A testing method and system for epoxy molding compounds based on production status monitoring

Abstract

This invention provides a testing method and system for epoxy molding compounds based on production status monitoring, belonging to the field of materials mechanics testing technology. The method includes: determining whether sampling is triggered based on a state vector S; sending a powder sample into a micro-cylinder testing chamber; isothermally locking the powder sample in the micro-cylinder testing chamber to obtain corresponding force-displacement curves, constant load relaxation curves, and unloading recovery curves; determining corresponding mechanical property indicators based on the force-displacement curves, constant load relaxation curves, and unloading recovery curves; and combining the mechanical property indicators with the state vector S to perform threshold adaptive regression to generate the test results for the current batch of epoxy molding compounds. This invention combines production status monitoring data with mechanical testing results to achieve accurate batch quality determination and adaptive optimization of process parameters for epoxy molding compounds.

Description

Technical Field

[0001] This invention relates to the field of materials mechanics testing technology, specifically to a method and system for testing epoxy molding compounds based on production status monitoring. Background Technology

[0002] Current epoxy molding compound testing primarily focuses on offline laboratory evaluations, such as melt flow index, spiral flow, gel time, differential scanning calorimetry (DSC), and post-cured hardness / strength testing. These methods mainly obtain "post-cured" or "high-shear rheological" indicators, which are insufficient to characterize mechanical behaviors more closely related to mold filling, such as short-term isothermal compaction before molding, stress relaxation, and unloading rebound. This results in insufficient predictive power for defects such as line scan defects, voids, and flash. Furthermore, offline testing has long cycles and delayed feedback, making it difficult to keep pace with production line batch changes and formulation fine-tuning.

[0003] On the other hand, issues of sample representativeness and consistency in judgment are common in production sites. Epoxy molding compounds are significantly affected by temperature and humidity fluctuations, negative pressure and mass flow rate fluctuations, static electricity accumulation, and historical shearing during vacuum conveying and drying. Traditional sampling methods at fixed times or in fixed quantities easily result in "dry shell with wet core," "orientation accumulation," or particle size stratification, thus distorting the compaction and rebound curves. Judgment thresholds are mostly empirical constants that are not adaptively adjusted according to batch conditions, leading to poor comparability of the same indicator across shifts and equipment. Existing countermeasures typically only strengthen environmental monitoring or increase sampling frequency, failing to simultaneously ensure sample representativeness and threshold stability within the same testing process. This results in a disconnect between test results and actual molding behavior, and low correction efficiency. Summary of the Invention

[0004] The purpose of this invention is to provide a testing method and system for epoxy molding compounds based on production status monitoring, so as to at least solve the problems of inconsistent cross-batch judgments caused by the lag in the evaluation of mechanical behavior before molding and insufficient sampling representativeness in existing solutions.

[0005] To achieve the above objectives, the first aspect of the present invention provides a method for testing epoxy molding compounds based on production status monitoring. The method includes: forming a state vector S based on multi-point status monitoring data collected from the epoxy molding compound conveying and pretreatment stages; determining whether sampling is triggered based on the state vector S; when sampling is triggered by the state vector S, obtaining a powder sample by isokinetic sampling from the material conveying path, and sending the powder sample into a micro-cylinder test chamber; isothermally locking the powder sample in the micro-cylinder test chamber, and sequentially performing mechanical tests of segmented loading, constant load relaxation, and isokinetic unloading based on the micro-cylinder test chamber to obtain corresponding force-displacement curves, constant load relaxation curves, and unloading recovery curves; determining corresponding mechanical property indicators based on the force-displacement curves, constant load relaxation curves, and unloading recovery curves, and combining the mechanical property indicators with the state vector S to perform threshold adaptive regression to generate the test results of the current batch of epoxy molding compounds.

[0006] Optionally, the state monitoring data includes at least one measurement data collected by a temperature sensor, a humidity sensor, a static pressure differential sensor, and a mass flow meter installed at the epoxy molding compound conveying main pipe, the feeding port, and the drying outlet; before forming the state vector S, the method further includes: performing time synchronization marking and noise filtering on the measurement data so that the state vector S corresponds to the mechanical property index in the time dimension when triggering sampling and threshold adaptive regression.

[0007] Optionally, a state vector S is formed based on the multi-point status monitoring data of the collected epoxy molding compound conveying and pre-processing links. This includes: performing time synchronization and interpolation processing on the measurement data collected from each measuring point to ensure that the measurement data from different measuring points correspond one-to-one within the same sampling period; spatially sorting the time-synchronized measurement data according to the measuring point location and material conveying direction, and calculating the difference features and rate of change features of adjacent measuring point measurement data; and concatenating the measurement data, difference features, and rate of change features of each measuring point into a multi-dimensional array in a preset order, and using a normalization algorithm to map each dimension to a unified numerical range to obtain the state vector S used to trigger sampling and threshold adaptive regression.

[0008] Optionally, the rule for determining whether to trigger sampling based on the state vector S is as follows: the measurement data of each dimension in the state vector S are compared one by one with the pre-calibrated stable interval threshold. When the measurement data of any dimension deviates from the corresponding stable interval by more than a preset deviation percentage, or the rate of change within a continuous sampling period exceeds a preset rate of change threshold, sampling is determined to be triggered. The preset deviation percentage and rate of change threshold are dynamically updated according to the distribution of mechanical property indicators of historical batches and are stored synchronously with the state vector S.

[0009] Optionally, when the state vector S triggers sampling, a powder sample is obtained by isokinetic sampling from the material conveying path, and the powder sample is sent into the micro-cylinder test chamber. This includes: sampling the powder sample through a bypass sampling branch connected to the feed end of the micro-cylinder test chamber on the main pipeline of the material conveying path; wherein the sampling branch is equipped with a differential pressure self-compensation device for automatically adjusting the bypass flow rate according to the flow rate change of the main pipeline to maintain a stable Reynolds number, so as to adjust the flow rate in real time during the sampling process using the differential pressure self-compensation device; powder sample accumulation is performed at the end of the sampling branch, and when the accumulated mass reaches the preset sampling amount, sampling is stopped and the sample is sent into the micro-cylinder test chamber through a one-way introduction step.

[0010] Optionally, the rules for obtaining the force-displacement curve are as follows: After the powder sample is isothermally locked in the micro-cylinder test chamber, a pre-loading stabilization phase is executed. The indenter of the micro-cylinder test chamber is advanced to a preset proportional range of the target loading stroke at a pre-loading rate lower than the target loading rate, and the acoustic emission event density and displacement fluctuation amplitude are detected in real time. When both meet the preset stability criteria, the main loading phase is switched. During the main loading phase, the force and displacement time series data are subjected to first-order and second-order differences. When a joint event occurs in which the first-order difference sign changes continuously and the second-order difference sign flips, accompanied by the acoustic emission sequence boundary crossing, it is marked as a densification initiation event. The subdivision sampling phase is activated from the densification initiation event until the absolute value of the first-order displacement difference returns to the first adaptive boundary. The measurement points of the pre-loading stabilization phase and the subdivision sampling phase are spliced ​​together in chronological order to form the force-displacement curve.

