Whole-period monitoring method and system for moisture absorption dynamics of flame-retardant curing agent for electrician

Abstract

The invention provides a moisture absorption dynamics full-period monitoring method and system for an electrician flame-retardant curing agent, and relates to the technical field of electrical equipment manufacturing, and the method comprises the steps: 1, selecting three barrels of samples of different positions of a flame-retardant curing agent raw material from the same flame-retardant curing agent raw material, and carrying out synchronous periodic monitoring to obtain moisture absorption rate data of the samples, constructing a moisture absorption characteristic calibration model; 2, integrating the obtained synchronous monitoring data according to a moisture absorption characteristic calibration model, and screening out an effective data point set which stably reflects the overall moisture absorption characteristic of the raw material through consistency analysis and outlier elimination to form a representative effective data set; 3, fusing the representative effective data sets, and calculating a statistical characteristic value of data at each moment to obtain an initial reference curve; according to the invention, integrated dynamic monitoring and analysis of moisture absorption dynamics of the whole process of the flame-retardant curing agent from a raw material end to a finished product end are realized.

Description

Technical Field

[0001] This invention relates to the field of power equipment manufacturing technology, and in particular to a method and system for full-cycle monitoring of the moisture absorption kinetics of electrical flame retardant curing agents. Background Technology

[0002] Flame retardant curing agents for electrical engineering are the core raw materials for preparing epoxy resin cable accessories. Their hygroscopic properties directly determine the cross-linking and curing effect of the adhesive, which in turn affects the insulation, flame retardant performance, and structural stability of the finished cable accessories. They are the core control point for production quality control. The hygroscopic monitoring of this curing agent generally adopts a segmented static detection mode, and an integrated monitoring system for hygroscopic dynamics covering the entire process from raw materials, adhesive preparation, curing, to finished products has not yet been established.

[0003] When flame retardant curing agent raw materials are received into the warehouse, single-point moisture absorption rate testing is performed by batch sampling. During the glue preparation stage, only the ambient temperature and humidity are recorded, and real-time moisture absorption monitoring is not carried out. During the curing process, temperature changes are tracked, but moisture absorption kinetic parameters are not linked. There are also no continuous moisture absorption tracking measures in the finished product storage and transportation process. Because the upper part of the raw material barrel is sampled and tested, the moisture absorption problem caused by the sealing defects of the raw material in the lower part of the barrel is not found. When the raw material in this part is mixed with epoxy resin during glue preparation, the moisture absorption rate of the glue changes instantaneously. The cross-linking and curing reaction of the glue is uneven. After the finished product is stored and transported, micropore defects in the insulation layer are detected. The batch moisture absorption rate exceeds the industry standard, which ultimately leads to the rework and rectification of the entire batch of products. The monitoring system is fragmented. It does not realize the dynamic monitoring and data linkage of the moisture absorption kinetics of the curing agent from the raw material end to the finished product end. There is a lack of distributed synchronous sampling and consistency verification mechanism for raw materials. There is no unified moisture absorption kinetics benchmark curve at the raw material end. There is no quantitative judgment basis for moisture absorption monitoring at each stage. Furthermore, the continuity of moisture absorption behavior at each stage and the correction effect of curing reaction and environmental changes on moisture absorption kinetics are not considered, making it difficult to identify abnormal moisture absorption changes. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a method and system for full-cycle monitoring of the moisture absorption kinetics of flame retardant curing agents for electrical applications, so as to realize integrated dynamic monitoring and analysis of the moisture absorption kinetics of flame retardant curing agents from the raw material end to the finished product end.

[0005] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows: Firstly, a method for full-cycle monitoring of the hygroscopic kinetics of electrical flame-retardant curing agents, the method comprising: Step 1: Select samples from different locations in three barrels of the same flame retardant curing agent raw material, perform synchronous periodic monitoring, obtain the moisture absorption rate data of the samples, and construct a moisture absorption characteristic calibration model. Step 2: Based on the hygroscopic characteristics calibration model, integrate the obtained synchronous monitoring data, and through consistency analysis and outlier removal, screen out a set of effective data points that stably reflect the overall hygroscopic characteristics of the raw materials to form a representative effective dataset; Step 3: Merge representative valid datasets, calculate statistical characteristic values ​​of data at each time point, and obtain an initial baseline curve; verify and optimize the curve by analyzing the uncertainty of the initial baseline curve and the residuals with the original data, and finally obtain the raw material end moisture absorption kinetic baseline curve. Step 4: Based on the moisture absorption kinetics baseline curve at the raw material end, conduct online real-time monitoring during the mixing and preparation of flame retardant curing agent and epoxy resin, and record the instantaneous moisture absorption response data during the glue preparation process. Step 5: Based on the instantaneous moisture absorption response data during the adhesive mixing process, monitor the mixed adhesive solution in real time throughout the crosslinking and curing reaction process, and record the moisture absorption kinetics change data under different curing conditions. Step 6: Based on the hygroscopic kinetics data during the curing process, continuously monitor the cured cable accessories during storage and transportation to obtain long-term hygroscopic behavior data of the finished product.

[0006] Furthermore, samples from different locations within three barrels of the same flame-retardant curing agent raw material were selected and simultaneously and periodically monitored to obtain moisture absorption rate data. A moisture absorption characteristic calibration model was then constructed, including: Synchronous sampling instructions are sent to the distributed humidity sensing nodes deployed at the top, middle and bottom of the three barrels of flame retardant curing agent raw materials to start periodic quality monitoring of the three barrels of samples under the same environmental conditions and obtain the original quality time series data returned by each sensing node. Based on the original mass time series data, time axis alignment was performed according to a unified timestamp to form three sets of synchronous mass change sequences with the same time reference. Based on the three sets of synchronous mass change sequences, normalization calculation was performed in combination with the initial dry basis mass value of each barrel to obtain the moisture absorption rate time series data of the three barrel samples at each monitoring time. Based on the time series data of moisture absorption rate, a horizontal consistency comparison was performed among the three barrel samples to identify and mark abnormal data segments that deviated from the average of the three barrels by more than a preset threshold. Based on the identified and marked time series data of moisture absorption rate, abnormal data segments were removed, and the remaining effective moisture absorption rate data was fused between barrels to extract the key kinetic parameters of the overall moisture absorption characteristics of the raw material. Based on key kinetic parameters, a calibration model for the moisture absorption characteristics of flame retardant curing agents was established by fitting and establishing the mapping relationship between the moisture absorption rate and the ambient humidity.

[0007] Furthermore, based on the hygroscopic characteristic calibration model, the obtained synchronous monitoring data are integrated. Through consistency analysis and outlier removal, a set of effective data points that stably reflect the overall hygroscopic characteristics of the raw material is selected, forming a representative and effective dataset, including: Using a moisture absorption characteristic calibration model, the time series data of moisture absorption rate of three barrel samples are input into the moisture absorption characteristic calibration model to calculate the deviation sequence between each sample data point and the model prediction value; based on the deviation sequence, the consistency analysis of the deviation of the three barrel samples at the same time is performed, and the standard deviation of the deviation values ​​of the three barrel samples at each time point is calculated. By statistically obtaining the standard deviation at each time point, the time point where the standard deviation exceeds the preset tolerance range of the moisture absorption characteristic calibration model is identified as an inconsistent time point, and all data points at the corresponding time point are marked. Based on the data points at the time of the consistency anomaly based on the label, combined with the identified abnormal data segments, a dual anomaly judgment is performed, and data points that simultaneously meet the conditions of excessive deviation and excessive inter-bucket dispersion are identified as outliers. Based on the outlier list, all data records corresponding to outliers were removed from the time series data of moisture absorption rate of the three samples, and the remaining valid data points were retained. Based on the retained valid data points, they were aggregated according to the monitoring time, and the weighted average of the valid data points of multiple buckets at each time was calculated to obtain the valid dataset.

[0008] Furthermore, representative and valid datasets are fused, and statistical characteristic values ​​of the data at each time point are calculated to obtain an initial baseline curve. The curve is then validated and optimized by analyzing the uncertainty of the initial baseline curve and the residuals compared to the original data, ultimately yielding the raw material end hygroscopic kinetic baseline curve, including: Based on representative and valid datasets, statistical fusion processing is performed on multiple valid data points at the same monitoring time to calculate the mean, standard deviation, and confidence interval of the moisture absorption rate at each time point, thereby obtaining a time series of statistical characteristic values. Based on time-series data, an initial baseline curve of the raw material's moisture absorption kinetics trend is obtained by curve fitting the relationship between moisture absorption rate and time. Based on the initial baseline curve and the confidence interval at each time point, the expanded uncertainty distribution of the initial baseline curve over the entire monitoring period is evaluated, and high-risk periods where uncertainty exceeds a preset threshold are identified. Based on the initial baseline curve and the original valid data points, the residual values ​​of each data point relative to the curve are calculated, a residual distribution histogram is constructed, and the deviation characteristics of the residuals are analyzed. Taking into account the high-risk period and the deviation characteristics, the fitting parameters of the initial baseline curve in the deviation period are locally corrected to obtain an optimized iterative version of the baseline curve. Based on the optimized benchmark curve iteration, the uncertainty assessment, residual analysis and parameter correction process is repeated until the change in curve parameters in two consecutive iterations is less than the convergence threshold, and finally the raw material end hygroscopic kinetic benchmark curve is obtained.

[0009] Further, step 4 includes: Based on the raw material end hygroscopic kinetics baseline curve, a real-time sampling command is sent to the online quality monitoring devices at the feed inlet, mixing chamber and discharge outlet of the mixing tank to start high-frequency monitoring of the mixing process and obtain the original quality time-series data stream including the mixing start time; Based on the original mass time-series data stream, environmental humidity compensation and correction are performed by combining the temperature and humidity sensor data of the glue mixing environment to obtain the net moisture absorption mass change sequence that eliminates environmental interference. Based on the corrected net moisture absorption mass change sequence, combined with the raw material mixing ratio and initial dry basis mass, the instantaneous moisture absorption rate of the entire rubber mixing process is calculated, and time normalization is performed with the mixing start time as zero point. By comparing the normalized instantaneous moisture absorption rate sequence with the raw material end moisture absorption kinetics baseline curve at each time step, the instantaneous abrupt change intervals that deviate from the baseline by more than the dynamic threshold are identified. Based on the marked instantaneous abrupt change intervals, the extreme values ​​of moisture absorption rate, integral moisture absorption amount, and duration are extracted, and an instantaneous response feature vector characterizing moisture absorption sensitivity is constructed by combining the rubber compounding process parameters. By matching the instantaneous response feature vector with the feature library of historical qualified batches, it is determined whether the current moisture absorption behavior of the adhesive is within the process tolerance range, and the moisture absorption quality status is identified. The quality status identifier, feature vector, mutation interval information, and instantaneous moisture absorption rate sequence are spatiotemporally correlated and encapsulated to form instantaneous moisture absorption response data during the glue mixing process.

[0010] Further, step 5 includes: Based on the instantaneous moisture absorption response data during the mixing process, a synchronous sampling command is sent to the quality monitoring device and temperature monitoring device on the surface and inside of the curing mold to start continuous monitoring of the mixed adhesive from the completion of mixing to the complete curing stage, and obtain the quality time-series data stream and temperature time-series data stream; Based on the quality time-series data stream and the temperature time-series data stream, alignment processing is performed according to a unified timestamp to identify the start and end times of the heating, isothermal, and cooling stages, and to divide the multi-stage time-series intervals of the curing reaction. Based on the multi-stage time interval, the moisture absorption increment and moisture absorption rate of each stage are calculated in combination with the initial mass value at the time of adhesive preparation completion, and the curing temperature conditions corresponding to each stage are marked; based on the stage moisture absorption parameters, combined with the instantaneous moisture absorption rate sequence, the continuity of moisture absorption behavior from adhesive preparation to the start of curing is analyzed, and the turning point of moisture absorption kinetics caused by the curing reaction is identified. Based on the identified turning points, the moisture absorption rate and time series of the entire curing process are segmented and fitted, and segmented feature parameters of each stage are extracted. The segmented feature parameters include the extreme value of moisture absorption rate, moisture absorption saturation time, and the proportion of total moisture absorption. Based on the extracted segmented feature parameters, and combined with curing temperature and ambient humidity conditions, a moisture absorption kinetic response matrix under different process parameter combinations is constructed. Based on the response matrix, the deviation integral is calculated with the raw material end moisture absorption kinetic baseline curve to quantify the correction effect of the curing reaction on the moisture absorption characteristics of the raw material, and the moisture absorption kinetic change data are obtained.

