, the capacitive properties of the internal structure, such as the dielectric constant and the thickness of the dielectric layer, can significantly affect the impedance readings. Moreover, due to the slower current flow, the capacitance charge and discharge cycle has a greater impact on the measurement results. P The composite circuit formed by Z and capacitor C reflects the internal heterogeneity and potential defects of the material, so Z l The calculation of (ω) is advantageous for detecting small physical changes such as cracks or voids;

[0081] At high frequencies, as the frequency increases, the influence of capacitor components decreases, and resistance becomes the main influencing factor. The current changes faster, and the charging and discharging of the capacitor has less influence on the impedance, making the influence of the resistance more significant. Directly using the ratio of voltage and current to calculate the impedance simplifies the calculation process, reduces the error introduced by complex circuit elements, and provides more intuitive impedance change information. For glass fiber 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, a frequency-impedance relationship graph was plotted in the charting software Matlab, and identification features were determined based on the peaks and inflection points of the graph, indicating that the material properties changed significantly at these frequencies, which is an indication of a sudden change 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 recognition features and is expressed as:

[0084]

[0085] Where SNR represents the calculated signal-to-noise ratio, Z s (ω) represents the impedance when the sample is connected at frequency ω, Z n (ω) represents the impedance of the background noise, which can be calculated by |Z s (ω)-Z n (ω) represents the absolute value of the difference between the sample signal and the noise signal at this frequency. The larger the difference, the more significant the impedance change of the sample at this frequency, and the easier it is to distinguish from the noise. At the same time, further divided by Z n (ω) can normalize the calculated differences;

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

[0087] Based on the selected identified characteristic frequency point, the change rate of the corresponding impedance at the adjacent frequency point is calculated, and the reference step size of the adjacent frequency step size is set based on the history, which is expressed as:

[0088]

[0089] Where R(ω k ) represents the rate of change of the kth identified characteristic frequency point, Z(ω k ) represents the impedance of the kth identified characteristic frequency point, s represents the reference step size, for example 1kHz;

[0090] Based on the 10th percentile of the initial change rate as the change threshold, the benchmark step size is gradually expanded to adjust the frequency point, which is expressed as:

[0091]

[0092] Where R(ω k±n) represents the frequency point ω k +n impedance change rate, Z(ω k ±n) represents the frequency point ω k +n impedance;

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

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

[0095] By using PZT sensors to perform refined measurements of the impedance of glass fiber epoxy boards at low and high frequencies, combined with the calculation and analysis of the signal-to-noise ratio, high-sensitivity detection of the internal structure and defects of the material can be achieved. Through preliminary impedance calculation at low frequencies, the impedance measurement can reflect the slight physical changes of the material, as well as the internal non-uniformity of the material reflected by the capacitance characteristics and parallel circuits. In addition, in the high-frequency range, the resistance effect can be highlighted, and the error of capacitor charging and discharging can be reduced, thereby more accurately reflecting the uniformity and continuity of the material and enhancing the defect detection accuracy. Through background noise measurement and signal-to-noise ratio calculation, the system can effectively separate meaningful signals and noise, allowing the selection of identification feature points with high signal-to-noise ratios in different frequency ranges. , ensuring that the real changes in the material state can be accurately captured during the monitoring process, reducing noise interference and achieving effective signal-noise separation. By calculating the change rate of the identified characteristic frequency points and setting the step size and change rate threshold based on historical experience, the system can automatically adjust and expand the frequency scanning range to ensure that the frequency scanning range covers the most sensitive areas throughout the production process. The integrated continuous frequency range not only enhances the ability to detect inherent defects in materials, but also provides a complete and coherent monitoring data chain for subsequent quality traceability. When material abnormalities are detected, the system can quickly trace back to the specific frequency point and its corresponding impedance change, locate the root cause of the problem, and improve the efficiency of problem solving.