[0011] Optionally, the rules for obtaining the constant load relaxation curve are as follows: after the powder sample is compacted to the density mark, switch to constant load control; when the load error enters the second adaptive boundary band constructed by the moving median and its absolute deviation, execute the recording pre-start; based on the recording pre-start, apply a test displacement step and determine whether there is a reaction intervention according to the transient stress response of the test displacement step. If it is determined to be a reaction intervention, perform segmented isothermal treatment and start recording after the load error and temperature shift return to their respective boundary bands simultaneously; during the recording period, perform thermal drift correction according to the cavity temperature sequence and remove the initial stage of constant load establishment and mechanical limit segment, retaining only the continuous measurement points that meet the stability criteria to form the constant load relaxation curve.

[0012] Optionally, the rules for obtaining the unloading recovery curve are as follows: after constant load relaxation is completed and the load error falls into the second adaptive boundary band, switch to constant speed unloading; determine the contact decoupling time by jointly judging the first adaptive boundary band using the first-order differential correlation coefficient of load displacement and the surface lateral expansion rate; continuously collect the original sequence of rebound displacement and time from the contact decoupling time, and simultaneously record the acoustic emission event density sequence; remove the intervals where the acoustic emission event density exceeds the first adaptive boundary band from the original sequence, and splice the remaining effective segments in chronological order to form an effective sequence of rebound displacement and time; construct a third adaptive boundary band for the displacement velocity of the effective sequence, and take the moment when the displacement velocity enters and remains within the third adaptive boundary band as the rebound termination point; construct the unloading recovery curve based on the effective sequence between the contact decoupling time and the rebound termination point.

[0013] Optionally, the corresponding mechanical property indices are determined based on the force-displacement curve, the constant load relaxation curve, and the unloading recovery curve. These mechanical property indices are then combined with the state vector S to perform threshold adaptive regression, generating the test results for the current batch of epoxy molding compound. This includes: identifying the yield point and densification slope based on the force-displacement curve; fitting an equivalent relaxation time constant based on the constant load relaxation curve; calculating the elastic recovery rate and hysteresis energy ratio based on the unloading recovery curve; forming a set of mechanical property indices based on the yield point, densification slope, equivalent relaxation time constant, elastic recovery rate, and hysteresis energy ratio; normalizing the set of mechanical property indices and forming a joint feature set with the state vector S; performing threshold adaptive regression using the joint feature set as input; and updating the judgment threshold based on historical batch calibration data; and generating the test results for the current batch of epoxy molding compound based on the boundary relationships of the set of mechanical property indices relative to the updated judgment threshold. The test results include at least the judgment level and the corresponding reference process parameter range.

[0014] A second aspect of the present invention provides an epoxy molding compound testing system based on production status monitoring. The system includes: a triggering unit, configured to form a state vector S based on multi-point status monitoring data collected from the epoxy molding compound conveying and pretreatment stages, and to determine whether sampling is triggered based on the state vector S; a sampling unit, configured to obtain a powder sample by isokinetic sampling from the material conveying path when sampling is triggered by the state vector S, and to send the powder sample into a micro-cylinder testing chamber; a testing unit, configured to perform isothermal locking on the powder sample in the micro-cylinder testing chamber, and to perform mechanical tests of segmented loading, constant load relaxation, and isokinetic unloading sequentially based on the micro-cylinder testing chamber to obtain corresponding force-displacement curves, constant load relaxation curves, and unloading recovery curves; and an output unit, configured to determine corresponding mechanical property indicators based on the force-displacement curves, constant load relaxation curves, and unloading recovery curves, and to combine the mechanical property indicators with the state vector S to perform threshold adaptive regression to generate the test results of the current batch of epoxy molding compound.

[0015] Through the above technical solution, the present invention ensures that the sample is consistent with the working conditions of the batch by using state vector-driven sampling triggering and isokinetic sampling; it obtains reproducible mechanical curves and extracts indicators through isothermal locking and segmented loading; and it outputs batch detection results by combining state vector with threshold adaptive regression, thereby improving the timeliness and accuracy of mechanical assessment before molding, reducing cross-shift judgment drift and strengthening the correlation with filling defects.

[0016] Other features and advantages of the embodiments of the present invention will be described in detail in the following detailed description section. Attached Figure Description

[0017] The accompanying drawings are provided to further illustrate embodiments of the present invention and form part of the specification. They are used together with the following detailed description to explain the embodiments of the present invention, but do not constitute a limitation thereof. In the drawings:

[0018] Figure 1 This is a flowchart of the steps of an epoxy molding compound testing method based on production status monitoring provided by one embodiment of the present invention;

[0019] Figure 2 This is a schematic diagram of the structure of a miniature cylindrical test cavity provided in one embodiment of the present invention;

[0020] Figure 3 This is a schematic diagram of a force-displacement curve provided by one embodiment of the present invention;

[0021] Figure 4 This is a schematic diagram of a constant load relaxation curve provided in one embodiment of the present invention;

[0022] Figure 5This is a schematic diagram of the unloading and recovery curve provided by one embodiment of the present invention;

[0023] Figure 6 This is a system structure diagram of an epoxy molding compound testing system based on production status monitoring, provided by one embodiment of the present invention.

[0024] Explanation of reference numerals in the attached figures

[0025] 10-High-precision linear drive unit; 20-Force sensor; 30-Controllable constant load holding unit; 40-Loading plunger; 50-Action section of unloading drive mechanism; 60-Cylindrical cavity body; 70-Rigid closed end face. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only for illustration and explanation of the embodiments of this application and are not intended to limit the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0027] It should be noted that if the embodiments of this application involve directional indicators (such as up, down, left, right, front, back, etc.), the directional indicators are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indicators will also change accordingly.

[0028] Furthermore, if the embodiments of this application involve descriptions such as "first" or "second," these descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, features defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the technical solutions of various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. If the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed in this application.

[0029] like Figure 1 As shown, this invention provides a method for testing epoxy molding compounds based on production status monitoring. The method includes:

[0030] Step S1: Based on the multi-point status monitoring data of the collected epoxy molding compound conveying and pretreatment process, a state vector S is formed, and based on the state vector S, it is determined whether to trigger sampling.

[0031] Specifically, the state monitoring data includes at least one measurement data collected by temperature sensors, humidity sensors, static pressure differential sensors, and mass flow meters installed at the epoxy molding compound conveying main pipe, feeding port, and drying discharge port; before forming the state vector S, the method further includes: performing time synchronization marking and noise filtering on the measurement data so that the state vector S corresponds to the mechanical property index in the time dimension when triggering sampling and threshold adaptive regression.

[0032] In this embodiment of the invention, the state monitoring data includes at least one measurement data collected by temperature sensors, humidity sensors, static pressure differential sensors, and mass flow meters deployed at the epoxy molding compound conveying main pipe, feeding port, and drying outlet. To ensure the time correspondence between the subsequent state vector S and the mechanical property indicators, these measurement data need to be uniformly synchronized and preprocessed before S is formed.