[0011] Furthermore, step 6 includes: Based on the hygroscopic dynamics change data, low-frequency periodic sampling commands are sent to the quality monitoring devices and environmental monitoring devices embedded on the surface and inside of the finished cable accessories to initiate long-term monitoring of the finished product in the storage and transportation environment from the time of demolding and production line, and obtain the time-series data stream of finished product quality and environmental temperature and humidity. The time-series data streams of finished product quality and environmental temperature and humidity are time-aligned and fused to identify environmental switching moments and divide the time-series intervals of multiple scenarios, including warehouse static storage, loading and transfer, unloading and temporary storage, and final warehousing. Based on multiple scenario intervals, the cumulative moisture absorption increment and average moisture absorption rate of each interval are calculated in combination with the initial dry base mass after demolding to obtain moisture absorption parameters; based on the moisture absorption parameters, the moisture absorption continuity from the completion of curing to the start of storage is analyzed, and the equilibrium time and equilibrium moisture absorption rate of the finished product when it reaches the moisture absorption equilibrium state are identified. Based on the equilibrium time and equilibrium moisture absorption rate, asymptotic fitting was performed on the moisture absorption rate and time series during the long-term storage stage to extract the moisture absorption characteristic parameters of the finished product. Based on the moisture absorption characteristic parameters of the finished product, and combined with environmental conditions in multiple scenarios, a quantitative relationship between the exposure time of different environments and the cumulative amount of moisture absorption is established. The contribution of each environmental stage to the final moisture absorption state of the finished product is quantified, and a full-cycle chain comparison is performed with the moisture absorption data of the raw material end and the curing process to generate behavioral data on the long-term moisture absorption stability of the finished product.

[0012] Secondly, a full-cycle monitoring system for the hygroscopic kinetics of electrical flame-retardant curing agents includes: The acquisition module is used to select samples from different locations of three barrels of flame retardant curing agent raw materials from the same flame retardant curing agent raw material, perform synchronous periodic monitoring, obtain moisture absorption rate data of the samples, and construct a moisture absorption characteristic calibration model. The screening module is used to integrate the obtained synchronous monitoring data according to the hygroscopic characteristics calibration model, and through consistency analysis and outlier removal, screen out a set of effective data points that stably reflect the overall hygroscopic characteristics of the raw materials to form a representative effective dataset. The optimization module is used to fuse representative and valid datasets, calculate the statistical characteristic values ​​of the data at each time point, and obtain the initial baseline curve. By analyzing the uncertainty of the initial baseline curve and the residuals with the original data, the curve is verified and optimized, and finally the raw material end moisture absorption kinetic baseline curve is obtained. The monitoring module is used to perform online real-time monitoring during the mixing and preparation of flame retardant curing agent and epoxy resin based on the moisture absorption kinetics baseline curve at the raw material end, and record the instantaneous moisture absorption response data during the glue preparation process. The recording module is used to monitor the mixed adhesive in real time during the entire crosslinking and curing process based on the instantaneous moisture absorption response data during the adhesive preparation process, and record the moisture absorption kinetics change data under different curing conditions. The processing module is used to continuously monitor the cured cable accessories during storage and transportation based on the hygroscopic kinetic changes during the curing reaction process, and obtain long-term hygroscopic behavior data of the finished product.

[0013] Thirdly, a computing device includes: One or more processors; A storage device for storing one or more programs that, when executed by one or more processors, cause the one or more processors to implement the method.

[0014] Fourthly, a computer-readable storage medium storing a program that, when executed by a processor, implements the method.

[0015] The above-described solution of the present invention has at least the following beneficial effects: This invention overcomes the technical problems of existing technologies in electrical flame retardant curing agent moisture absorption monitoring systems. These problems include fragmentation, lack of full-cycle dynamic monitoring, lack of a distributed synchronous sampling mechanism and unified quantitative benchmark curve, lack of monitoring basis for each stage, and difficulty in identifying abnormal moisture absorption changes. The invention employs a method of synchronously and periodically monitoring and analyzing the moisture absorption dynamics of three barrels from the same flame retardant curing agent raw material to the finished product, effectively capturing the moisture absorption change patterns and abnormal changes at each stage, avoiding chain quality defects caused by localized moisture absorption problems, improving the quality control of cable accessory production, and ensuring the insulation, flame retardant performance, and structural stability of the finished cable accessories. Attached Figure Description

[0016] Figure 1 This is a schematic flowchart of a method for full-cycle monitoring of the moisture absorption kinetics of electrical flame retardant curing agents provided in an embodiment of the present invention.

[0017] Figure 2 This is a schematic diagram of a full-cycle monitoring system for the moisture absorption kinetics of an electrical flame retardant curing agent provided in an embodiment of the present invention. Detailed Implementation

[0018] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough understanding of the present disclosure and to fully convey its scope to those skilled in the art.

[0019] like Figure 1 As shown, embodiments of the present invention propose a method for full-cycle monitoring of the moisture absorption kinetics of electrical flame retardant curing agents, the method comprising the following steps: Step 1: Select samples from different locations in three barrels of the same flame retardant curing agent raw material, perform synchronous periodic monitoring, obtain the moisture absorption rate data of the samples, and construct a moisture absorption characteristic calibration model. Step 2: Based on the hygroscopic characteristics calibration model, integrate the obtained synchronous monitoring data, and through consistency analysis and outlier removal, screen out a set of effective data points that stably reflect the overall hygroscopic characteristics of the raw materials to form a representative effective dataset; Step 3: Merge representative valid datasets, calculate statistical characteristic values ​​of data at each time point, and obtain an initial baseline curve; verify and optimize the curve by analyzing the uncertainty of the initial baseline curve and the residuals with the original data, and finally obtain the raw material end moisture absorption kinetic baseline curve. Step 4: Based on the moisture absorption kinetics baseline curve at the raw material end, conduct online real-time monitoring during the mixing and preparation of flame retardant curing agent and epoxy resin, and record the instantaneous moisture absorption response data during the glue preparation process. Step 5: Based on the instantaneous moisture absorption response data during the adhesive mixing process, monitor the mixed adhesive solution in real time throughout the crosslinking and curing reaction process, and record the moisture absorption kinetics change data under different curing conditions. Step 6: Based on the hygroscopic kinetics data during the curing process, continuously monitor the cured cable accessories during storage and transportation to obtain long-term hygroscopic behavior data of the finished product. In this embodiment of the invention, the present invention employs a method of synchronously and periodically monitoring three samples from different locations within the same flame-retardant curing agent raw material to obtain moisture absorption rate data and construct a moisture absorption characteristic calibration model. Then, through consistency analysis and outlier removal, effective data is integrated and screened to form a representative and effective dataset. After data fusion, statistical feature value calculation, and curve verification optimization, a baseline curve for the moisture absorption kinetics of the raw material is obtained. Finally, based on this baseline curve, the entire process of adhesive mixing, cross-linking and curing of the mixed adhesive, and the storage and transportation of the finished product are monitored in real-time at different stages, and the corresponding moisture absorption data is recorded. This overcomes the technical problems of fragmented moisture absorption monitoring of flame-retardant curing agents in existing technologies, such as the lack of a full-cycle dynamic monitoring mechanism, the lack of distributed synchronous sampling and unified quantitative benchmarks for raw materials, and the lack of basis for monitoring at each stage, making it difficult to capture abnormal moisture absorption changes. This achieves integrated dynamic monitoring of the moisture absorption kinetics of flame-retardant curing agents from the raw material end to the finished product end, tracking the moisture absorption change patterns at each stage, effectively identifying moisture absorption anomalies, providing reliable data support for moisture absorption quality control at each stage of cable accessory production, avoiding finished product quality defects caused by moisture absorption problems, and ensuring the production quality and performance of cable accessories.

[0020] In a preferred embodiment of the present invention, step 1 above may include: Step 1.1: Send a synchronous sampling command to the distributed humidity sensing nodes deployed on the upper, middle, and lower parts of the three barrels of flame retardant curing agent raw materials. Initiate periodic quality monitoring of the three barrels under the same environmental conditions to obtain the raw quality time-series data returned by each sensing node. Specifically, this includes: deploying distributed humidity sensing nodes on the three barrels of the same batch of flame retardant curing agent raw materials. One humidity sensing node is fixed at the top (10 cm from the barrel opening), the midpoint of the barrel height, and the bottom (10 cm from the barrel bottom) of each barrel. After all sensing nodes have completed preliminary accuracy calibration, a unified synchronous sampling command is sent to all distributed humidity sensing nodes through the data processing terminal. This initiates periodic quality monitoring of the three barrels with deployed sensing nodes. The curing agent raw materials were transferred to the same monitoring environment, with the temperature controlled at 25 degrees Celsius and the relative humidity at 60%. The monitoring environment was kept free from significant airflow interference. Periodic quality monitoring of the three raw material samples was initiated, with a sampling period of 30 minutes and a single continuous monitoring duration of 72 hours. Each distributed humidity sensor node collected the quality data of the raw material at the corresponding location at each sampling moment according to the synchronous sampling instruction, and transmitted the collected quality data back to the data processing terminal in real time. The terminal classified and stored the transmitted data according to the raw material barrel number and the sensor node location, forming the original quality time series data corresponding to each sensor node. The data record content includes the specific sampling time and the corresponding raw material quality value at that time.

[0021] Step 1.2: Based on the original quality time-series data, time axis alignment is performed according to a unified timestamp to form three sets of synchronous quality change sequences with the same time reference. Based on the three sets of synchronous quality change sequences, normalization calculation is performed in combination with the initial dry basis quality value of each barrel to obtain the moisture absorption rate time-series data of the three barrel samples at each monitoring time. Specifically, this includes: using the system time of the data processing terminal as a unified reference, matching a unique unified timestamp for all original quality time-series data, with the timestamp accuracy set to 1 minute; calibrating and aligning the original quality time-series data returned by each sensor node according to the unified timestamp time step by step; removing invalid data without corresponding timestamps and abnormal data with mismatched timestamps; and fusing the aligned data from the sensor nodes at the top, middle, and bottom positions of each barrel to form three sets corresponding to the three barrels respectively. The synchronous mass change sequence of raw materials includes consecutive sampling times and corresponding mass change values ​​at each time. Initial dry basis mass measurements were performed on three barrels of flame retardant curing agent raw materials beforehand. The method involved drying the raw materials in a 105°C drying device until constant weight, then weighing them and recording the initial dry basis mass value for each barrel. Based on this initial dry basis mass value, normalization calculations were performed on the mass data at each sampling time in the three synchronous mass change sequences. The moisture absorption rate of each raw material barrel at each monitoring time was obtained through calculation. Then, the arithmetic mean of the moisture absorption rate values ​​at the top, middle, and bottom positions of each barrel was taken. Finally, the time-series data of the moisture absorption rate corresponding to each of the three barrel samples at each monitoring time was obtained. The time-series data of the moisture absorption rate completely records the moisture absorption rate values ​​corresponding to each sampling time and the three barrel samples.