[0096] S2, based on the introduction of bonding defects in the sample to perform impedance measurement, 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 the RMSD value is calculated based on historical data and the frequency points are marked;

[0097] Preferably, impedance measurement is performed based on the introduction of bonding defects into 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, and the defect threshold of the RMSD value is calculated based on historical data and the frequency point is marked, including,

[0098] Introduce bonding defects based on the glass fiber epoxy board sample, use a cutting tool to cut the defects on the glass fiber epoxy board sample, and use a marking pen to accurately mark the position and size of each defect on the glass fiber epoxy board;

[0099] Use a PZT sensor to measure impedance over a scanning frequency range, and obtain information about the impedance, position, and size of each measurement frequency point;

[0100] The voltage V is measured using an impedance analyzer at the measurement frequency point k where the bonding defect is introduced. k (t) and current I k The voltage and current signals at each frequency point k are Fourier transformed to obtain the voltage phase and current phase By voltage phase and current phase Subtract to calculate the phase angle θ k ;

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

[0102]

[0103] Where Re(Z k ) represents the real impedance at frequency point k, represents the voltage at frequency point k, Represents the current at frequency point k;

[0104] The specific effect of the introduced defects on the impedance characteristics of the glass fiber epoxy board sample is quantified as follows:

[0105]

[0106] Where Re(Z k,i ) po Re(Z k,i ) pr represents the real impedance value of the i-th measurement at the k-th frequency point before the defect is introduced, N represents the number of measurements at each frequency point, and RMSD k Represents the root mean square error value of the kth frequency point;

[0107] Apply RMSD calculation to the data before and after each group of experiments with the introduction of bonding defects, calculate the RMSD values ​​of all frequency points, and organize the calculation results into a table, indicating the RMSD values ​​under different defect states;

[0108] The average value and the sum of two times the standard deviation of the RMSD value based on the sample historical data are used as the defect threshold, and the frequency points with RMSD values ​​greater than or equal to the defect threshold are marked as obvious damage.

[0109] By artificially introducing bonding defects, the damage that may occur to the glass fiber epoxy board during actual use can be simulated, and the impedance data of the material in different damage states can be collected. By performing RMSD analysis on the electrical impedance measurement results at different frequency points, the system can identify the slight damage or defects that occur to the material during the production process. In particular, by comparing with historical data, RMSD analysis can quantify the extent of these damages and provide accurate numerical results to help further confirm the severity of the damage. By setting defect thresholds based on historical data, defects can be identified in advance before they have significantly affected the overall performance of the material, helping to adjust the production process in time, 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 by comparing with historical data, reasonable defect thresholds can be set to achieve the effect of increasing sensitivity to potential quality problems. Flexible use of this precise quantification and marking capability can still maintain a high sensitivity to quality changes of samples under non-sensing operating conditions, thereby achieving accurate quality traceability.

[0110] S3, obtain material parameters and calculate electrical impedance according to the scanning frequency range, analyze the correlation of RMSD value and electrical impedance according to the marked frequency points, use t distribution for testing, calculate t statistic and screen frequency points with strong linear relationship, and integrate sensitive frequency range based on frequency interval;

[0111] Preferably, material parameters are obtained and electrical impedance is calculated according to the scanning frequency range, the correlation of RMSD value and electrical impedance is analyzed according to the marked frequency points, a t distribution is used for testing, and t statistics are calculated and frequency points with strong linear relationships are screened, and the sensitive frequency range is integrated based on the frequency interval, including,

[0112] Based on the PZT sensor and the analog-to-digital converter, the glass fiber epoxy board to which the PZT sensor is attached is subjected to a frequency scan. The scanning range is based on the scanning frequency range. The material parameters including the dielectric constant, dielectric loss factor, piezoelectric coefficient and elastic modulus are obtained, and the electrical impedance is calculated, which is expressed as:

[0113] Z s (ω) = pvA;

[0114]