[0033] Measurement data from each measuring point are assigned a unified time stamp, and sampling jitter is corrected using a local monotonic clock and periodic heartbeat packets. When sampling gaps or discontinuous time stamps are detected in the time series, interpolation and resampling are used to ensure that all measuring points maintain a one-to-one correspondence within the same sampling period.

[0034] Furthermore, noise filtering is performed on the time-aligned measurement data. Abnormal spikes and drift values ​​are identified and removed using the Hampel criterion. High-frequency random disturbances are suppressed using median filtering combined with low-pass smoothing, while long-term slow drifts are restored to a stable baseline using a sliding detrending method. To reduce the impact of dimensional differences, the measurement values ​​of each channel are mapped to a uniform numerical range, and the mapping parameters are retained for subsequent reproduction.

[0035] Considering the time lag in material transport, the effective time for material to reach each measuring point is estimated based on the mass flow rate and pipe section length. Time lag compensation is then applied to the measurement data of the main pipe, feed inlet, and drying outlet. The compensated data is spatially sorted according to the material transport direction, and the differential characteristics and rate of change characteristics between adjacent measuring points are calculated. Simultaneously, stability indicators (such as sliding window variance, coefficient of variation, and differential pressure ripple intensity) and cross-measuring point consistency indicators (such as temperature and humidity gradient and mass conservation deviation) are extracted.

[0036] Finally, the original normalized values, difference and rate of change features, and stability and consistency indices are concatenated into a multidimensional array in a preset order to form a state vector S. To ensure that S is strictly synchronized with the mechanical property indices, key boundary moments of the testing phase (compaction start, constant load start, unloading start) are marked while recording the sampling trigger moment. The aggregated values ​​(mean, quantiles, or extreme values) of S are extracted within the time windows corresponding to these anchor moments as inputs for trigger determination and threshold adaptive regression.

[0037] Through the above steps, the generated state vector S retains the instantaneous working condition characteristics, contains the stability and gradient information of the conveying process, and can strictly correspond in time with the subsequent force-displacement curve, constant load relaxation curve and unloading recovery curve, thereby reducing mismatch and noise interference, and improving the representativeness of sampling and the consistency of detection and judgment.

[0038] Preferably, the state vector S is formed based on the multi-point status monitoring data of the epoxy molding compound conveying and pretreatment process. This includes: performing time synchronization and interpolation processing on the measurement data collected from each measuring point to ensure that the measurement data from different measuring points correspond one-to-one within the same sampling period; spatially sorting the time-synchronized measurement data according to the measuring point location and material conveying direction, and calculating the difference features and rate of change features of adjacent measuring point measurement data; and concatenating the measurement data, difference features, and rate of change features of each measuring point into a multi-dimensional array in a preset order, and using a normalization algorithm to map each dimension to a unified numerical range to obtain the state vector S used to trigger sampling and threshold adaptive regression.

[0039] In this embodiment of the invention, it is necessary to perform time synchronization processing on the measurement data collected from each measuring point. Since the sampling frequency and response time of sensors at locations such as the conveying main, feeding port, and drying outlet may differ, misalignment errors will occur when calculating differential features later if a unified time reference is not established. Specifically, a unified timestamp can be applied to each measurement data point, with the time reference provided by the same clock source or synchronization signal. If sampling gaps or discontinuous timestamps are found at individual measuring points, interpolation methods, such as linear interpolation or spline interpolation, can be used to fill in the missing points with the same sampling period as other measuring points, thereby achieving a one-to-one correspondence between measurement data from different measuring points within the same sampling period.

[0040] After time synchronization is completed, all measurement data are spatially sorted according to the physical arrangement of the measuring points, i.e., the actual positions of the measuring points on the conveying path and the direction of material conveying. This step ensures that the subsequent calculation of differential characteristics and rate of change characteristics conforms to the spatial transmission logic of the material. For example, the data at the feed inlet is sorted first, then the data in the middle section of the main pipe, and finally the data at the drying outlet. After sorting, differential characteristics (such as changes in temperature difference, humidity difference, and static pressure difference) and rate of change characteristics (such as the percentage change in flow rate per unit time) are calculated for the measurement data of adjacent measuring points to reflect the state gradient and dynamic changes of the material during the conveying process.

[0041] The original measurement data, difference features, and rate of change features are concatenated into a multidimensional array according to a preset field order. This order can be fixed according to the input requirements of subsequent models or decision algorithms to ensure the comparability of features at the same location across different batches. After concatenation, to eliminate differences in dimensions and numerical ranges between different physical quantities, a normalization algorithm is used to map all dimensions to a unified numerical range, such as [0,1] or [-1,1]. Minimum-maximum scaling, Z-score normalization, or robust scaling methods can be used for normalization, and the normalization parameters are recorded to maintain consistency in subsequent data processing.

[0042] The state vector S obtained through the above steps contains not only the absolute state information of each measuring point, but also spatial gradient and dynamic change information. This state vector can accurately reflect the overall working condition of the material in the conveying and preprocessing stages, and after time synchronization and normalization, it can be directly used as the input data source for triggering sampling and executing threshold adaptive regression. This construction method ensures a high degree of consistency between sampling decisions and actual production conditions, and improves the comparability of cross-batch and cross-equipment testing results, thereby effectively enhancing the stability and representativeness of the testing.

[0043] Preferably, the rule for determining whether to trigger sampling based on the state vector S is as follows: the measurement data of each dimension in the state vector S are compared one by one with the pre-calibrated stable interval threshold. When the measurement data of any dimension deviates from the corresponding stable interval by more than a preset deviation percentage, or the rate of change within a continuous sampling period exceeds a preset rate of change threshold, sampling is determined to be triggered. The preset deviation percentage and rate of change threshold are dynamically updated according to the distribution of mechanical property indicators of historical batches and are stored synchronously with the state vector S.

[0044] In this embodiment of the invention, each dimension of the state vector S is mapped to a specific source of measurement data, such as temperature, humidity, static pressure difference, and mass flow rate located at the conveying main pipe, feeding port, and drying outlet. For each dimension, the range of values ​​for that dimension under normal and stable production conditions is pre-defined using a large amount of historical batch production operation data and corresponding mechanical property index results; that is, the stable interval threshold. These stable interval thresholds not only reflect the normal fluctuation range under the process but can also be recalibrated according to changes in equipment, formula, or environmental conditions.

[0045] In the specific judgment process, the measurement data of each dimension of the state vector S is compared with its corresponding stable interval one by one. When the deviation of the measurement data of a certain dimension from the stable interval exceeds the set deviation percentage threshold, it can be considered that the state of that dimension has fluctuated significantly. At the same time, in order to avoid short-term spikes or instantaneous noise triggering sampling, it is also necessary to evaluate the trend of change within continuous sampling periods, that is, to calculate the rate of change of that dimension in multiple continuous sampling periods. When the absolute value of the rate of change continuously exceeds the preset rate of change threshold, it is determined that there is a continuous dynamic shift in that dimension.