[0022] Step 1.3: Based on the moisture absorption rate time-series data, perform a horizontal consistency comparison among the three barrel samples, identify and mark abnormal data segments that deviate from the average of the three barrels by more than a preset threshold; based on the identified and marked moisture absorption rate time-series data, remove abnormal data segments, and perform inter-barrel fusion processing on the remaining effective moisture absorption rate data to extract the key kinetic parameters of the overall moisture absorption characteristics of the raw material. Specifically, this includes: using the moisture absorption rate time-series data of the three barrel samples as the processing basis, extracting data according to the same sampling time, obtaining the moisture absorption rate values ​​of the three barrel samples at each time step, calculating the arithmetic mean of the three moisture absorption rate values ​​at the same time step, identifying continuous data segments where the deviation of the moisture absorption rate value of a single sample from the average of the three barrels at that time step exceeds 5% as abnormal data segments, and processing all identified abnormal data segments. The data is marked with the start and end sampling times of the abnormal data segments. Based on the marking results, all marked abnormal data segments are directly removed from the moisture absorption rate time series data of the three barrels of samples, and the remaining unmarked effective moisture absorption rate data is retained. The retained effective moisture absorption rate data is then processed by inter-barrel fusion. The effective moisture absorption rate values ​​of the three barrels of samples are integrated according to the sampling time. The overall arithmetic mean and moisture absorption rate change rate of the three barrels are calculated time by time. By performing trend analysis and feature extraction on the overall mean and change rate over the entire monitoring period, key kinetic parameters that stably reflect the overall moisture absorption characteristics of the flame retardant curing agent raw material are obtained. The key kinetic parameters include the initial moisture absorption rate of the raw material, the slope of the moisture absorption rate change, and the characteristic parameters of the moisture absorption amount changing with time.

[0023] Step 1.4: Based on key kinetic parameters, a calibration model for the moisture absorption characteristics of the flame retardant curing agent, mapping its moisture absorption rate to ambient humidity, is fitted and established. This specifically includes: using the extracted key kinetic parameters of the overall moisture absorption characteristics of the raw material as core basic data. These key kinetic parameters specifically include the initial moisture absorption rate, the slope of the moisture absorption rate change, and characteristic parameters of the moisture absorption amount changing over time for the flame retardant curing agent raw material. These parameters are organized according to the monitoring time sequence to form a time-series sequence of key kinetic parameters, ensuring that each sampling time corresponds to a complete set of key kinetic parameters. Simultaneously, the data for each sampling time within the entire monitoring period is retrieved synchronously. The corresponding environmental humidity monitoring data consists of relative humidity values ​​collected synchronously in the monitored environment, which correspond one-to-one with the sampling times of the key kinetic parameter time series, forming an environmental humidity time series data sequence. During model construction, the data correspondence is established, with environmental humidity data as the independent variable and the moisture absorption rate data of the flame retardant curing agent as the dependent variable. The moisture absorption rate data is calculated from the initial moisture absorption rate and the slope of the rate change in the key kinetic parameter time series. The environmental humidity value is matched with the corresponding moisture absorption rate value at each time step, forming a complete set of independent and dependent variable data pairs, comprising... The data includes all sampling times within the entire monitoring period to ensure the integrity and correlation of the data pairs. A nonlinear fitting method is used to perform correlation analysis and model construction on the two sets of data. During the fitting process, the actual moisture absorption characteristics of the flame-retardant curing agent raw material are fully considered, namely, the moisture absorption rate of the raw material increases nonlinearly with increasing ambient humidity, and the rate of increase gradually slows down, to avoid the fitted curve deviating from the actual moisture absorption characteristics. To eliminate fitting bias caused by data fluctuations, all data pairs are preprocessed, removing data pairs whose moisture absorption rate deviates from the ambient humidity by more than 0.001, retaining the valid data pairs; then, the valid data pairs are iterated step by step. For the fitting process, after each iteration, the mean deviation between the fitted curve and the effective data pairs is calculated. If the mean deviation is > 0.0005, the fitting parameters are adjusted and the iteration is repeated until the mean deviation is ≤ 0.0005, at which point the iteration stops and the fitting calculation is complete. After fitting, the reliability of the model is further verified by selecting environmental humidity data from 10 random sampling times within the entire monitoring period and inputting them into the fitted model to obtain the corresponding predicted moisture absorption rate. The predicted value is then compared with the actual moisture absorption rate at that time. If the deviation of all compared data is < 0.0008, the model is considered to be fit; otherwise, the model is considered to be fit.In case 0008, the fitting parameters are readjusted, and the above fitting and verification process is repeated until the model prediction deviation meets the requirements. Finally, a precise mapping relationship is established between the moisture absorption rate of the flame-retardant curing agent and the ambient humidity, forming a calibration model for the moisture absorption characteristics of the flame-retardant curing agent raw material. This calibration model contains complete fitting parameters and mapping rules. Based on any input ambient humidity value (range consistent with the monitored ambient humidity, i.e., 40%-80%), it can quickly output the corresponding predicted value of the flame-retardant curing agent's moisture absorption rate through built-in mapping rules, accurately reflecting the variation law of the raw material's moisture absorption rate under different ambient humidity conditions.

[0024] In this embodiment of the invention, a distributed humidity sensing node deployed at the top, middle, and bottom of three barrels of flame-retardant curing agent raw materials is used to send synchronous sampling commands. Periodic quality monitoring is conducted under the same environmental conditions to obtain raw quality time-series data. Then, moisture absorption rate time-series data is obtained through unified timestamp alignment and normalization calculation of the initial dry basis mass. After horizontal consistency comparison and marking of the three barrel samples and removal of abnormal data segments, key kinetic parameters are extracted from the valid data through inter-barrel fusion. Finally, a moisture absorption characteristic calibration model that fits the mapping relationship between moisture absorption rate and ambient humidity is established based on these parameters. Therefore, it overcomes the technical problems in existing raw material moisture absorption monitoring, such as the lack of a distributed synchronous sampling mechanism, the absence of a unified time benchmark for data, the inability to capture differences in moisture absorption at different locations of the raw material, the interference of abnormal data with monitoring accuracy, and the lack of calibration basis for moisture absorption characteristics, which lead to unrepresentative raw material moisture absorption data and the inability to reflect the overall moisture absorption characteristics of the raw material. It achieves multi-location, synchronous, and periodic monitoring of raw materials, ensures the synchronicity, accuracy, and completeness of moisture absorption rate data, effectively eliminates interference from abnormal data, extracts key kinetic parameters of the overall moisture absorption characteristics of the raw material, and constructs a moisture absorption characteristic calibration model that fits the actual moisture absorption law of the raw material.

[0025] In a preferred embodiment of the present invention, step 2 above may include: Step 2.1: Using the moisture absorption characteristic calibration model, input the time series data of the moisture absorption rate of the three barrels of samples into the moisture absorption characteristic calibration model, and calculate the deviation sequence between each sample data point and the model prediction value; based on the deviation sequence, perform inter-barrel deviation consistency analysis on the data points of the three barrels of samples at the same time, and calculate the standard deviation of the deviation values ​​of the three barrels of samples at each time. Specifically, this includes: inputting the complete time series data of the moisture absorption rate of the three barrels of flame retardant curing agent samples into the constructed flame retardant curing agent moisture absorption characteristic calibration model; the model outputs the predicted moisture absorption rate value corresponding to each data point of each sample based on the sampling time and the environmental humidity parameter at that time; and calculating the actual value of the sample moisture absorption rate and the model prediction value for each data point. The differences are used to form the deviation sequences of the three buckets of samples. The deviation sequences of each bucket are sorted by sampling time and include all sampling times and the deviation values ​​at the corresponding times. After the deviation sequences are constructed, the deviation sequences of the three buckets of samples are integrated according to the same sampling time. The deviation values ​​of the three buckets of samples at each time are extracted. Inter-bucket deviation consistency analysis is performed on the three buckets of samples. Statistical calculations are performed on the three deviation values ​​at each sampling time to obtain the standard deviation of the deviation values ​​of the three buckets of samples at each time. The standard deviations of all sampling times are sorted in chronological order to form the time series data of the standard deviations at each time. This data completely records each sampling time and the standard deviation value calculated at the corresponding time.

[0026] Step 2.2: By statistically analyzing the standard deviations at each time point, moments with standard deviations exceeding the preset tolerance range of the moisture absorption characteristic calibration model are identified as consistency anomalies, and all data points at the corresponding moments are marked. Specifically, this includes: pre-setting the preset tolerance range of the moisture absorption characteristic calibration model to 0.02; comparing the time-series data of the obtained standard deviations at each time point with this preset tolerance range; identifying sampling moments with standard deviations exceeding 0.02 as consistency anomalies; uniformly numbering and marking all identified consistency anomalies; and simultaneously marking all moisture absorption rate data points corresponding to the three samples at each consistency anomaly moment. The marking information includes the consistency anomaly moment number, the specific sampling time, the raw material barrel number to which the corresponding data point belongs, and the actual moisture absorption rate value of the data point. All marking information is organized and stored in chronological order to ensure that the marked consistency anomalies and corresponding data points are traceable.

[0027] Step 2.3: Based on the marked consistency anomaly time data points, combined with the identified abnormal data segments, perform dual anomaly judgment. Data points that simultaneously meet the criteria of excessive deviation and excessive dispersion between barrels are identified as outliers. Specifically, this includes: retrieving the identified and marked raw material moisture absorption rate abnormal data segments, organizing all sampling time intervals contained in the abnormal data segments and all sampling times within the corresponding intervals, and then retrieving the marked consistency anomaly times and corresponding data point information. The two types of marked data are then cross-compared and integrated. Check all marked data points with inconsistent anomalies at each time step to determine whether the sampling time of the data point falls within the sampling time interval corresponding to the identified abnormal data segment. If a data point is both at a time with a standard deviation exceeding 0.02 and its sampling time falls within the time interval of the abnormal data segment, thus satisfying both the excessive deviation and excessive dispersion between barrels, the data point is identified as an outlier. After completing the determination of all marked data points with inconsistent anomalies, compile an outlier list, which includes the specific sampling time of the outlier, the barrel number to which it belongs, the actual value of the moisture absorption rate, the deviation value, and the information of the abnormal data segment interval to which it belongs.

[0028] Step 2.4: Based on the outlier list, remove all data records corresponding to outliers from the moisture absorption rate time series data of the three barrel samples, retaining the remaining valid data points. Aggregate the retained valid data points according to the monitoring time, and calculate the weighted average of the valid data points from multiple barrels at each time point to obtain the valid dataset. Specifically, this includes: using the compiled outlier list as a basis, searching line by line in the moisture absorption rate time series data of the three barrel samples for all data records corresponding to outliers, and completely removing the retrieved outlier data records from the original time series data. The remaining data points after the removal operation are the valid data points that can reflect the true moisture absorption characteristics of the raw material. All retained valid data points are classified and aggregated according to the sampling time, grouping the valid data points of the three barrel samples at the same sampling time into one group, which is the valid dataset corresponding to the sensor nodes at different locations on the raw material barrel. Data points are assigned weights: the weight of data points corresponding to upper sensor nodes is 0.3, the weight of data points corresponding to middle sensor nodes is 0.4, and the weight of data points corresponding to lower sensor nodes is 0.3. If valid data points of some raw material barrels are missing at a certain sampling time, the weights of the actual valid data points are normalized to ensure that the sum of the weights of all valid data points at that time is 1. At each time step, the weighted average of the moisture absorption rate of the valid data points of multiple barrels at that time is calculated according to the set weights. After the calculation is completed for all sampling times, the weighted average of the moisture absorption rate of all sampling times is sorted in chronological order. At the same time, the number of valid data points participating in the weighted calculation, the corresponding raw material barrel number, and the weight configuration information are recorded for each sampling time. Finally, a representative and valid dataset that can stably reflect the overall moisture absorption characteristics of the flame retardant curing agent raw material is formed.

[0029] In this embodiment of the invention, because the preferred embodiment uses the following technical means to overcome the technical problems of lack of scientific deviation analysis and anomaly judgment mechanism, inaccurate outlier identification, and lax effective data screening in the existing data integration process, which leads to the selected data being unable to stably reflect the overall hygroscopic characteristics of the raw material and is easily interfered with by abnormal data, thus affecting the accuracy of the baseline curve construction, the invention achieves the identification and removal of outliers, ensures the accuracy, consistency and representativeness of effective data points, and selects an effective dataset that can stably reflect the overall hygroscopic characteristics of the raw material.