[0115] Where p represents the material density, v represents the speed at which the vibration sound wave propagates through the material, A represents the cross-sectional area of ​​the material, and Y(ω) represents the electrical impedance. Represents the relative dielectric constant. In glass fiber epoxy board, the dielectric constant can indicate the electrical insulation performance and internal structural uniformity of the material. δ represents the dielectric loss factor, which is related to the energy loss caused by polarization inside the material. Z s (ω) and Z α (ω) represent the mechanical impedance of the material itself and the additional impedance, d 3x Represents the piezoelectric coefficient. In glass fiber epoxy boards, the piezoelectric effect can be used to monitor the behavior of materials under load. It represents the elastic modulus, which reflects the material's ability to respond to mechanical stress, i represents the imaginary unit, and α represents the sensor scale factor. The sensitivity coefficient of the sensor is determined by the sensor calibration experiment;

[0116] Where the calculated machine impedance Z s (ω) represents the material's ability to block mechanical vibration, where the impedance Z s (ω), the application of v helps to determine the speed of sound wave propagation in the material and helps to calculate the corresponding mechanical impedance of the material, the additional impedance Z α (ω) includes additional impedance due to the measurement device, environmental factors, or sample assembly method. The total impedance of the system can be measured without connecting the sample and the additional impedance can be confirmed. The piezoelectric coefficient d in the calculation 3x Describes the interaction efficiency between electric field and mechanical stress, obtained through piezoelectric material property testing, elastic modulus in calculation Describes the ability of a material to resist deformation when subjected to external forces. It can be obtained through material mechanical testing. The dielectric loss factor δ can be measured by an LCR meter or an impedance analyzer.

[0117] The RMSD values ​​of the frequency points marked as obvious damage and the corresponding electrical impedance Y(ω) are collected and correlation is calculated, which is expressed as:

[0118]

[0119] Where M represents the total number of frequency points, Y(ω k ) represents the electrical impedance value at frequency point k, represents the mean value of electrical impedance, represents the RMSD mean;

[0120] The t-distribution is used for the test and the t-statistic is calculated as:

[0121]

[0122] Based on the universal significance level β of 0.05 and the degree of freedom M-2, the corresponding t distribution critical value t is found by consulting the t distribution table. c , and further calculate the correlation threshold, expressed as:

[0123]

[0124] Where t represents the t statistic, r c represents the correlation threshold;

[0125] If the calculated r value is greater than or equal to r c , then judge Y(ω k ) and RMSD k 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 r value is less than r c , then judge Y(ω k ) and RMSD k There is no strong linear relationship between them;

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

[0128]

[0129] Where Δω represents the average frequency interval, B represents the total number of frequency points k with strong linear relationship, represents the frequency value of the i+1th frequency point k, represents the frequency value of the i-th frequency point k;

[0130] The sensitive range of each frequency point with strong linear relationship is determined by taking Δω as a floating value, and the overlapping sensitive ranges are merged to form a sensitive frequency range.

[0131] By real-time monitoring of the electrical impedance value of the glass fiber epoxy board at key frequency points, it is possible to instantly detect small changes in the material, thereby achieving early damage identification, realizing high-precision monitoring technology and improving the overall quality and reliability of the product. By comparing historical data, the health history of the material can be tracked, providing strong data support for the realization of dynamic traceability, making the production process more transparent and controllable. By calculating and applying correlation thresholds, the system can automatically identify possible quality problems, prompt operators or automatically adjust the production process, improve the reliability and safety of the production process, and help companies take the initiative to prevent potential problems during the production process rather than responding to them afterwards. Through correlation analysis, the system can automatically screen out the frequency points that are most sensitive to damage, and calculate the sensitive range of these frequency points, and use these sensitive frequency ranges in subsequent monitoring to ensure that the system can capture the subtlest changes in the material state. By combining electrical impedance measurement, RMSD analysis and 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 locate the root cause of the problem, and thus improve the efficiency of problem solving.