[0046] The logical combination of triggering conditions adopts an "OR" relationship, meaning that as long as either dimension meets one of the two conditions mentioned above—exceeding the deviation limit or exceeding the rate of change limit—sampling is triggered. This combination method ensures dual sensitivity to sudden anomalies and gradual drifts, preventing the missed detection of short-term but large-amplitude changes in operating conditions, and also enabling timely detection of slowly accumulating abnormal trends.

[0047] It is worth noting that the deviation percentage threshold and the rate of change threshold are not fixed values, but are dynamically updated based on the distribution of mechanical property indicators of historical batches. For example, over a period of time, the mechanical property indicators and corresponding state vector S data of each batch are collected, the distribution characteristics of these indicators in qualified batches are statistically analyzed, and the boundary conditions of this distribution are reflected in the threshold settings of each dimension of the state vector. In this way, the sampling triggering conditions can adapt to long-term changes in production status, avoiding false triggering or missed triggering due to factors such as seasons, equipment wear and tear, or formula fine-tuning.

[0048] Finally, the updated stability interval threshold, deviation percentage threshold, and rate of change threshold are stored together with the state vector S. This not only allows for direct use of this historical data for model optimization in subsequent threshold adaptive regression, but also enables the restoration of the original triggering conditions and production status during quality traceability, ensuring the interpretability and traceability of detection decisions.

[0049] Step S20: When the state vector S triggers sampling, a powder sample is obtained by constant-velocity sampling from the material conveying path, and the powder sample is sent into the micro-cylinder test chamber.

[0050] Specifically, a bypass sampling branch connected to the feed end of the micro-cylinder test chamber is set on the main material conveying route to collect powder samples. The sampling branch is equipped with a differential pressure self-compensation device to automatically adjust the bypass flow rate according to the flow rate change of the main pipeline to maintain a stable Reynolds number. The differential pressure self-compensation device is used to adjust the flow rate in real time during the sampling process. Powder sample accumulation is performed at the end of the sampling branch. When the accumulated mass reaches the preset sampling amount, sampling is stopped and the sample is sent into the micro-cylinder test chamber through a one-way introduction step.

[0051] In this embodiment of the invention, a bypass sampling branch is reserved on the main material conveying route, connecting to the feed end of the micro-cylinder test chamber. The interface of this branch should be located as close as possible to the main material flow center area of ​​the target batch to reduce the possibility of particle stratification or flow disturbance during material diversion. The flow channel size at the branch inlet matches the main pipeline to avoid excessive local pressure drop during sampling that could affect the flow pattern in the main pipeline.

[0052] During sampling, a differential pressure self-compensation device is installed in the bypass sampling branch. This device automatically adjusts the flow cross-section in the branch according to changes in the mass flow rate or static pressure difference in the main pipeline, thereby maintaining a relatively constant Reynolds number in the branch. This adjustment can be achieved through the linkage of an elastic diaphragm and a flow-limiting valve, or through a differential pressure-driven variable orifice mechanism. Since the stability of the Reynolds number directly affects the conveying state and particle velocity distribution of the powder in the branch, real-time flow rate adjustment helps to avoid sudden changes in sampling speed caused by fluctuations in the main pipeline flow rate, thus ensuring constant sampling velocity and sample representativeness.

[0053] A powder accumulation unit is installed at the end of the sampling branch. This unit can be a quantitative metering chamber or a closed cavity, and its volume and shape are designed according to the preset sampling amount to ensure that the sample accumulates stably without additional compaction or loosening disturbance. During the accumulation process, the mass of the accumulated powder is monitored in real time by a built-in mass detection unit (such as a mass flow meter or a high-precision weighing module). When the accumulated mass is detected to have reached the preset sampling amount, the branch inlet valve is immediately closed to cut off the sampling source and prevent excessive sample from overloading the test chamber or introducing additional flow impact.

[0054] After sampling is stopped, the powder sample from the accumulation unit is introduced into the micro-cylinder test chamber in one go through a one-way introduction mechanism. This one-way introduction mechanism should ensure that no backflow occurs during the material flow into the test chamber and reduce sudden changes in the flow rate of the material during the introduction process, thereby avoiding particle breakage or agglomeration during the introduction stage. To this end, a slow-release valve or a gradually opening gate structure can be used to allow the powder to enter the bottom of the test chamber smoothly and naturally accumulate and form under the action of gravity.

[0055] The design of the above sampling and introduction process ensures the controllability and stability of the entire process of powder sample delivery, from main pipeline diversion, constant velocity conveying, quantitative accumulation to one-time introduction into the test chamber. This not only guarantees a high degree of consistency between the physical state of the sample and the real-time operating conditions of the main pipeline, but also avoids sample deviations caused by flow disturbances, particle size stratification, or changes in the external environment during the sampling process. The sample that finally enters the test chamber can accurately reflect the material conveying and pretreatment state at the trigger moment, thereby improving the reliability and comparability of subsequent mechanical test results and reducing detection errors and batch-to-batch differences.

[0056] Step S30: The powder sample is isothermally locked in the micro cylindrical test chamber, and mechanical tests of segmented loading, constant load relaxation and constant speed unloading are carried out in sequence based on the micro cylindrical test chamber to obtain the corresponding force-displacement curve, constant load relaxation curve and unloading recovery curve.

[0057] Specifically, the inner wall and base of the micro-cylindrical test chamber are preheated to the target set temperature, which can be determined according to the simulated process conditions or material properties required for the test. A closed-loop temperature control strategy should be adopted during preheating, using a temperature sensor to monitor the chamber temperature in real time and adjust the power of the heating element to ensure a smooth temperature rise and avoid overshooting beyond the set temperature. After the powder sample is introduced into the test chamber, the air inlet or outlet connected to the outside is immediately closed to maintain a relatively sealed environment within the test chamber, reducing heat exchange between the sample and the outside air. Simultaneously, measures to homogenize the thermal field inside the test chamber are activated, such as through a surrounding heating jacket or a multi-zone temperature control module on the inner wall, to control the temperature difference of the sample in the vertical and horizontal, and internal and external directions within a preset range.

[0058] During the locking phase, the temperature at multiple locations within the cavity is continuously monitored. When the temperature at all monitoring points stabilizes within the set value ± allowable fluctuation range, and the fluctuation duration exceeds a preset stability criterion (e.g., no significant change for several consecutive sampling cycles), isothermal locking is considered complete. If necessary, temperature sensing elements can be installed on the sample surface to directly monitor the sample's temperature, ensuring thermal equilibrium is reflected not only in the cavity environment but also within the sample particles. Through the aforementioned isothermal locking steps, stable and uniform temperature conditions can be provided for the powder sample before subsequent segmented loading, constant load relaxation, and constant rate unloading tests. This reduces stress fluctuations or material property drift caused by thermal gradients, making the obtained mechanical curves more comparable and reproducible.