[0030] In a preferred embodiment of the present invention, step 3 above may include: Step 3.1: Based on the representative valid dataset, perform statistical fusion processing on multiple valid data points from the same monitoring time, calculate the mean, standard deviation, and confidence interval of the moisture absorption rate at each time point, and obtain a time-series data sequence of statistical characteristic values. Specifically, this includes: using the obtained representative valid dataset as the processing basis, classifying and integrating all valid data points within the dataset according to the monitoring time, grouping the valid data points from three samples at the same monitoring time into one group, and performing statistical fusion processing on each group of data. Calculate the arithmetic mean of the moisture absorption rate of all valid data points at each time point, then calculate the standard deviation of the moisture absorption rate value at that time point relative to the mean, and simultaneously calculate the confidence interval of the moisture absorption rate at that time point at a 95% confidence level, including the upper and lower limits of the confidence interval. After completing the statistical calculation for a single time point, organize the mean, standard deviation, upper and lower limits of the confidence interval of the moisture absorption rate corresponding to all monitoring times according to the chronological order of monitoring time to form a time-series data sequence of statistical characteristic values.

[0031] Step 3.2: Based on the time-series data sequence, an initial baseline curve for the moisture absorption kinetics trend of the raw material is obtained by curve fitting the relationship between moisture absorption rate and time. Based on the initial baseline curve and the confidence intervals at each time point, the expanded uncertainty distribution of the initial baseline curve over the entire monitoring period is evaluated, and high-risk periods where the uncertainty exceeds a preset threshold are identified. Specifically, this includes: based on the time-series data sequence of the obtained statistical characteristic values, extracting the average moisture absorption rate at each monitoring time and the corresponding time point in the sequence; using the monitoring time as the horizontal axis and the average moisture absorption rate as the vertical axis, performing nonlinear trend fitting on the relationship between moisture absorption rate and time to conform to the actual moisture absorption law of the flame retardant curing agent raw material, generating an initial baseline curve that reflects the overall moisture absorption kinetics trend of the raw material. The system can output the predicted value of the raw material moisture absorption rate at any monitoring time. Based on the fitting result of the initial baseline curve and combined with the confidence interval of each time in the time series data sequence, the system evaluates the expanded uncertainty distribution of the initial baseline curve in the entire monitoring period, calculates the expanded uncertainty value corresponding to the curve prediction value at each time, and organizes it into time series distribution data of expanded uncertainty in chronological order. The preset threshold of expanded uncertainty is set to 0.03. The values ​​in the time series distribution data of expanded uncertainty are compared with the preset threshold at each time. Continuous monitoring periods with expanded uncertainty values ​​exceeding 0.03 are identified as high-risk periods. The start and end times of all high-risk periods are recorded, as well as the monitoring intervals with high fitting deviation of the initial baseline curve.

[0032] Step 3.3: Based on the initial baseline curve and the original valid data points, calculate the residual value of each data point relative to the curve, construct a residual distribution histogram, and analyze the deviation characteristics of the residuals; combining the high-risk period and deviation characteristics, locally correct the fitting parameters of the initial baseline curve during the deviation period to obtain an optimized iterative version of the baseline curve. Specifically, this includes: retrieving the obtained initial baseline curve, extracting the predicted moisture absorption rate corresponding to each monitoring time on the curve, and simultaneously retrieving the original valid data points from the representative valid dataset to obtain the monitoring time and actual moisture absorption rate value corresponding to each data point; calculating the difference between the actual moisture absorption rate value of each data point and the predicted value of the corresponding time on the initial baseline curve for each data point to obtain the residual value of each data point relative to the curve, and organizing it into a residual sequence according to the monitoring time order; based on the residuals... A histogram of residual distribution was constructed, and all residual values ​​were grouped according to the residual value interval of 0.005. The number of residual data points in each interval was counted, and a histogram was drawn with the residual value interval as the horizontal axis and the data point frequency as the vertical axis. The deviation characteristics of the residuals were analyzed by the histogram, including the positive and negative distribution patterns and the range of numerical values, to determine whether the residuals were random or systematic biases. Based on the identified high-risk periods and the residual deviation characteristics obtained in this analysis, the fitting parameters of the initial baseline curve were locally corrected for high-risk periods and monitoring periods with significant residual deviations. The slope, intercept, and other core fitting parameters of the curve were adjusted as needed to make the corrected curve more closely match the actual values ​​of the original valid data points during the deviation periods. After the correction was completed, an optimized iterative version of the baseline curve was generated.

[0033] Step 3.4: Based on the optimized benchmark curve iteration version, repeat the uncertainty assessment, residual analysis, and parameter correction process until the change in curve parameters for two consecutive iterations is less than the convergence threshold, and finally obtain the raw material end hygroscopic kinetic benchmark curve. Specifically, this includes: using the obtained optimized benchmark curve iteration version as the new processing object, repeating the expanded uncertainty assessment operation to obtain the expanded uncertainty distribution of the iteration version within the entire monitoring period and re-identifying high-risk periods; then repeating the residual calculation, residual distribution histogram construction, and deviation characteristic analysis operations; and combining the new high-risk periods and deviation characteristics to locally correct the fitting parameters of the iteration version again, generating a new benchmark curve iteration version. After completing one iteration of correction, all core fitting parameters of the two consecutive benchmark curve iterations are extracted, and the parameter changes between the two versions are calculated for each parameter. It is then determined whether the changes of all parameters are less than the preset convergence threshold of 0.001. If the parameter changes do not meet the convergence requirements, the complete process of uncertainty assessment, residual analysis, and parameter correction is repeated until the changes of all core fitting parameters between two consecutive benchmark curve iterations are less than 0.001. Once the convergence requirements are met, the iteration is stopped, and the final benchmark curve iteration is determined as the raw material moisture absorption kinetics benchmark curve. This curve can fit the actual moisture absorption law of the flame retardant curing agent raw material and fully reflect the moisture absorption kinetics change characteristics of the raw material throughout the entire monitoring period.

[0034] In this embodiment of the invention, because this preferred embodiment uses statistical fusion of multiple valid data points from the same monitoring time based on a representative valid dataset, calculates the mean, standard deviation, and confidence interval of the moisture absorption rate at each time point to obtain a time series of statistical characteristic values, obtains an initial baseline curve through curve fitting, evaluates the curve expansion uncertainty based on the confidence interval and identifies high-risk periods, calculates the residual value, analyzes the residual deviation characteristics, locally corrects the fitting parameters to obtain an iterative version of the baseline curve, and repeats the uncertainty evaluation, residual analysis, and parameter correction process until convergence, finally obtaining the raw material end moisture absorption kinetic baseline curve, this technical means overcomes the technical problems of existing raw material end moisture absorption baseline curve construction lacking a scientific statistical fusion and iterative optimization mechanism, not considering the influence of curve uncertainty and data residuals, resulting in insufficient accuracy and reliability of the baseline curve, failing to truly reflect the overall moisture absorption kinetic law of the raw material, and being difficult to use as a quantitative judgment basis for moisture absorption monitoring at each stage, thus achieving the construction of a reliable raw material end moisture absorption kinetic baseline curve that fits the actual moisture absorption characteristics of the raw material.

[0035] In a preferred embodiment of the present invention, step 4 above may include: Step 4.1: Based on the raw material end hygroscopic kinetics baseline curve, send real-time sampling commands to the online quality monitoring devices at the feed inlet, mixing chamber, and discharge outlet of the mixing reactor to initiate high-frequency monitoring of the mixing process and obtain the original quality time-series data stream including the start time of mixing. Specifically, this includes: using the established raw material end hygroscopic kinetics baseline curve as the quantitative basis for hygroscopic monitoring during the mixing stage; deploying one online quality monitoring device at each of the material feeding positions at the feed inlet of the mixing reactor, the stirring position in the middle of the mixing chamber, and the material discharge position at the discharge outlet; after all online quality monitoring devices have completed preliminary accuracy calibration, sending a unified real-time sampling command to the three devices through the data processing terminal to initiate flame-retardant curing. High-frequency moisture absorption monitoring is performed during the mixing process of the flame retardant curing agent and epoxy resin. The sampling interval is set to 1 minute to achieve continuous data acquisition throughout the mixing process. The moment when the flame retardant curing agent and epoxy resin are simultaneously added to the mixing tank is marked as the mixing start time. This moment serves as the time reference point for the original quality time-series data stream. Each online quality monitoring device collects the quality data of the mixed materials at the corresponding location according to the sampling instructions and transmits it back to the data processing terminal in real time. The terminal classifies and stores the transmitted data according to the monitoring location, forming an original quality time-series data stream that includes the mixing start time and continuous sampling times. The data record content includes the specific sampling time, monitoring location, and the material quality value at the corresponding time.

[0036] Step 4.2: Based on the original mass time-series data stream, environmental humidity compensation and correction are performed using data from the temperature and humidity sensors in the mixing environment to obtain a net moisture-absorbing mass change sequence that eliminates environmental interference. Specifically, this includes: deploying one temperature and humidity sensor around the perimeter of the mixing workshop and outside the mixing tank; ensuring that the sampling interval of all temperature and humidity sensors is consistent with the online quality monitoring device, which is 1 minute, to achieve synchronous acquisition of temperature and humidity data in the mixing environment and material mass data; retrieving the obtained original mass time-series data stream, and simultaneously retrieving the synchronous temperature and humidity data collected by the temperature and humidity sensors in the mixing environment; based on the variation pattern of environmental temperature and humidity, performing environmental humidity compensation and correction on the original mass time-series data stream; and making targeted corrections to the material mass data for different environmental temperature and relative humidity conditions to eliminate the interference of spurious changes in material mass caused by environmental water vapor condensation and temperature and humidity fluctuations, retaining only the true mass change data caused by moisture absorption. After correction, the sequence is organized according to the sampling time to form a net moisture-absorbing mass change sequence that eliminates environmental interference. This sequence includes the net moisture-absorbing mass change values ​​of the material at each continuous sampling time.

[0037] Step 4.3: Based on the corrected net moisture absorption mass change sequence, combined with the raw material mixing ratio and initial dry basis mass, calculate the instantaneous moisture absorption rate of the entire adhesive preparation process, and perform time normalization processing with the mixing start time as the zero point. Specifically, this includes: pre-determining the mixing ratio of flame retardant curing agent and epoxy resin in the adhesive preparation process as 1:4, measuring the initial dry basis mass of the flame retardant curing agent and the initial dry basis mass of the epoxy resin added to the adhesive preparation tank in advance, and calculating the total initial dry basis mass of the mixture; based on the obtained net moisture absorption mass change sequence, calculating the instantaneous moisture absorption rate corresponding to each sampling time in the entire adhesive preparation process according to the net moisture absorption mass change value of the mixture and the total initial dry basis mass; after completing the instantaneous moisture absorption rate calculation, using the mixing start time as the time zero point, converting all sampling times into relative time relative to this zero point, with the relative time measured in minutes, and organizing them in chronological order of relative time to form a sequence containing relative time and corresponding instantaneous moisture absorption rate, thus completing the time normalization processing of the moisture absorption rate data of the entire adhesive preparation process.

[0038] Step 4.4 involves comparing the normalized instantaneous moisture absorption rate sequence with the raw material end moisture absorption kinetics baseline curve on a time-by-time basis to identify instantaneous abrupt change intervals that deviate from the baseline by more than a dynamic threshold. Specifically, this includes: retrieving the time-normalized instantaneous moisture absorption rate sequence and simultaneously retrieving the raw material end moisture absorption kinetics baseline curve; matching the relative time in the normalized sequence with the time dimension of the baseline curve; extracting the actual instantaneous moisture absorption rate value of the normalized sequence and the predicted moisture absorption rate value of the corresponding moment on the baseline curve on a time-by-time basis; calculating the deviation between the two; pre-setting a dynamic threshold of 0.04; comparing the time-by-time calculated deviation value with this dynamic threshold; if the deviation value at a certain moment exceeds 0.04, and this exceeding state continues for multiple consecutive moments, then this consecutive time interval is identified as an instantaneous abrupt change interval; uniformly marking all identified instantaneous abrupt change intervals, with the marking information including the start relative time, end relative time, actual moisture absorption rate value at each moment within the interval, deviation value, and the corresponding predicted value of the baseline curve, ensuring that the location and moisture absorption change characteristics of the abrupt change interval can be clearly traced.