[0132] S4, construct a random forest model and train it through K-fold cross validation, input the model based on the frequency points in the sensitive frequency range, determine the prediction threshold through the cumulative distribution function, and determine the model output;

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

[0134] Based on the electrical impedance Y(ω calculated within the sensitive frequency range k ) values, construct the model data set, including Y(ω k ) and the corresponding status label;

[0135] Construct a random forest model and input Y(ω) at different frequency points in the model dataset into the model's decision tree. k )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 takes a weighted average of the prediction results of all decision trees and selects the maximum value as the final prediction output based on the probability value output by the model.

[0137] K-fold cross validation is used during model training. The data set is split and one part is used as the validation set, and the rest is used as the training set. This is repeated K times, and different data is selected as the validation set each time. After each training, the model is scored with the validation set, the performance indicators of the model are recorded, and the best model parameter combination is selected for final training and the model parameters are confirmed.

[0138] The model output probabilities during the model training process are sorted, and the cumulative distribution function value F(P) of each probability value is calculated. The prediction threshold is calculated by taking the universal significance level β as 0.05, which is expressed as:

[0139]

[0140] P c =min{PF(P)≥1-β};

[0141] Where X represents the total number of samples, P i represents the probability value of the i-th sample, O(P i ≤P) is an indicator function, which takes 1 when the condition is met, otherwise it takes 0; P c represents the prediction threshold;

[0142] In actual monitoring, the system will collect the electrical impedance Y(ω k ) is input into the random forest model, and the maximum value is selected as the final prediction output according to 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 damage.

[0143] Through real-time monitoring of the impedance value within the sensitive frequency range, the model can detect subtle changes in the material in time and predict the possibility of damage. When the model output probability value exceeds the predetermined threshold, the system automatically triggers an alarm, warns the operator in advance or automatically adjusts the production process to prevent the damage from expanding or developing into a more serious problem, and realizes the damage prediction of glass fiber epoxy board materials. Through the analysis and prediction of glass fiber epoxy board materials by the model, the material performance and life cycle can be understood in time, so that in the reproduction process, the actual performance differences of different proportions of materials or even different batches of materials can be understood in time, and the behavior and life of materials can be better understood, and the material use and maintenance strategies can be optimized. Through quality prediction based on non-sensing operation, material defects can be discovered in time, and quality traceability can be carried out to improve product quality and safety, ensure the efficiency of the production process and the high quality of products. Through real-time impedance monitoring within the sensitive frequency range, the model can accurately track every change in the material state, thereby realizing continuous quality control during non-sensing operation. When the model output probability exceeds the set threshold, the system immediately issues an alarm, which allows instant response to avoid the spread of quality defects, ensures the quality and consistency of the production process, and realizes quality control and traceability of the entire process from raw materials to finished products.

[0144] S5, based on the output judgment of the model, remote early warning is carried out and early warning information is sent through multiple terminals and early warning alarms are issued. Data is stored in a distributed manner and encrypted and transmitted in real time. Access rights are defined and log data is backed up based on DES encryption.

[0145] Preferably, based on the output judgment of the model, remote early warning is carried out and early warning information is sent through multiple terminals and early warning alarms are issued, which means that through the judgment results of the random forest model, remote monitoring and early warning are carried out based on the IoT Internet of Things. If a signal of quality damage is detected, a text message is sent through the SMS gateway, and an email is sent through the SMTP server to warn the operator, and specific 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] Through the integration of early warning functions, the system can notify relevant personnel through multiple channels at the moment of potential risks to ensure the timeliness and accuracy of information transmission, including through SMS and email, on-site early warning reminders and real-time early warning information. The use of multiple early warning methods and the synchronous triggering of on-site alarm devices enable operators to take quick action to avoid accidents or further material damage. The system not only improves the safety of the operating process, but also provides strong support for achieving quality traceability under non-sensing operation, ensuring the stability and reliability of the industrial production process.