[0059] like Figure 2 The miniature cylindrical test chamber includes a vertically arranged cylindrical cavity body 60, a loading plunger 40 assembly coaxially arranged therewith, a controllable constant load holding unit 30, and a precision unloading drive mechanism, so as to continuously complete three mechanical tests in a single cavity: segmented loading, constant load relaxation, and constant speed unloading.

[0060] The inner wall of the cavity body is made of a high-strength material with a low coefficient of thermal expansion, and the inner diameter tolerance is controlled within a very small range to ensure a constant gap between the pressure head and the inner wall during loading, avoiding uneven force distribution caused by radial sway. The bottom of the cavity body is a rigid closed end face 70, and the top is equipped with a sealing ring seat, which seals with the outer diameter of the loading plunger 40 to ensure stable force transmission and prevent powder leakage during loading and unloading.

[0061] The loading plunger 40 assembly is coaxially connected to the force sensor 20 via a high-precision linear drive unit 10. The drive unit enables closed-loop speed control, allowing the plunger to advance according to a preset speed curve during the segmented loading stage and maintain a constant unloading speed during the constant-speed unloading stage. The plunger end face can be selected with a flat, spherical, or textured contact surface according to the powder characteristics to optimize stress distribution during the compaction process.

[0062] The constant load holding unit is arranged between the drive system and the plunger. It can adopt a lever-type loading buffer or a closed-loop force control servo mechanism to realize real-time feedback and compensation of force signals, so that the loading force is kept within the set value range during the constant load relaxation stage, and the force attenuation caused by powder creep or structural loosening is offset.

[0063] The unloading drive mechanism's action section 50 shares a guide system with the loading plunger 40, but has an independent speed control channel. It can seamlessly switch to the constant speed unloading mode after the constant load stage ends, avoiding sample disturbance caused by replacement or transfer.

[0064] Preferably, the rules for obtaining the force-displacement curve are as follows: After the powder sample is isothermally locked in the micro-cylinder test chamber, a pre-loading stabilization phase is executed. The indenter of the micro-cylinder test chamber is advanced to a preset ratio range of the target loading stroke at a pre-loading rate lower than the target loading rate, and the acoustic emission event density and displacement fluctuation amplitude are detected in real time. When both meet the preset stability criteria, the main loading phase is switched. During the main loading phase, the force and displacement time series data are subjected to first-order and second-order differences. When a joint event occurs in which the first-order difference sign changes continuously and the second-order difference sign flips, accompanied by the acoustic emission sequence boundary crossing, it is marked as a densification initiation event. From the densification initiation event, a subdivision sampling phase is activated until the absolute value of the first-order displacement difference returns to the first adaptive boundary. The measurement points of the pre-loading stabilization phase and the subdivision sampling phase are spliced ​​together in chronological order to form the force-displacement curve.

[0065] In this embodiment of the invention, after the powder sample is introduced into the micro-cylindrical test chamber and isothermally locked, the pre-loading stage begins. The loading rate during this stage should be lower than the target loading rate to reduce the impact of inertial impact and local particle rearrangement during the initial compaction process on measurement accuracy. During loading, the acoustic emission event density sequence and displacement fluctuation amplitude sequence are acquired in real time. The acoustic emission signal can be obtained through a high-sensitivity sensor deployed on the outer wall of the test chamber, and the displacement fluctuation amplitude is calculated from the maximum-minimum difference of the displacement sensor within a short time window. These two sequences are simultaneously compared with a preset stability criterion. For example, if the acoustic emission event density is lower than a set threshold and the displacement fluctuation amplitude is less than the allowable range within several consecutive sampling periods, the current compaction state is considered to be stabilizing. When this condition is met, pre-loading ends, this data segment is recorded as the "pre-loading stable segment," and the process switches to the main loading stage.

[0066] After entering the main loading stage, the loading rate is adjusted to the target loading rate to ensure stable propulsion speed and force conditions in the middle section of the force-displacement curve. During this period, the first-order and second-order differences are calculated for the collected force and displacement time-series data, respectively. The first-order difference reflects the trend of force or displacement change, and the second-order difference reflects the acceleration of the trend change. When the characteristic of "continuous change of the sign of the first-order difference and flipping of the sign of the second-order difference" is detected, and the acoustic emission event density exceeds its first adaptive boundary (constructed by the absolute deviation of the moving median and the median), it can be determined that the current loading process has entered the initial stage of particle densification.

[0067] Starting from the densification initiation event, the process enters the subdivision sampling phase. In this phase, the sampling interval should be shortened, and the acquisition frequency of force and displacement should be increased to capture minute changes in the mechanical response during densification. Subdivision sampling continues until the absolute value of the first-order difference of the displacement returns to the first adaptive boundary band, indicating that the loading rate and displacement change rate have recovered to the steady state at the end of the densification phase. Throughout the curve generation process, only two key data segments are retained: one is the pre-loading stable segment, reflecting the mechanical response of the powder during initial compaction under low-speed loading; the other is the subdivision sampling segment of the densification phase, providing high-resolution details of force-displacement changes. Stitching these two data segments in chronological order yields the final force-displacement curve.

[0068] Preferably, the rules for obtaining the constant load relaxation curve are as follows: after the powder sample is compacted to the density mark, the constant load control is switched; when the load error enters the second adaptive boundary band constructed by the moving median and its absolute deviation, the recording pre-start is executed; based on the recording pre-start, a test displacement step is applied and the transient stress response of the test displacement step is used to determine whether there is a reaction intervention. If it is determined to be a reaction intervention, segmented isothermal treatment is performed and the recording is started after the load error and temperature shift return to their respective boundary bands; during the recording period, thermal drift correction is performed according to the cavity temperature sequence and the initial stage of constant load establishment and mechanical limit segment are removed, and only continuous measurement points that meet the stability criteria are retained to form the constant load relaxation curve.

[0069] In this embodiment of the invention, after the segmented loading stage is completed, the indenter is continuously advanced until the force-displacement curve reaches a preset density threshold. This threshold can be determined by a sudden change in the loading force, a sharp decrease in the displacement increment, or a characteristic change in the acoustic emission signal, indicating that the porosity between sample particles has been largely eliminated and the process has entered the final stage of densification. At this point, the loading mode is immediately switched to constant load control, and the force sensor 20 adjusts the indenter position or the output of the loading mechanism in a closed loop to maintain a constant loading force.

[0070] After entering the constant load control phase, the load error, i.e., the difference between the current actual applied force and the target constant load force, is calculated in real time, and a second adaptive boundary is constructed based on the moving median and the median absolute deviation (MAD). This boundary dynamically adjusts according to the statistical characteristics of the data within the time window to adapt to the natural relaxation process of the material during the initial constant load stage. When the load error enters and remains within the range of the second adaptive boundary, a pre-start recording is performed, which means that the system has entered the initial plateau region where the force value is relatively stable, and the next step of judgment can be performed.