[0039] Step 4.5: Based on the marked instantaneous abrupt change intervals, extract the extreme value of the moisture absorption rate, the integral moisture absorption amount, and the duration. Combine this with the adhesive preparation process parameters to construct an instantaneous response feature vector characterizing the moisture absorption sensitivity. Specifically, this includes: for each marked instantaneous abrupt change interval, extracting key parameters of the moisture absorption kinetics for each interval, including the maximum and minimum values ​​of the moisture absorption rate (i.e., the extreme value of the moisture absorption rate), the cumulative value of the moisture absorption amount at all times within the interval (i.e., the integral moisture absorption amount), and the difference between the relative time at the end of the interval and the relative time at the beginning of the interval (i.e., the duration); simultaneously, retrieving the core process parameters for this adhesive preparation, including the mixing speed of the adhesive preparation kettle (300 rpm), the mixing temperature of the adhesive preparation process (28 degrees Celsius), and the material feeding rate (5 kg / min). The extracted key parameters of moisture absorption kinetics and the parameters of the compounding process are arranged and integrated in a fixed order. The parameter order is: maximum moisture absorption rate, minimum moisture absorption rate, integral moisture absorption, duration, mixing speed, mixing temperature, and feeding rate. In this order, an instantaneous response feature vector that can characterize the moisture absorption sensitivity of this compounding process is constructed to ensure that the parameter dimensions of the feature vector are fixed and the information is complete.

[0040] Step 4.6: By matching the instantaneous response feature vector with the historical qualified batch feature library, it is determined whether the current adhesive mixing moisture absorption behavior is within the process tolerance range, and a moisture absorption quality status identifier is obtained. Specifically, this includes: pre-constructing a historical qualified batch feature library for the adhesive mixing process, which contains the instantaneous response feature vectors of all past qualified adhesive mixing batches. The parameter dimensions of all vectors are consistent with the vector constructed this time, and the database supports real-time query and update; inputting the instantaneous response feature vector constructed in the current adhesive mixing process into the historical qualified batch feature library, calculating the similarity between this vector and the feature vectors of all qualified batches in the library, and calculating the average similarity between the current vector and the qualified batch vectors using the arithmetic mean method; setting the similarity threshold corresponding to the process tolerance to 90% in advance. If the average similarity calculated this time is higher than 90%, it is determined that the current adhesive mixing moisture absorption behavior is within the process tolerance range, and the generated moisture absorption quality status identifier is qualified; if the average similarity is lower than 90%, it is determined that the current adhesive mixing moisture absorption behavior exceeds the process tolerance range, and the generated moisture absorption quality status identifier is abnormal. The moisture absorption quality status label includes the specific judgment result, the average similarity value, and the number of qualified batches compared.

[0041] Step 4.7 involves spatiotemporally associating and encapsulating the quality status identifier, feature vector, mutation interval information, and instantaneous moisture absorption rate sequence to form instantaneous moisture absorption response data for the rubber compounding process. Specifically, this includes: the retrieved moisture absorption quality status identifier, the constructed instantaneous response feature vector, all information of the marked instantaneous mutation interval, and the instantaneous moisture absorption rate sequence after time normalization. Using the batch number of this rubber compounding as the core identifier, all the above data are spatiotemporally associating and encapsulating. The data is then linked and integrated according to the relative time dimension, ensuring that each relative time node corresponds to the instantaneous moisture absorption rate value, the baseline curve prediction value, and the deviation value. Simultaneously, the position of the instantaneous mutation interval is associated with the parameters of the feature vector. The moisture absorption quality status identifier is used as the overall judgment result of the moisture absorption monitoring for this rubber compounding batch, forming a unified whole with all detailed data. After the association and integration are completed, the data is stored in a standardized data format, forming instantaneous moisture absorption response data for the rubber compounding process that contains all information on moisture absorption monitoring throughout the entire rubber compounding process. This data enables one-click traceability and linked analysis of information at each stage.

[0042] In this embodiment of the invention, the preferred embodiment uses the raw material end moisture absorption kinetics baseline curve as a basis to send real-time sampling commands to the online quality monitoring devices at the feed inlet, mixing chamber, and discharge outlet of the mixing reactor to conduct high-frequency monitoring of the mixing process and acquire the original quality time-series data stream including the mixing start time. Environmental humidity compensation correction is performed using data from temperature and humidity sensors in the mixing environment to eliminate environmental interference. Then, the instantaneous moisture absorption rate of the entire mixing process is calculated based on the raw material mixing ratio and initial dry basis mass, and time normalization is performed. The normalized instantaneous moisture absorption rate sequence is compared with the raw material end baseline curve at each time step to identify instantaneous abrupt change intervals. Relevant parameters of the abrupt change intervals are extracted and combined with the mixing process parameters to construct an instantaneous response feature vector. This feature vector is then matched with a historical qualified batch feature library to determine the moisture absorption quality status of the mixed batch. Finally, the quality is... This technology, which encapsulates status identifiers, feature vectors, and other related data in a spatiotemporal manner to form instantaneous moisture absorption response data during the glue mixing process, overcomes the technical problems of existing glue mixing stages that only record ambient temperature and humidity without real-time moisture absorption monitoring, lack of environmental interference correction mechanisms, lack of quantitative comparison with the moisture absorption benchmark at the raw material end, inability to identify instantaneous moisture absorption mutations, lack of scientific moisture absorption quality judgment standards, and lack of standardized integration and encapsulation of glue mixing moisture absorption-related data. These problems lead to the inability to detect moisture absorption anomalies in a timely manner and the inability of glue mixing moisture absorption data to provide effective support for the process. This technology achieves online real-time moisture absorption monitoring at multiple locations, high frequency, and without environmental interference during the glue mixing process, identifies instantaneous moisture absorption mutation intervals during the glue mixing process, achieves scientific quantitative judgment of glue mixing moisture absorption behavior based on the feature database of historical qualified batches, and standardizes, integrates, and encapsulates instantaneous moisture absorption response data of the glue mixing process.

[0043] In a preferred embodiment of the present invention, step 5 above may include: Step 5.1: Based on the instantaneous moisture absorption response data during the adhesive mixing process, a synchronous sampling command is sent to the quality monitoring device and temperature monitoring device on the surface and inside the curing mold to initiate continuous monitoring of the mixed adhesive from the completion of adhesive mixing to the complete curing stage, obtaining quality time-series data streams and temperature time-series data streams. Specifically, this includes: using the obtained instantaneous moisture absorption response data during the adhesive mixing process as the basis for monitoring during the curing stage; deploying monitoring devices at key locations on the cable accessory curing mold; fixing one quality monitoring device and one temperature monitoring device at two symmetrical locations on the middle of the outer wall of the mold surface; and fixing one quality monitoring device and one temperature monitoring device at the center and edge locations of the adhesive after pouring inside the mold. All quality monitoring devices and temperature monitoring devices complete their functions. Preliminary precision calibration ensures accurate and reliable monitoring data. A synchronous sampling command is issued to all deployed quality and temperature monitoring devices via the data processing terminal, with a sampling interval of 2 minutes. Continuous monitoring of the mixed adhesive is initiated from the moment of adhesive preparation to the complete curing stage, with the moment of adhesive preparation serving as the start time for curing monitoring and a time reference point. Each monitoring device synchronously collects the mixed adhesive quality and temperature data at its corresponding location according to the sampling command, transmitting the data back to the data processing terminal in real time. The terminal categorizes and stores the transmitted data according to monitoring location and monitoring type, forming a quality time-series data stream containing continuous sampling time, monitoring location, and corresponding quality values, and a temperature time-series data stream containing continuous sampling time, monitoring location, and corresponding temperature values.

[0044] Step 5.2: Based on the mass time-series data stream and the temperature time-series data stream, alignment processing is performed according to a unified timestamp. The start and end times of the heating, isothermal, and cooling phases are identified, and the multi-stage time-series intervals of the curing reaction are divided. Specifically, this includes: retrieving the obtained mass and temperature time-series data streams, using the system time of the data processing terminal as a unified reference, and matching a unique unified timestamp to each data record in both data streams. The timestamp accuracy is set to 1 minute to ensure that each mass data and the corresponding temperature data have a consistent time identifier; performing time-by-time alignment processing on the two data streams based on the unified timestamp, removing invalid data without a corresponding timestamp, abnormal data with mismatched timestamps, and incomplete data with missing sampling times, to obtain aligned mass and temperature synchronization data. The process involves identifying the start and end times of each stage of the curing reaction, based on the preset requirements of the cable accessory curing process. The criteria for the heating stage are a continuous temperature increase of at least 0.5 degrees Celsius per minute. When the temperature reaches the preset curing temperature and remains stable, the heating stage ends and the isothermal stage begins. The criteria for the isothermal stage are temperature fluctuations controlled within ±1 degree Celsius. When the temperature begins to decrease continuously with a decrease of at least 0.3 degrees Celsius per minute, the isothermal stage ends and the cooling stage begins. The cooling stage continues until the temperature drops to room temperature and stops changing; at this point, complete curing is considered achieved, and the cooling stage ends. Based on the identified start and end times of each stage, multiple time intervals are divided into heating, isothermal, and cooling stages. The start and end times and duration of each time interval are recorded.

[0045] Step 5.3: Based on the multi-stage time intervals and the initial mass value at the completion of adhesive mixing, calculate the moisture absorption increment and rate for each stage, and label the corresponding curing temperature conditions for each stage. Based on the stage moisture absorption parameters and the instantaneous moisture absorption rate sequence, analyze the continuity of moisture absorption behavior from adhesive mixing to the start of curing, and identify the turning point of moisture absorption kinetics caused by the curing reaction. Specifically, this includes: retrieving the divided multi-stage time intervals of the curing reaction, and simultaneously retrieving the initial mass value of the mixed adhesive at the completion of adhesive mixing. The initial mass value is the corrected net moisture absorption mass value at the completion of adhesive mixing. For each time interval, calculate the moisture absorption increment and rate for that stage. The moisture absorption increment is the mass of the mixed adhesive at the end of the interval minus the mass of the mixed adhesive at the beginning of the interval, and the moisture absorption rate is the moisture absorption increment of the interval divided by the duration of the interval. Calculate and record for each stage. Simultaneously, collect the curing temperature data in each time interval in real time, calculate the average curing temperature for each stage, and use the average curing temperature as the average temperature for that stage. The corresponding curing temperature conditions are marked next to the moisture absorption parameters at each stage to establish a correlation between the moisture absorption parameters and the curing temperature. Based on the moisture absorption increment, moisture absorption rate, and other stage moisture absorption parameters calculated at each stage, the instantaneous moisture absorption rate sequence of the adhesive mixing process after normalization of the completion time is retrieved. The moisture absorption rate data of the last 10 minutes before the completion of adhesive mixing is extracted and compared with the moisture absorption rate data of the first 10 minutes before the heating stage of the curing stage. It is determined whether the deviation between the two is less than 0.02. If the deviation is less than 0.02, it is determined that the moisture absorption behavior from adhesive mixing to the start of curing has good continuity. If the deviation is greater than or equal to 0.02, it is determined that the continuity is poor. Combining the change law of moisture absorption rate at each stage, the moment when the change amplitude of moisture absorption rate exceeds 0.03 is identified as the turning point of moisture absorption kinetics caused by the curing reaction. The turning point is the moment when the curing reaction affects the moisture absorption characteristics. The specific time, corresponding temperature, and moisture absorption rate value of the turning point are recorded, which is the node of influence of the curing reaction on moisture absorption kinetics.