[0147] Furthermore, by distributing data storage and encrypting data in real time, defining access rights and backing up log data based on DES encryption, including:

[0148] The distributed storage method is used to store the collected data and the calculated data in real time and securely. The collected data and the calculated data are encrypted through the end-to-end encryption protocol TLS. When the collected data and the calculated data are transmitted to the storage device, they are distributedly stored in a multi-node cloud database, and the multi-node data in the cloud database is replicated and backed up.

[0149] The role-based access control (RBAC) mechanism defines access rights, clarifies data access rights in the cloud database, records all access and operation record data, and records the access and operation record data in log files. The log data is processed through DES encryption and periodically backed up to off-site storage.

[0150] By adopting distributed storage, the system effectively improves the security and reliability of data. The combination of encrypted transmission and distributed storage ensures that data can remain intact and secure even during transmission or when the storage node is attacked. The role-based access control and encrypted log mechanisms ensure the security and transparency of data access, providing a reliable basis for subsequent auditing and tracing. At the same time, by storing encrypted logs off-site, the security of personnel access control can be further improved. The off-site storage of log files improves security, facilitates access to and tracing of personnel's operation records, and improves traceability capabilities.

[0151] Example 2

[0152] If the functions are implemented in the form of 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 the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program codes.

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

[0154] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0155] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used to implement: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0156] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A non-sensing operation quality tracing method for the glass fiber industry, characterized by: include, Select PZT sensor for installation, cooperate with analog-to-digital converter to obtain collected data, calculate impedance value and identify characteristics in low-frequency range and high-frequency range respectively, calculate signal-to-noise ratio to confirm and identify characteristic frequency points, analyze the change rate of adjacent identified characteristic frequency points, determine frequency interval and integrate scanning frequency range; Impedance measurement is performed based on the introduction of bonding defects into 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. The defect threshold of the RMSD value is calculated based on historical data and the frequency points are marked; Obtain material parameters and calculate electrical impedance according to the scanning frequency range, analyze the correlation of RMSD value and electrical impedance according to the marked frequency points, use t distribution for testing, calculate t statistics and screen frequency points with strong linear relationship, and integrate sensitive frequency range based on frequency interval; A random forest model was constructed and trained through K-fold cross validation. The frequency points in the sensitive frequency range were input into the model, the prediction threshold was determined through the cumulative distribution function, and the model output was determined. Based on the output of the model, remote early warning is carried out and early warning information is sent through multiple terminals and early warning alarms are issued. Data is stored in a distributed manner and encrypted in real time. Access rights are defined and log data is backed up based on DES encryption.

2. The non-sensing operation quality tracing method for the glass fiber industry as claimed in claim 1, characterized in that: The PZT sensor is selected for installation and matched with the analog-to-digital converter to obtain the collected data. include, Select the PZT sensor for installation, clean the selected pasting position of the glass fiber epoxy board, use high-performance epoxy resin adhesive to paste the cut PZT sensor to the selected pasting position, and let it stand for 24 hours until the adhesive is completely cured and pasting is completed; 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; Configure the computer to drive and read the analog-to-digital converter output data.