[0071] After the initial recording, one or more small-amplitude test displacement steps are applied, and the transient stress response curve is monitored in real time. If characteristic waveforms related to thermal reaction and residual stress release (such as non-monotonic decay, delayed rise, etc.) appear in the stress changes caused by these steps, it is determined that a reaction has occurred. In this case, to avoid confusion of the relaxation curve by chemical reaction or physical rearrangement process, a segmented isothermal strategy should be implemented. That is, while maintaining constant load conditions, the temperature of local areas of the cavity is adjusted through the temperature control module to slow down or stabilize the material reaction rate. This process requires simultaneous monitoring of load error and cavity temperature deviation. The recording of the relaxation curve is only formally started when both return to their respective adaptive boundary ranges.

[0072] During the formal recording phase, thermal drift correction must be performed throughout the entire process based on the cavity temperature sequence. Specifically, a correction model can be established between temperature and the baseline drift of the applied force to remove spurious signals caused by temperature fluctuations from the original force data. In addition, data segments in the initial stage of constant load establishment (where force values ​​fluctuate significantly) and near the mechanical limit stage (where displacement hardly changes and structural interference may occur) should be deleted to prevent these unstable segments from interfering with curve analysis.

[0073] Ultimately, only continuous measurement points that meet stability criteria during the constant load phase are retained. These criteria may include load fluctuations below a certain range, temperature fluctuations within allowable ranges, and no sudden displacement jumps. Arranging these filtered and calibrated data in chronological order constitutes the constant load relaxation curve.

[0074] Preferably, the rules for obtaining the unloading recovery curve are as follows: after constant load relaxation is completed and the load error falls into the second adaptive boundary band, switch to constant speed unloading; determine the contact decoupling time by jointly judging the first adaptive boundary band with the first differential correlation coefficient of load displacement and the surface lateral expansion rate; continuously collect the original sequence of rebound displacement and time from the contact decoupling time, and simultaneously record the acoustic emission event density sequence; remove the intervals where the acoustic emission event density exceeds the first adaptive boundary band from the original sequence, and splice the remaining effective segments in chronological order to form an effective sequence of rebound displacement time; construct a third adaptive boundary band for the displacement velocity of the effective sequence, and take the moment when the displacement velocity enters and remains within the third adaptive boundary band as the rebound termination point; construct the unloading recovery curve based on the effective sequence between the contact decoupling time and the rebound termination point.

[0075] In this embodiment of the invention, after the constant load relaxation phase is completed, the load error is monitored in real time. When the error falls into the second adaptive boundary band constructed by the moving median and the median absolute deviation (MAD) and remains stable, the loading mode is switched to constant-rate unloading. The unloading speed should be pre-calibrated to ensure that the speed is constant throughout the unloading section to avoid inconsistent mechanical responses caused by speed fluctuations.

[0076] During constant-rate unloading, load and displacement data are continuously collected, and their first-order difference is calculated in real time. Simultaneously, the change in the lateral expansion rate of the sample surface is acquired through external optical or displacement sensing methods. The first-order difference correlation coefficient of load-displacement and the lateral expansion rate of the surface are compared with the first adaptive boundary band, and the contact decoupling moment is determined by a joint judgment method. This moment usually corresponds to the point where the contact surface between the indenter and the sample begins to lose effective constraint, the load drops sharply, and the recovery trend of the lateral expansion rate reverses.

[0077] Once the contact decoupling moment is determined, the rebound phase begins immediately. From this moment onward, the original sequence of rebound displacement and time is continuously recorded, along with the acoustic emission event density sequence, for subsequent identification of interference segments involving internal structural adjustments or particle rearrangement. During rebound data processing, intervals where the acoustic emission event density exceeds the first adaptive boundary are considered interference segments potentially indicating structural rearrangement and are removed from the original displacement-time sequence. The remaining valid data segments are then concatenated in chronological order to obtain a continuous valid rebound displacement-time sequence, ensuring the overall continuity and physical consistency of the curve. Next, the displacement velocity of the valid sequence is calculated, and a third adaptive boundary is constructed based on the absolute deviation between the median and the middle value within its sliding window. When the displacement velocity enters and remains within this third adaptive boundary for a duration that meets a preset stability criterion, the corresponding moment is defined as the rebound termination point. This termination point typically represents the sample rebound process approaching a stable plateau, where displacement changes tend to cease.

[0078] Finally, the effective displacement-time series from the contact decoupling moment to the springback termination point is used as the unloading recovery curve. This curve can intuitively reflect the sample's elastic recovery capability, residual deformation ratio, and the process of energy release from the internal structure after unloading, providing an accurate data basis for subsequent calculations of mechanical properties such as elastic recovery rate and hysteresis energy ratio.

[0079] In one possible implementation, such as Figure 3 As shown, this curve reflects the compaction process of epoxy molding compound powder under isothermal conditions. In the initial loading stage (0–1.5 mm displacement), the curve slope is small, representing the initial compaction stage where the voids between powder particles are gradually compressed, with the loading pressure increasing from 0 MPa to approximately 35 MPa. A yield point appears at approximately 1.5 mm displacement, at which point relative slippage and rearrangement begin at the particle contact surfaces, and the curve slope decreases. After the yield point, the densification stage begins (approximately 1.8–4.5 mm displacement), where the particle arrangement becomes more compact, porosity decreases significantly, and the loading pressure increases rapidly with displacement, eventually reaching approximately 110 MPa at 4.5 mm displacement. This curve can be used to identify mechanical characteristic indicators such as the yield point location and the slope of the densification stage, providing basic data for subsequent threshold regression analysis.

[0080] Furthermore, such as Figure 4 At the start of constant load, the displacement was approximately 4.00 mm, then increased slowly over time, reaching approximately 4.25 mm at 150 s, indicating that the powder underwent creep behavior under constant pressure. The slope of the curve was relatively large in the initial stage, then gradually decreased, indicating that stress redistribution and structural loosening within the powder were mainly concentrated in the initial stage, and tended to stabilize in the later stage. This curve can be used to fit the equivalent relaxation time constant and evaluate the stability and rheological characteristics of materials under constant load conditions.

[0081] Furthermore, such as Figure 5 The rebound process of the powder sample was observed. The displacement at the end of unloading was approximately 4.25 mm, which then gradually decreased over time, stabilizing at approximately 3.90 mm at 60 s, indicating an elastic recovery of about 0.35 mm. The initial rebound rate was high, gradually slowing down and plateauing later, reflecting the time characteristics of stress release and structural rebound within the powder. This curve can be used to calculate indicators such as elastic recovery rate and hysteresis energy ratio, thereby aiding in the assessment of the material's compressive rebound performance and internal structural stability.

[0082] Step S40: Determine the corresponding mechanical property index based on the force-displacement curve, constant load relaxation curve and unloading recovery curve, and combine the mechanical property index with the state vector S to perform threshold adaptive regression to generate the test results of the current batch of epoxy molding compound.