[0046] Step 5.4: Based on the identified inflection points, perform piecewise fitting of the moisture absorption rate and time series of the entire curing process, and extract the piecewise feature parameters for each stage. These feature parameters include the extreme value of the moisture absorption rate, the moisture saturation time, and the proportion of total moisture absorption. Specifically, this involves: using all identified moisture absorption kinetic inflection points as boundaries, dividing the moisture absorption rate and time series of the entire curing process into multiple continuous fitting intervals. The start and end times of each fitting interval are the times of two adjacent inflection points. If no inflection point exists, the entire curing process is considered as one fitting interval. For each fitting interval, plot the moisture absorption rate and time correlation data within the interval with time as the horizontal axis and moisture absorption rate as the vertical axis. Nonlinear fitting is performed to match the actual moisture absorption patterns of the mixed adhesive at each stage of curing, generating piecewise fitting curves for each fitting interval. Based on each piecewise fitting curve, piecewise feature parameters for each stage are extracted, where the extreme values ​​of moisture absorption rate include the maximum and minimum values ​​of moisture absorption rate within the fitting interval. Initial extreme points of moisture absorption rate are obtained by analyzing the geometric slope changes at each point on the curve. Then, a geometric profile extreme value determination algorithm is added for secondary verification. The moisture absorption rate sequence obtained by slope recursion is considered as a new geometric profile curve. For each initially selected extreme point, three consecutive geometric points before and after it are selected to construct a verification window, containing a total of seven consecutive points. For geometric contour points, analyze the changing trend of the contour curve window by window. If the contour curve within a window shows a monotonic trend, i.e., a continuous increasing trend from the start point to the initial extreme point and a continuous decreasing trend from the initial extreme point to the end point, then the initial extreme point is determined to be the maximum effective moisture absorption rate. If the contour curve within a window shows a monotonic decreasing trend, i.e., a continuous decreasing trend from the start point to the initial extreme point and a continuous increasing trend from the initial extreme point to the end point, then the initial extreme point is determined to be the minimum effective moisture absorption rate. If the contour curve within a window fluctuates, i.e., there is a change of first increasing then decreasing then increasing or first decreasing then increasing then decreasing, without a clear monotonic trend, then... If the initial extreme point is determined to be a false extreme point caused by local geometric protrusion, it will be removed to ensure that the extracted extreme value of the moisture absorption rate conforms to the true moisture absorption law of the curing stage. The moisture absorption saturation time is the moment when the moisture absorption rate drops below 0.001 and lasts for more than 5 minutes within the fitting interval, which is determined as the time when the moisture absorption of this stage reaches saturation. The proportion of total moisture absorption is the ratio of the moisture absorption increment within the fitting interval to the total moisture absorption increment of the entire curing process, which is obtained by the ratio of the moisture absorption increment of the stage to the sum of the moisture absorption increments of all stages. The segmented feature parameters are extracted for each interval and arranged in the order of the fitting interval to form a set of segmented feature parameters to ensure that the moisture absorption characteristics of each stage can be characterized.

[0047] Step 5.5: Based on the extracted segmented feature parameters and combined with curing temperature and ambient humidity conditions, construct a hygroscopic kinetic response matrix under different process parameter combinations. Based on the response matrix, perform deviation integral calculation with the raw material end hygroscopic kinetic baseline curve to quantify the correction effect of the curing reaction on the hygroscopic properties of the raw material, and obtain hygroscopic kinetic change data. Specifically, this includes: retrieving the extracted segmented feature parameters for each stage, and simultaneously collecting environmental data under different process parameter combinations during the curing process. The curing temperature is set with three gradients: 80°C, 100°C, and 120°C; the ambient humidity is set with three gradients: 50%, 60%, and 70%, forming nine different process parameter combinations. Construct a hygroscopic kinetic response matrix under different process parameter combinations, with each process parameter combination as a row and the segmented feature parameters for each stage as a column. Each element in the matrix corresponds to a specific process. The specific values ​​of a segmented characteristic parameter under a parameter combination are used to establish the correlation between process parameters and moisture absorption characteristic parameters. The obtained raw material end moisture absorption kinetics baseline curve is retrieved, and the moisture absorption characteristic parameters corresponding to each stage of the baseline curve are extracted. Each segmented characteristic parameter in the moisture absorption kinetic response matrix is ​​compared with the corresponding parameter of the baseline curve, and the deviation value between the two is calculated. Then, all deviation values ​​are integrated to obtain the integral value of the deviation of the curing reaction on the moisture absorption characteristics of the raw material under each process parameter combination. The magnitude of the integral value of the deviation is used to quantify the correction effect of the curing reaction on the moisture absorption characteristics of the raw material. The larger the integral value of the deviation, the more significant the correction effect of the curing reaction on the moisture absorption characteristics of the raw material. The integral values ​​of the deviation, the deviation of the segmented characteristic parameters, and the quantitative results of the correction effect under all process parameter combinations are compiled to finally form moisture absorption kinetic change data that reflects the moisture absorption change law of the curing process and the correction effect of the curing reaction.

[0048] In this embodiment of the invention, based on the instantaneous moisture absorption response data of the adhesive mixing process, synchronous sampling commands are sent to the quality monitoring device and temperature monitoring device on the surface and inside of the curing mold. This allows for continuous monitoring of temperature and humidity and moisture absorption of the mixed adhesive from the completion of mixing to the complete curing stage, acquiring time-series data streams of quality and temperature. Data is aligned using a unified timestamp and divided into multiple time-series intervals for the curing reaction (heating, isothermal, cooling). The moisture absorption parameters for each stage are calculated based on the initial mass value at the completion of adhesive mixing, and the continuity of moisture absorption behavior from mixing to the start of curing is analyzed. The moisture absorption kinetics inflection points triggered by the curing reaction are identified. Based on these inflection points, the moisture absorption rate and time series of the entire curing process are piecewise fitted, and characteristic parameters for each stage are extracted. Furthermore, a moisture absorption kinetic response matrix is ​​constructed under different process parameter combinations, combined with curing temperature and ambient humidity conditions. This matrix is ​​then compared with the raw material moisture absorption kinetics baseline curve using deviation integration to quantify the correction effect of the curing reaction on the moisture absorption characteristics of the raw material and to obtain the changes in moisture absorption kinetics. This data-driven approach overcomes the limitations of existing curing processes, which only track temperature changes without linking them to hygroscopic kinetic parameters, lack a synchronous monitoring mechanism for temperature and humidity, fail to analyze hygroscopic characteristics by dividing the curing reaction into stages, cannot identify hygroscopic kinetic inflection points triggered by the curing reaction, fail to establish a correlation between process parameters and hygroscopic characteristics, and cannot quantify the corrective effect of the curing reaction on the hygroscopic characteristics of raw materials. This results in the inability to capture the hygroscopic kinetic changes during the curing stage and a disconnect between hygroscopic data from the raw material and adhesive mixing stages. The new approach achieves synchronous and continuous monitoring of temperature and hygroscopic kinetics throughout the entire curing process, divides the curing reaction into stages and analyzes the hygroscopic characteristics of each stage, identifies hygroscopic kinetic inflection points triggered by the curing reaction, quantifies the corrective effect of the curing reaction on the hygroscopic characteristics of raw materials, constructs a correlation response matrix between curing process parameters and hygroscopic kinetics, and realizes full-process data linkage between hygroscopic monitoring during the curing stage and the raw material and adhesive mixing stages, capturing the hygroscopic kinetic changes throughout the entire curing process.

[0049] In a preferred embodiment of the present invention, step 6 above may include: Step 6.1: Based on the hygroscopic kinetic change data, issue low-frequency periodic sampling commands to the quality monitoring devices and environmental monitoring devices pre-embedded on the surface and inside the finished cable accessories. This initiates long-term monitoring of the finished product in the storage and transportation environment from the time of demolding. The goal is to obtain time-series data streams of finished product quality and environmental temperature and humidity. Specifically, the obtained hygroscopic kinetic change data is used as the basis for hygroscopic monitoring at the finished product stage. After the finished cable accessories are demolded, the deployment status of the monitoring devices on the surface and inside the finished product is immediately confirmed. Specifically, one quality monitoring device is fixed at each of the two end joints and the middle insulation layer on the surface of the finished product. Inside the finished product, one quality monitoring device is pre-embedded at the center of the insulation layer and at the junction of the core wire and the insulation layer. All pre-embedded quality monitoring devices are installed and calibrated during the curing stage. Simultaneously, one environmental temperature and humidity monitoring device is deployed in different areas of the finished product storage warehouse, in the cargo hold of the transport vehicle, and in the unloading temporary storage area to ensure coverage of the entire environment during finished product storage and transportation. The data processing terminal issues a unified low-frequency periodic sampling command to all quality monitoring devices on the surface and inside of finished products, as well as to each environmental temperature and humidity monitoring device. Considering the need for long-term monitoring of finished products, the sampling interval is set to 1 hour, and the duration of a single sampling is 1 minute. Long-term continuous monitoring of finished products is initiated from the moment of demolding and removal from the production line. The moment of demolding and removal from the production line is the starting moment of finished product monitoring and serves as the time reference point. Each quality monitoring device synchronously collects the quality data of the finished product at the corresponding location, and each environmental temperature and humidity monitoring device synchronously collects the temperature and relative humidity data of its area. All monitoring data are transmitted back to the data processing terminal in real time. The terminal classifies and stores the transmitted data according to monitoring type, finished product number, and monitoring location, forming a finished product quality time-series data stream containing continuous sampling time, finished product number, monitoring location, and corresponding quality value, and an environmental temperature and humidity time-series data stream containing continuous sampling time, monitoring area, and corresponding temperature and humidity value.

[0050] Step 6.2 involves aligning and fusing the finished product quality time-series data stream and the environmental temperature and humidity time-series data stream to identify environmental switching moments and divide the time-series into multiple scenarios, including warehousing, loading and transfer, unloading and temporary storage, and final warehousing. Specifically, this includes retrieving the finished product quality time-series data stream and the environmental temperature and humidity time-series data stream, using the system time of the data processing terminal as a unified benchmark, and matching a unique unified timestamp to each data record in both data streams. The timestamp accuracy is set to 1 minute to ensure that each piece of finished product quality data corresponds to the environmental temperature and humidity data at the same moment. Based on the unified time... The system performs time-by-time alignment on the two types of data streams, removing invalid data without corresponding timestamps, abnormal data with mismatched timestamps, and missing data caused by sampling signal interruptions, resulting in an aligned synchronized data stream of finished product quality and environmental temperature and humidity. Based on the actual process of finished product storage and transportation, environmental switching moments are identified. The criteria for environmental switching moments are a sudden change in relative humidity exceeding 5% for more than 10 minutes, or a significant change in the monitored area, such as when finished products are transferred from the warehouse to the transport vehicle, and the monitored area changes from the warehouse to the cargo hold. This is marked as an environmental switching moment. Based on all identified environmental switching moments, multiple time-series intervals for finished product storage and transportation are divided. The warehouse static interval is the period from when the finished product is demolded and enters the warehouse until it is loaded and transported. The loading and transportation interval is the period from when the finished product is loaded onto the vehicle until it is unloaded. The unloading and temporary storage interval is the period from when the finished product is unloaded until it is put into storage. The final storage interval is the period from when the finished product is put into storage until the end of monitoring. The start time, end time, duration, and corresponding monitoring area are recorded for each interval, along with the time range and environmental boundaries of each scenario.