3. The non-sensing operation quality tracing method for the glass fiber industry as claimed in claim 2, characterized in that: The method of calculating impedance values ​​and identifying features in the low-frequency interval and the high-frequency interval, calculating the signal-to-noise ratio to confirm the identification of characteristic frequency points, analyzing the change rate of adjacent identification characteristic frequency points, determining the frequency interval and integrating the scanning frequency range includes: Based on the scenario where the PZT sensor is not connected to the glass fiber epoxy board sample, the background noise is measured by the PZT sensor and the analog-to-digital converter, and the noise level in the range of 1kHz to 1MHz is recorded; Based on the scenario where the PZT sensor is connected to the glass fiber epoxy board sample, the PZT sensor and the analog-to-digital converter are respectively located in the low-frequency range and the high-frequency range for measurement, and the impedance value corresponding to each frequency point is recorded. The impedance calculation in the low-frequency range is expressed as: Where Z l (ω) represents the impedance of the low-frequency range frequency ω, R represents resistance, C represents capacitance, R P represents the parallel resistance, i represents the imaginary unit; The impedance in the high frequency range is calculated by dividing the voltage at frequency ω by the current at frequency ω; Based on the calculated impedance data at different frequencies, a frequency-impedance relationship graph was plotted in the chart software Matlab, and the identification features were determined based on the peak and inflection points of the relationship graph; Calculate the signal-to-noise ratio based on the recognition feature, use the historical experience value as the signal-to-noise ratio threshold, and select the recognition feature that is greater than or equal to the signal-to-noise ratio threshold; Based on the selected identification characteristic frequency point, the change rate of the corresponding impedance adjacent frequency point is calculated, and the reference step size of the adjacent frequency step size is set based on the history. The 10th percentile of the initial change rate is used as the change threshold, and the reference step size is gradually expanded to adjust the frequency point, which is expressed as: Where R(ω k ±n) represents the frequency point ω k +n impedance change rate, Z(ω k ±n) represents the frequency point ω k +n impedance; Compare the change rate calculated each time with the threshold. When the change rate is lower than the set change threshold, stop adjusting the frequency point and confirm the frequency interval based on the identified characteristic frequency point. Integration is performed based on the expanded frequency intervals, and the overlapping parts of the frequency intervals are integrated to form a continuous frequency range, which is set as the scanning frequency range.

4. The non-sensing operation quality tracing method for the glass fiber industry as claimed in claim 3, characterized in that: The method includes: introducing bonding defects into the sample to perform impedance measurement, calculating the real part of the impedance based on the measured voltage and current, and quantifying the specific impact of the introduced defects on the impedance characteristics of the glass fiber epoxy board sample, calculating the defect threshold of the RMSD value based on historical data and marking the frequency point, including: Introduce bonding defects based on the glass fiber epoxy board sample, use a cutting tool to cut the defects on the glass fiber epoxy board sample, and use a marking pen to accurately mark the position and size of each defect on the glass fiber epoxy board; Use a PZT sensor to measure impedance over a scanning frequency range, and obtain information about the impedance, position, and size of each measurement frequency point; The voltage V is measured using an impedance analyzer at the measurement frequency point k where the bonding defect is introduced. k (t) and current I k The voltage and current signals at each frequency point k are Fourier transformed to obtain the voltage phase and current phase By voltage phase and current phase Subtract to calculate the phase angle θ k ; At the frequency point k, the real part of the impedance is calculated based on the measured voltage and current values, and the specific effect of the introduced defects on the impedance characteristics of the glass fiber epoxy board sample is quantified and the RMSD is calculated. k value; Apply RMSD calculation to the data before and after each group of experiments with the introduction of bonding defects, calculate the RMSD values ​​of all frequency points, and organize the calculation results into a table, indicating the RMSD values ​​under different defect states; The average value and the sum of two times the standard deviation of the RMSD value based on the sample historical data are used as the defect threshold, and the frequency points with RMSD values ​​greater than or equal to the defect threshold are marked as obvious damage.