[0083] Specifically, the yield point and densification slope are identified based on the force-displacement curve; an equivalent relaxation time constant is fitted based on the constant load relaxation curve; the elastic recovery rate and hysteresis energy ratio are calculated based on the unloading recovery curve; a set of mechanical property indicators is formed based on the yield point, densification slope, equivalent relaxation time constant, elastic recovery rate, and hysteresis energy ratio; the set of mechanical property indicators is normalized and combined with the state vector S to form a joint feature set; threshold adaptive regression is performed using the joint feature set as input, and the judgment threshold is updated based on historical batch calibration data; the detection result of the current batch of epoxy molding compound is generated based on the boundary relationship of the set of mechanical property indicators relative to the updated judgment threshold, and the detection result includes at least the judgment level and the corresponding reference process parameter range.

[0084] In this embodiment of the invention, the force-displacement curve is processed and its features are extracted. Using the force and displacement data from the initial and middle segments of the curve, combined with the changing trends of the first and second order differences, the yield point is accurately identified. This point typically corresponds to the moment when a significant rearrangement of the particle contact network occurs. The yield point can be determined by detecting abrupt changes in the loading force-displacement slope, and if necessary, by incorporating the increasing trend of acoustic emission events as an auxiliary signal. Subsequently, linear or piecewise linear fitting is performed on the densification segment (the section of the curve from the yield point to the end of loading) to obtain the slope of the densification segment. This slope reflects the contribution of changes in particle packing density to macroscopic stiffness.

[0085] For the constant load relaxation curve, the equivalent relaxation time constant is fitted using the load decay data during the constant load holding stage. This constant can be obtained by fitting an exponential decay model or a generalized viscoelastic model. During fitting, the transition segment at the initial stage of constant load establishment and the end segment near the mechanical limit should be removed first, and temperature drift correction should be performed to ensure that the relaxation time constant only reflects the intrinsic viscoelastic relaxation characteristics of the material.

[0086] The unloading recovery curve is processed. First, the elastic recovery rate from the end of unloading to the rebound termination point is calculated, which is the ratio of the rebound displacement to the total displacement before unloading, reflecting the material's ability to return to its original state after unloading. Then, the hysteresis energy ratio is calculated, obtained by the ratio of the hysteresis area enclosed by the force-displacement curve throughout the loading-unloading process to the total energy during the loading process, used to measure the degree of energy loss and internal friction. Both of these indicators must be calculated based on valid curve data after removing interference segments to ensure accuracy.

[0087] After obtaining five fundamental characteristics—yield initiation point, compaction slope, equivalent relaxation time constant, elastic recovery rate, and hysteresis energy ratio—these are combined into a set of mechanical property indices. To eliminate differences in dimensions and numerical ranges among different indices, the indices set is normalized. Min-max normalization, Z-score standardization, or other robust scaling methods can be used, while retaining the normalization parameters to ensure comparability across batches.

[0088] The normalized set of mechanical property indices is concatenated with the corresponding state vector S to form a joint feature set. The state vector S records the real-time operating conditions of the conveying and pretreatment process, including features such as temperature, humidity, static pressure difference, mass flow rate and its difference, rate of change, stability, and consistency. Therefore, the joint feature set not only includes the intrinsic mechanical response of the material but also incorporates the production operating conditions, providing a more comprehensive input for the regression model.

[0089] Using a joint feature set as input, threshold adaptive regression is performed. The core of this regression is to establish a decision threshold function based on the calibration dataset of historical batches, and dynamically update it as production data accumulates, allowing the threshold to adapt to changes in statistical distribution under different equipment, formulations, and environmental conditions. The regression can employ quantile regression, constrained regression, or other robust methods to reduce the interference of abnormal batches on the threshold. Finally, based on the boundary relationships between the current batch's mechanical property index set and the updated decision threshold, the test results for that batch are generated. The test results include at least the decision level (e.g., pass, warning, rejection) and the corresponding reference process parameter range, providing an actionable adjustment basis for the production process.

[0090] Figure 6 This is a system structure diagram of an epoxy molding compound testing system based on production status monitoring, provided by one embodiment of the present invention. Figure 6As shown, this invention provides an epoxy molding compound testing system based on production status monitoring. The system includes: a triggering unit, used to form a state vector S based on multi-point status monitoring data collected from the epoxy molding compound conveying and pretreatment stages, and to determine whether to trigger sampling based on the state vector S; a sampling unit, used to obtain powder samples by isokinetic sampling from the material conveying path when sampling is triggered by the state vector S, and to send the powder samples into a micro-cylinder testing chamber; a testing unit, used to perform isothermal locking on the powder samples in the micro-cylinder testing chamber, and to perform mechanical tests of segmented loading, constant load relaxation, and isokinetic unloading sequentially based on the micro-cylinder testing chamber to obtain corresponding force-displacement curves, constant load relaxation curves, and unloading recovery curves; and an output unit, used to determine corresponding mechanical property indicators based on the force-displacement curves, constant load relaxation curves, and unloading recovery curves, and to combine the mechanical property indicators with the state vector S to perform threshold adaptive regression to generate the test results of the current batch of epoxy molding compound.

[0091] Those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. This program is stored in a storage medium and includes several instructions to cause a microcontroller, chip, or processor to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0092] The optional embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the embodiments of the present invention are not limited to the specific details described above. Within the scope of the technical concept of the embodiments of the present invention, various simple modifications can be made to the technical solutions of the embodiments of the present invention, and these simple modifications all fall within the protection scope of the embodiments of the present invention. It should also be noted that the various specific technical features described in the above specific embodiments can be combined in any suitable manner without contradiction. To avoid unnecessary repetition, the embodiments of the present invention will not further describe the various possible combinations.

[0093] Furthermore, various different embodiments of the present invention can be combined in any way, as long as they do not violate the spirit of the embodiments of the present invention, they should also be regarded as the content disclosed by the embodiments of the present invention.