[0051] Step 6.3: Based on multiple scenario intervals, calculate the cumulative moisture absorption increment and average moisture absorption rate for each interval, combined with the initial dry basis mass after demolding, to obtain moisture absorption parameters. Based on these parameters, analyze the continuity of moisture absorption from the completion of curing to the start of storage, identifying the equilibrium moment and equilibrium moisture absorption rate when the finished product reaches moisture absorption equilibrium. Specifically, this includes: retrieving the divided time-series intervals for finished product storage and transportation across multiple scenarios, and simultaneously determining the initial dry basis mass of the finished product at the demolding moment in advance. The determination method is as follows: immediately after the finished product is demolded and removed from the production line, select three samples of the same batch of finished products, dry them in a 105°C drying device until constant weight, and weigh them. Take the average mass of the three samples as the initial dry basis mass of the batch of finished products after demolding. For each scenario time-series interval, calculate the cumulative moisture absorption increment and average moisture absorption rate for that interval. The cumulative moisture absorption increment is the finished product mass at the end of the interval minus the finished product mass at the beginning of the interval, and the average moisture absorption rate is the cumulative moisture absorption increment divided by the duration of the interval. Calculations are completed and recorded for each scenario, and the results are labeled. The average ambient temperature and average relative humidity within each scenario range are recorded to establish a correlation between moisture absorption parameters and environmental conditions. Based on the cumulative moisture absorption increment and average moisture absorption rate calculated for each scenario, the moisture absorption kinetics data of the curing process are retrieved, and the moisture absorption rate data of the finished product at the moment of curing completion is extracted. This data is compared and analyzed with the average moisture absorption rate data of the first hour before the start of the finished product storage and resting period. The deviation between the two is calculated. If the deviation value is <0.01, the moisture absorption behavior from the completion of curing to the start of storage is considered to have good continuity. If the deviation value is ≥0.01, the continuity is considered to be poor, and the continuity analysis results are recorded. Combining the long-term monitoring of the moisture absorption rate change pattern of the finished product, the moment when the finished product's moisture absorption rate remains below 0.0005 for 8 consecutive hours without significant fluctuations is identified as the equilibrium moment when the finished product reaches a moisture absorption equilibrium state. The moisture absorption rate value of the finished product at this equilibrium moment is read as the equilibrium moisture absorption rate of the finished product, and the equilibrium moment, equilibrium moisture absorption rate, and corresponding environmental conditions are recorded.

[0052] Step 6.4: Based on the equilibrium time and equilibrium moisture absorption rate, perform asymptotic fitting on the moisture absorption rate and time series during the long-term storage period to extract the moisture absorption characteristic parameters of the finished product. Specifically, this includes: using the identified moisture absorption equilibrium time and equilibrium moisture absorption rate of the finished product as the core basis, retrieving the moisture absorption rate and time series data of the finished product during the long-term storage period. The long-term storage period specifically refers to the time from when the finished product reaches the moisture absorption equilibrium time to the end of monitoring. Extract the finished product moisture absorption rate values ​​and corresponding time data for all sampling times within this period, and remove abnormal data points with data fluctuations exceeding 0.0001 to ensure data stability; using time as the horizontal axis and moisture absorption rate as the vertical axis, perform asymptotic fitting on the moisture absorption rate and time-related data within this period, ensuring that the fitting process fully matches the finished product's equilibrium time. The actual law that the moisture absorption rate tends to stabilize after moisture absorption equilibrium is used to generate an asymptotic curve of the moisture absorption rate change during long-term storage. The asymptotic curve can reflect the long-term trend of the finished product after moisture absorption equilibrium. Based on this asymptotic curve, the moisture absorption characteristic parameters of the finished product are extracted, including the finished product equilibrium moisture absorption rate, the moisture absorption equilibrium rate, and the long-term moisture absorption decay coefficient. The moisture absorption equilibrium rate is the average moisture absorption rate of the finished product after reaching equilibrium. The long-term moisture absorption decay coefficient is obtained by analyzing the slope change of the asymptotic curve and is used to characterize the stability of the moisture absorption rate of the finished product after equilibrium. The moisture absorption characteristic parameters of the finished product are extracted batch by batch and organized by finished product number to form a set of finished product moisture absorption characteristic parameters, ensuring that the long-term moisture absorption characteristics of each finished product can be characterized.

[0053] Step 6.5: Based on the finished product's moisture absorption characteristic parameters and combined with multiple environmental conditions, establish a quantitative relationship between exposure duration and cumulative moisture absorption in different environments. Quantify the contribution of each environmental stage to the final moisture absorption state of the finished product, and perform a full-cycle chain comparison with moisture absorption data from the raw material end and the curing process to generate behavioral data on the long-term moisture absorption stability of the finished product. Specifically, this includes: retrieving the extracted finished product's moisture absorption characteristic parameters, and simultaneously retrieving the data from the divided time-series intervals and corresponding environmental conditions for each scenario, including key information such as environmental temperature and humidity, and exposure duration for each scenario. The relative humidity control in the warehouse static storage scenario is also included. The relative humidity is controlled at around 55% in loading and transshipment scenarios, fluctuating between 50% and 65%; in unloading and temporary storage scenarios, it is controlled at around 60%; and in the final warehousing scenario, it is controlled at around 55%. Based on the environmental conditions and exposure duration of each scenario, combined with the moisture absorption characteristics of the finished product, a quantitative relationship between the exposure duration of different environments and the cumulative moisture absorption of the finished product is calculated. Specifically, the cumulative moisture absorption of the finished product within the exposure duration of each scenario is calculated, the variation law of the cumulative moisture absorption under different exposure durations is analyzed, and a correlation model between exposure duration and cumulative moisture absorption is established. By calculating the proportion of the cumulative moisture absorption in each scenario to the total cumulative moisture absorption of the finished product, the contribution of each environmental stage to the final moisture absorption state of the finished product is quantified. For example, the contribution ratios of the warehousing static scenario, loading and transshipment scenario, unloading and temporary storage scenario, and final warehousing scenario are calculated and recorded separately. Simultaneously, the obtained raw material end moisture absorption kinetics baseline curve and curing process moisture absorption kinetics change data are retrieved. The finished product end moisture absorption characteristic parameters, moisture absorption parameters in each scenario, and raw material end baseline curve and curing process moisture absorption data are compared in a full-cycle chain. Specifically, this includes comparing the finished product equilibrium moisture absorption rate with the corresponding equilibrium value of the raw material end baseline curve and the predicted equilibrium value of the curing process, and comparing the moisture absorption rate of the finished product at each stage with the corresponding moisture absorption rates at the raw material end and the curing end. If the deviation value of all comparison items is <0.02, the finished product is judged to have good long-term moisture absorption stability. If there are items with a deviation value ≥0.02, they are marked as abnormal and the cause is analyzed. All information such as finished product end moisture absorption characteristic parameters, environmental conditions in each scenario, quantitative relationship between exposure time and moisture accumulation, contribution degree of each environmental stage, and full-cycle chain comparison results are integrated to finally generate behavioral data that can accurately reflect the long-term moisture absorption stability of the finished product.

[0054] In this embodiment of the invention, because this preferred embodiment uses the moisture absorption kinetics change data during the curing stage as a basis, it sends low-frequency periodic sampling commands to the quality monitoring devices and environmental monitoring devices embedded on the surface and inside of the finished cable accessories, initiates long-term monitoring of the finished product storage and transportation process, and obtains time-series data streams of finished product quality and environmental temperature and humidity. It performs time alignment and fusion processing on the two types of data streams, identifies environmental switching moments and divides multiple scenario time-series intervals, calculates moisture absorption parameters for each scenario based on the initial dry base mass after demolding, analyzes the moisture absorption continuity from curing to the start of storage, and identifies the finished product's moisture absorption equilibrium moment and equilibrium moisture absorption rate. Based on equilibrium-related parameters, it performs asymptotic fitting on the moisture absorption rate and time series of the finished product during long-term storage and extracts finished product-end moisture absorption characteristic parameters. Furthermore, it establishes a quantitative relationship between environmental exposure duration and cumulative moisture absorption, quantifies the contribution of each environmental stage to the finished product's moisture absorption state, and conducts a full-cycle chain comparison between the finished product-end moisture absorption data and the raw material-end and curing process moisture absorption data to generate long-term moisture absorption stability behavior data of the finished product. Therefore, it overcomes the limitations of existing finished product storage and transportation processes. The existing technical problems include: lack of continuous moisture absorption tracking measures; failure to combine finished product moisture absorption monitoring with curing stage data; lack of environmental scenario segmentation analysis mechanism; inability to identify the finished product's moisture absorption equilibrium state; lack of quantitative correlation between environmental conditions and finished product moisture absorption; and lack of chain comparison of moisture absorption data throughout the entire life cycle of raw materials, curing, and finished products. These issues lead to the inability to track changes in finished product moisture absorption, quantify the impact of each environmental stage on finished product moisture absorption, and disconnected moisture absorption data throughout the entire life cycle, making it difficult to assess the long-term moisture absorption stability of finished products. This solution aims to achieve synchronous long-term monitoring of moisture absorption and environment during finished product storage and transportation. It divides the time-series intervals of finished product storage and transportation into multiple scenarios and quantifies the contribution of each environmental stage to finished product moisture absorption. It identifies the finished product's moisture absorption equilibrium state and extracts core moisture absorption characteristic parameters from the finished product end. It completes chain comparison and linkage analysis of moisture absorption data throughout the entire life cycle of flame retardant curing agents from raw materials to finished products, generates behavioral data that reflects the long-term moisture absorption stability of finished products, grasps the moisture absorption change patterns of finished products, provides full-cycle moisture absorption data support for the quality control of finished cable accessories, assesses the moisture absorption status of finished products, and avoids quality defects such as insulation and structure caused by excessive moisture absorption.

[0055] like Figure 2 As shown, embodiments of the present invention also provide a full-cycle monitoring system for the hygroscopic kinetics of electrical flame retardant curing agents, comprising: The acquisition module is used to select samples from different locations of three barrels of flame retardant curing agent raw materials from the same flame retardant curing agent raw material, perform synchronous periodic monitoring, obtain moisture absorption rate data of the samples, and construct a moisture absorption characteristic calibration model. The screening module is used to integrate the obtained synchronous monitoring data according to the hygroscopic characteristics calibration model, and through consistency analysis and outlier removal, screen out a set of effective data points that stably reflect the overall hygroscopic characteristics of the raw materials to form a representative effective dataset. The optimization module is used to fuse representative and valid datasets, calculate the statistical characteristic values ​​of the data at each time point, and obtain the initial baseline curve. By analyzing the uncertainty of the initial baseline curve and the residuals with the original data, the curve is verified and optimized, and finally the raw material end moisture absorption kinetic baseline curve is obtained. The monitoring module is used to perform online real-time monitoring during the mixing and preparation of flame retardant curing agent and epoxy resin based on the moisture absorption kinetics baseline curve at the raw material end, and record the instantaneous moisture absorption response data during the glue preparation process. The recording module is used to monitor the mixed adhesive in real time during the entire crosslinking and curing process based on the instantaneous moisture absorption response data during the adhesive preparation process, and record the moisture absorption kinetics change data under different curing conditions. The processing module is used to continuously monitor the cured cable accessories during storage and transportation based on the hygroscopic kinetic changes during the curing reaction process, and obtain long-term hygroscopic behavior data of the finished product.

[0056] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for full-cycle monitoring of the hygroscopic kinetics of electrical flame-retardant curing agents, characterized in that, The method includes: Step 1: Select samples from different locations in three barrels of the same flame retardant curing agent raw material, perform synchronous periodic monitoring, obtain the moisture absorption rate data of the samples, and construct a moisture absorption characteristic calibration model. Step 2: Based on the hygroscopic characteristics calibration model, integrate the obtained synchronous monitoring data, and through consistency analysis and outlier removal, screen out a set of effective data points that stably reflect the overall hygroscopic characteristics of the raw materials to form a representative effective dataset; Step 3: Merge representative valid datasets, calculate statistical characteristic values ​​of data at each time point, and obtain an initial baseline curve; verify and optimize the curve by analyzing the uncertainty of the initial baseline curve and the residuals with the original data, and finally obtain the raw material end moisture absorption kinetic baseline curve. Step 4: Based on the moisture absorption kinetics baseline curve at the raw material end, conduct online real-time monitoring during the mixing and preparation of flame retardant curing agent and epoxy resin, and record the instantaneous moisture absorption response data during the glue preparation process. Step 5: Based on the instantaneous moisture absorption response data during the adhesive mixing process, monitor the mixed adhesive solution in real time throughout the crosslinking and curing reaction process, and record the moisture absorption kinetics change data under different curing conditions. Step 6: Based on the hygroscopic kinetics data during the curing reaction, continuously monitor the cured cable accessories during storage and transportation to obtain long-term hygroscopic behavior data of the finished product.