5. The non-sensing operation quality tracing method for the glass fiber industry as claimed in claim 4, characterized in that: The method of obtaining material parameters and calculating electrical impedance according to the scanning frequency range, analyzing the correlation of RMSD value and electrical impedance according to the marked frequency points, using t distribution for testing, calculating t statistics and screening frequency points with strong linear relationships, and integrating the sensitive frequency range based on the frequency interval includes: Based on the PZT sensor and the analog-to-digital converter, the glass fiber epoxy board to which the PZT sensor is attached is subjected to a frequency scan. The scanning range is based on the scanning frequency range. The material parameters including the dielectric constant, dielectric loss factor, piezoelectric coefficient and elastic modulus are obtained, and the electrical impedance is calculated, which is expressed as: Z s (ω)=pvA; Where p represents the material density, v represents the speed at which the vibration sound wave propagates through the material, A represents the cross-sectional area of ​​the material, and Y(ω) represents the electrical impedance. represents the relative dielectric constant, δ represents the dielectric loss factor, and Z represents the energy loss caused by the polarization inside the material. s (ω) and Z α (ω) represent the mechanical impedance of the material itself and the additional impedance, d 3x represents the piezoelectric coefficient, represents the elastic modulus, i represents the imaginary unit, and α represents the sensor scale factor; Collect the RMSD values ​​of the frequency points marked as obvious damage and the corresponding electrical impedance Y(ω), calculate the correlation through the Pearson correlation coefficient, use the t distribution for testing, calculate the t statistic, and find the corresponding t distribution critical value t by consulting the t distribution table. c , and further calculate the correlation threshold r c ; If the calculated r value is greater than or equal to r c , then judge Y(ω k ) and RMSD k There is a strong linear relationship between them, and it is determined that the corresponding frequency point k is more sensitive to damage; Based on the frequency intervals between all adjacent frequency points with strong linear relationships, the average value Δω of the frequency intervals is calculated, Δω is used as a floating value to determine the sensitive range of each frequency point with strong linear relationships, and the overlapping sensitive ranges are merged to form a sensitive frequency range.

6. The non-sensing operation quality tracing method for the glass fiber industry as claimed in claim 5, 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: Based on the electrical impedance Y(ω calculated within the sensitive frequency range k ) values, construct the model data set, including Y(ω k ) and the corresponding status label; Construct a random forest model and input Y(ω) at different frequency points in the model dataset into the model's decision tree. k )value; The model samples the input data multiple times and constructs multiple decision trees. Each decision tree independently classifies the input data. The model takes a weighted average of the prediction results of all decision trees and selects the maximum value as the final prediction output based on the probability value output by the model. K-fold cross validation is used during model training. The data set is split and one part is used as the validation set, and the rest is used as the training set. This is repeated K times, and different data is selected as the validation set each time. After each training, the model is scored with the validation set, the performance indicators of the model are recorded, and the best model parameter combination is selected for final training and the model parameters are confirmed. Sort the model output probabilities during model training, calculate the cumulative distribution function value F(P) of each probability value and determine the prediction threshold; In actual monitoring, the system will collect the electrical impedance Y(ω k ) is input into the random forest model, and the maximum value is selected as the final prediction output according to 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 damage.

7. The non-sensing operation quality tracing method for the glass fiber industry as claimed in claim 6, characterized in that: The remote early warning and sending of early warning information through multiple terminals and early warning alarm according to the output judgment of the model refers to remote monitoring and early warning based on the IoT Internet of Things through the judgment results of the random forest model. If a signal of quality damage is detected, a text message is sent through the SMS gateway, and an email is sent through the SMTP server to warn the operator, and 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.

8. The non-sensing operation quality tracing method for the glass fiber industry as claimed in claim 7, characterized in that: The distributed storage of data and real-time data encryption transmission, definition of access rights and DES-based encryption backup of log data, include, The distributed storage method is used to store the collected data and the calculated data in real time and securely. The collected data and the calculated data are encrypted through the end-to-end encryption protocol TLS. When the collected data and the calculated data are transmitted to the storage device, they are distributedly stored in a multi-node cloud database, and the multi-node data in the cloud database is replicated and backed up. The role-based access control (RBAC) mechanism defines access rights, clarifies data access rights in the cloud database, records all access and operation record data, and records the access and operation record data in log files. The log data is processed through DES encryption and periodically backed up to off-site storage.

9. A computer device comprising: Memory and processor; The memory stores a computer program, characterized in that: when the processor executes the computer program, the steps of the non-sensing operation quality tracing method for the glass fiber industry according to any one of claims 1 to 8 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the non-sensing operation quality tracing method for the glass fiber industry according to any one of claims 1 to 8 are implemented.

Images