Claims

1. An epoxy molding compound test method based on production status monitoring, characterized in that, The method comprises: forming a state vector S based on the collected multi-point state monitoring data of the epoxy plastic packaging material conveying and pretreatment link, and determining whether to trigger sampling based on the state vector S; wherein The state monitoring data includes at least one measurement data collected by the temperature sensor, humidity sensor, static pressure difference sensor and mass flow meter arranged at the epoxy plastic packaging material conveying dry pipe, feeding port and dry discharge port; before forming the state vector S, the method further comprises: time synchronization marking and noise filtering processing of the measurement data, so that the state vector S corresponds to the mechanical property index in the time dimension when triggering sampling and threshold adaptive regression; Forming a state vector S based on the collected multi-point state monitoring data of the epoxy plastic packaging material conveying and pretreatment link comprises: performing time synchronization and interpolation completion processing on the measurement data collected by each measuring point, so that the measurement data of different measuring points correspond to each other in the same sampling period; spatially sorting each measurement data after time synchronization according to the measuring point position and the material conveying direction, and calculating the difference feature and the change rate feature of the adjacent measuring point measurement data; splicing the measurement data, the difference feature and the change rate feature of each measuring point into a multi-dimensional array according to a predetermined order, and mapping each dimension to a unified numerical interval using a normalization algorithm to obtain a state vector S for triggering sampling and threshold adaptive regression; When the state vector S triggers sampling, a powder sample is obtained from the material conveying path at a constant speed, and the powder sample is sent into a micro-cylinder test chamber; Isothermal locking of the powder sample is performed in the micro-cylinder test chamber, and mechanical testing of segmented loading, constant load relaxation and constant speed unloading is sequentially performed in the micro-cylinder test chamber to obtain corresponding force-displacement curves, constant load relaxation curves and unloading recovery curves; Based on the force-displacement curve, the constant load relaxation curve and the unloading recovery curve, the corresponding mechanical property index is determined, and the mechanical property index is combined with the state vector S to perform threshold adaptive regression to generate the detection result of the current batch of epoxy plastic packaging material, including: Based on the force-displacement curve, the yield starting point and the densification segment slope are identified; based on the constant load relaxation curve, the equivalent relaxation time constant is fitted; based on the unloading recovery curve, the elastic recovery rate and the hysteresis energy ratio are calculated; based on the yield starting point, the densification segment slope, the equivalent relaxation time constant, the elastic recovery rate and the hysteresis energy ratio, a set of mechanical property indexes is formed; the set of mechanical property indexes is normalized and combined with the state vector S to form a joint feature set, the threshold adaptive regression is performed with the joint feature set as input, and the determination threshold is updated according to the historical batch calibration data; the detection result of the current batch of epoxy plastic packaging material is generated according to the out-of-bound relationship of the set of mechanical property indexes relative to the updated determination threshold, and the detection result at least includes the determination level and the corresponding reference process parameter interval.

2. The method for testing epoxy molding compound based on production status monitoring according to claim 1, wherein, The rule for determining whether to trigger sampling based on the state vector S is: The dimensional measurement data in the state vector S is compared with the pre-calibrated stable interval threshold one by one, and when the measurement data of any dimension deviates from the corresponding stable interval by more than the preset deviation percentage or the change rate in the continuous sampling period exceeds the preset change rate threshold, it is determined that sampling is triggered; Wherein, the preset deviation percentage and change rate threshold are dynamically updated according to the mechanical property index distribution of historical batches, and are stored synchronously with the state vector S.

3. The method for testing epoxy molding compound based on production status monitoring according to claim 1, wherein, When the state vector S triggers sampling, a powder sample is obtained from the isokinetic sampling of the material conveying path, and the powder sample is sent into the micro-cylinder test cavity, including: The powder sample is sampled based on the bypass sampling branch connected to the inlet end of the micro-cylinder test cavity and arranged on the main pipeline of the material conveying path; wherein, The sampling branch is provided with a differential pressure self-compensation device for automatically adjusting the bypass flow rate according to the change of the main pipeline flow rate to maintain the stability of the Reynolds number, so as to adjust the flow rate in real time during sampling by using the differential pressure self-compensation device; At the end of the sampling branch, the powder sample is accumulated, and when the accumulated mass reaches the preset sampling amount, the sampling is stopped and the sample is sent into the micro-cylinder test cavity through the one-way introduction step.

4. The method for testing epoxy molding compound based on production status monitoring according to claim 1, wherein, The acquisition rule of the force-displacement curve is: After the powder sample completes isothermal locking in the micro-cylinder test cavity, a preloading stabilization segment is executed, the pressure head of the micro-cylinder test cavity is pushed to a preset proportion interval of the target loading stroke at a preloading lower than the target loading rate, and the acoustic emission event density and displacement fluctuation amplitude are detected in real time, and when both satisfy the preset stability criterion, switch to main loading; During the main loading, first-order and second-order differences of force and displacement time series data are performed, and when a combined event of first-order difference sign continuous change and second-order difference sign flip occurs with acoustic emission sequence boundary out-of-bounds, it is marked as a densification initiation event; From the densification initiation event, a subdivision sampling stage is enabled until the displacement first-order difference absolute value returns to the first adaptive boundary; The measurement points of the preloading stabilization segment and the subdivision sampling segment are spliced in time sequence as the force-displacement curve.

5. The method for testing epoxy molding compound based on production status monitoring according to claim 1, wherein, The acquisition rule of the constant load relaxation curve is: After the powder sample is compacted to the density mark, switch to constant load control; When the load error enters the second adaptive boundary constructed by the moving median and the absolute deviation of the median, record pre-start is executed; Based on the record pre-start, a test displacement step is applied, and whether there is a reaction intervention is judged according to the test displacement step transient stress response, if it is judged that there is a reaction intervention, segmented isothermal is executed and the record is started after the load error and temperature deviation return to their respective boundaries at the same time; During the recording period, thermal drift correction is implemented according to the cavity temperature sequence, and the initial stage of constant load establishment and the mechanical limiting segment are removed, only the continuous measurement points that satisfy the stability criterion are retained to form the constant load relaxation curve.

6. The method for testing epoxy molding compound based on production status monitoring according to claim 1, wherein, The acquisition rule of the unloading recovery curve is: After the constant load relaxation is completed and the load error falls into the second adaptive boundary, switch to isokinetic unloading; The contact decoupling time is determined by the joint determination of the first adaptive boundary by the load displacement first-order difference correlation coefficient and the surface transverse expansion rate; From the contact decoupling time, the original sequence of rebound displacement and time is continuously collected, and the acoustic emission event density sequence is recorded synchronously. The intervals of the acoustic emission event density exceeding the first adaptive boundary band are removed from the original sequence, and the remaining effective segments are spliced in chronological order to form a rebound displacement time effective sequence; A third adaptive boundary band is constructed for the displacement speed of the effective sequence, and the time when the displacement speed enters and continuously remains in the third adaptive boundary band is taken as the rebound termination point; An unloading recovery curve is constructed based on the effective sequence between the contact decoupling time and the rebound termination point.

7. An epoxy molding compound test system based on production status monitoring, characterized by, The system is used to perform the production state monitoring based epoxy plastic packaging material testing method in any one of claims 1-6, and the system comprises: a triggering unit configured to form a state vector S based on the collected multi-measurement-point state monitoring data of the epoxy plastic packaging material conveying and pretreatment links, and determine whether to trigger sampling based on the state vector S; a sampling unit configured to obtain a powder sample from the material conveying path at a constant speed when the state vector S triggers sampling, and send the powder sample into a micro-cylinder test chamber; a testing unit configured to perform isothermal locking on the powder sample in the micro-cylinder test chamber, and sequentially implement segmented loading, constant load relaxation and constant speed unloading mechanical tests based on the micro-cylinder test chamber to obtain corresponding force-displacement curves, constant load relaxation curves and unloading recovery curves; an output unit configured to determine corresponding mechanical property indexes based on the force-displacement curves, constant load relaxation curves and unloading recovery curves, and combine the mechanical property indexes with the state vector S to perform threshold adaptive regression to generate a detection result of the current batch of epoxy plastic packaging materials.

Images