2. The method for full-cycle monitoring of the hygroscopic kinetics of electrical flame retardant curing agents according to claim 1, characterized in that, Samples from different locations in three barrels of the same flame retardant curing agent raw material were selected and simultaneously and periodically monitored to obtain the moisture absorption rate data of the samples. A calibration model for the moisture absorption characteristics was then constructed, including: Synchronous sampling instructions are sent to the distributed humidity sensing nodes deployed at the top, middle and bottom of the three barrels of flame retardant curing agent raw materials to start periodic quality monitoring of the three barrels of samples under the same environmental conditions and obtain the original quality time series data returned by each sensing node. Based on the original mass time series data, time axis alignment was performed according to a unified timestamp to form three sets of synchronous mass change sequences with the same time reference. Based on the three sets of synchronous mass change sequences, normalization calculation was performed in combination with the initial dry basis mass value of each barrel to obtain the moisture absorption rate time series data of the three barrel samples at each monitoring time. Based on the time series data of moisture absorption rate, a horizontal consistency comparison was performed among the three barrel samples to identify and mark abnormal data segments that deviated from the average of the three barrels by more than a preset threshold. Based on the identified and marked time series data of moisture absorption rate, abnormal data segments were removed, and the remaining effective moisture absorption rate data was fused between barrels to extract the key kinetic parameters of the overall moisture absorption characteristics of the raw material. Based on key kinetic parameters, a calibration model for the moisture absorption characteristics of flame retardant curing agents was established by fitting and establishing the mapping relationship between the moisture absorption rate and the ambient humidity.

3. The method for full-cycle monitoring of the hygroscopic kinetics of electrical flame retardant curing agents according to claim 2, characterized in that, Based on the hygroscopic characteristic calibration model, the obtained synchronous monitoring data were integrated. Through consistency analysis and outlier removal, a set of effective data points that stably reflect the overall hygroscopic characteristics of the raw material was selected, forming a representative and effective dataset, including: Using a moisture absorption characteristic calibration model, the time series data of moisture absorption rate of three barrel samples are input into the moisture absorption characteristic calibration model to calculate the deviation sequence between each sample data point and the model prediction value; based on the deviation sequence, the consistency analysis of the deviation of the three barrel samples at the same time is performed, and the standard deviation of the deviation values ​​of the three barrel samples at each time point is calculated. By statistically obtaining the standard deviation at each time point, the time point where the standard deviation exceeds the preset tolerance range of the moisture absorption characteristic calibration model is identified as an inconsistent time point, and all data points at the corresponding time point are marked. Based on the data points at the time of the consistency anomaly based on the label, combined with the identified abnormal data segments, a dual anomaly judgment is performed, and data points that simultaneously meet the conditions of excessive deviation and excessive inter-bucket dispersion are identified as outliers. Based on the outlier list, all data records corresponding to outliers were removed from the time series data of moisture absorption rate of the three samples, and the remaining valid data points were retained. Based on the retained valid data points, they were aggregated according to the monitoring time, and the weighted average of the valid data points of multiple buckets at each time was calculated to obtain the valid dataset.

4. The method for full-cycle monitoring of the hygroscopic kinetics of electrical flame retardant curing agents according to claim 3, characterized in that, By fusing representative and valid datasets and calculating the statistical characteristic values ​​of the data at each time point, an initial baseline curve is obtained. By analyzing the uncertainty of the initial baseline curve and the residuals between it and the original data, the curve is verified and optimized, ultimately yielding the baseline curve of moisture absorption kinetics at the raw material end, including: Based on representative and valid datasets, statistical fusion processing is performed on multiple valid data points at the same monitoring time to calculate the mean, standard deviation, and confidence interval of the moisture absorption rate at each time point, thereby obtaining a time series of statistical characteristic values. Based on time-series data, an initial baseline curve of the raw material's moisture absorption kinetics trend is obtained by curve fitting the relationship between moisture absorption rate and time. Based on the initial baseline curve and the confidence interval at each time point, the expanded uncertainty distribution of the initial baseline curve over the entire monitoring period is evaluated, and high-risk periods where uncertainty exceeds a preset threshold are identified. Based on the initial baseline curve and the original valid data points, the residual values ​​of each data point relative to the curve are calculated, a residual distribution histogram is constructed, and the deviation characteristics of the residuals are analyzed. Taking into account the high-risk period and the deviation characteristics, the fitting parameters of the initial baseline curve in the deviation period are locally corrected to obtain an optimized iterative version of the baseline curve. Based on the optimized benchmark curve iteration, the uncertainty assessment, residual analysis and parameter correction process is repeated until the change in curve parameters in two consecutive iterations is less than the convergence threshold, and finally the raw material end hygroscopic kinetic benchmark curve is obtained.

5. The method for full-cycle monitoring of the hygroscopic kinetics of electrical flame retardant curing agents according to claim 4, characterized in that, Step 4 includes: Based on the raw material end hygroscopic kinetics baseline curve, a real-time sampling command is sent to the online quality monitoring devices at the feed inlet, mixing chamber and discharge outlet of the mixing tank to start high-frequency monitoring of the mixing process and obtain the original quality time-series data stream including the mixing start time; Based on the original mass time-series data stream, environmental humidity compensation and correction are performed by combining the temperature and humidity sensor data of the glue mixing environment to obtain the net moisture absorption mass change sequence that eliminates environmental interference. Based on the corrected net moisture absorption mass change sequence, combined with the raw material mixing ratio and initial dry basis mass, the instantaneous moisture absorption rate of the entire rubber mixing process is calculated, and time normalization is performed with the mixing start time as zero point. By comparing the normalized instantaneous moisture absorption rate sequence with the raw material end moisture absorption kinetics baseline curve at each time step, the instantaneous abrupt change intervals that deviate from the baseline by more than the dynamic threshold are identified. Based on the marked instantaneous abrupt change intervals, the extreme values ​​of moisture absorption rate, integral moisture absorption amount, and duration are extracted, and an instantaneous response feature vector characterizing moisture absorption sensitivity is constructed by combining the rubber compounding process parameters. By matching the instantaneous response feature vector with the feature library of historical qualified batches, it is determined whether the current moisture absorption behavior of the adhesive is within the process tolerance range, and the moisture absorption quality status is identified. The quality status identifier, feature vector, mutation interval information, and instantaneous moisture absorption rate sequence are spatiotemporally correlated and encapsulated to form instantaneous moisture absorption response data during the glue mixing process.

6. The method for full-cycle monitoring of the hygroscopic kinetics of electrical flame retardant curing agents according to claim 5, characterized in that, Step 5 includes: Based on the instantaneous moisture absorption response data during the mixing process, a synchronous sampling command is sent to the quality monitoring device and temperature monitoring device on the surface and inside of the curing mold to start continuous monitoring of the mixed adhesive from the completion of mixing to the complete curing stage, and obtain the quality time-series data stream and temperature time-series data stream; Based on the quality time-series data stream and the temperature time-series data stream, alignment processing is performed according to a unified timestamp to identify the start and end times of the heating, isothermal, and cooling stages, and to divide the multi-stage time-series intervals of the curing reaction. Based on the multi-stage time interval, the moisture absorption increment and moisture absorption rate of each stage are calculated in combination with the initial mass value at the time of adhesive preparation completion, and the curing temperature conditions corresponding to each stage are marked; based on the stage moisture absorption parameters, combined with the instantaneous moisture absorption rate sequence, the continuity of moisture absorption behavior from adhesive preparation to the start of curing is analyzed, and the turning point of moisture absorption kinetics caused by the curing reaction is identified. Based on the identified turning points, the moisture absorption rate and time series of the entire curing process are segmented and fitted, and segmented feature parameters of each stage are extracted. The segmented feature parameters include the extreme value of moisture absorption rate, moisture absorption saturation time, and the proportion of total moisture absorption. Based on the extracted segmented feature parameters, and combined with curing temperature and ambient humidity conditions, a moisture absorption kinetic response matrix under different process parameter combinations is constructed. Based on the response matrix, the deviation integral is calculated with the raw material end moisture absorption kinetic baseline curve to quantify the correction effect of the curing reaction on the moisture absorption characteristics of the raw material, and the moisture absorption kinetic change data are obtained.

7. The method for full-cycle monitoring of the hygroscopic kinetics of electrical flame retardant curing agents according to claim 6, characterized in that, Step 6 includes: Based on the hygroscopic dynamics change data, low-frequency periodic sampling commands are sent to the quality monitoring devices and environmental monitoring devices embedded on the surface and inside of the finished cable accessories to initiate long-term monitoring of the finished product in the storage and transportation environment from the time of demolding and production line, and obtain the time-series data stream of finished product quality and environmental temperature and humidity. The time-series data streams of finished product quality and environmental temperature and humidity are time-aligned and fused to identify environmental switching moments and divide the time-series intervals of multiple scenarios, including warehouse static storage, loading and transfer, unloading and temporary storage, and final warehousing. Based on multiple scenario intervals, the cumulative moisture absorption increment and average moisture absorption rate of each interval are calculated in combination with the initial dry base mass after demolding to obtain moisture absorption parameters; based on the moisture absorption parameters, the moisture absorption continuity from the completion of curing to the start of storage is analyzed, and the equilibrium time and equilibrium moisture absorption rate of the finished product when it reaches the moisture absorption equilibrium state are identified. Based on the equilibrium time and equilibrium moisture absorption rate, asymptotic fitting was performed on the moisture absorption rate and time series during the long-term storage stage to extract the moisture absorption characteristic parameters of the finished product. Based on the moisture absorption characteristic parameters of the finished product, and combined with environmental conditions in multiple scenarios, a quantitative relationship between the exposure time of different environments and the cumulative amount of moisture absorption is established. The contribution of each environmental stage to the final moisture absorption state of the finished product is quantified, and a full-cycle chain comparison is performed with the moisture absorption data of the raw material end and the curing process to generate behavioral data on the long-term moisture absorption stability of the finished product.

8. A full-cycle monitoring system for the hygroscopic kinetics of flame-retardant curing agents for electrical applications, wherein the system implements the method as described in any one of claims 1 to 7, characterized in that, include: The acquisition module is used to select samples from different locations of three barrels of flame retardant curing agent raw materials from the same flame retardant curing agent raw material, perform synchronous periodic monitoring, obtain moisture absorption rate data of the samples, and construct a moisture absorption characteristic calibration model. The screening module is used to integrate the obtained synchronous monitoring data according to the hygroscopic characteristics calibration model, and through consistency analysis and outlier removal, screen out a set of effective data points that stably reflect the overall hygroscopic characteristics of the raw materials to form a representative effective dataset. The optimization module is used to fuse representative and valid datasets, calculate the statistical characteristic values ​​of the data at each time point, and obtain the initial baseline curve. By analyzing the uncertainty of the initial baseline curve and the residuals with the original data, the curve is verified and optimized, and finally the raw material end moisture absorption kinetic baseline curve is obtained. The monitoring module is used to perform online real-time monitoring during the mixing and preparation of flame retardant curing agent and epoxy resin based on the moisture absorption kinetics baseline curve at the raw material end, and record the instantaneous moisture absorption response data during the glue preparation process. The recording module is used to monitor the mixed adhesive in real time during the entire crosslinking and curing process based on the instantaneous moisture absorption response data during the adhesive preparation process, and record the moisture absorption kinetics change data under different curing conditions. The processing module is used to continuously monitor the cured cable accessories during storage and transportation based on the hygroscopic kinetic changes during the curing reaction process, and obtain long-term hygroscopic behavior data of the finished product.

9. A computing device, characterized in that, include: One or more processors; A storage device for storing one or more programs, which, when executed by one or more processors, cause the one or more processors to perform the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a program that, when executed by a processor, implements the method as described in any one of claims 1 to 7.

Images