Sterile barrier seal strength detection method and system based on servo pull force

Abstract

The application belongs to the technical field of sealing detection, and specifically relates to a sterile barrier sealing strength detection method and system based on servo tension, which comprises the following steps: cutting a sterile sealing sample to form an initial detection sample; applying a constant displacement constraint to construct a constant mechanical test reference data set; synchronously collecting servo force values and dielectric parameters after temperature control stabilization to obtain homologous time sequence data and integrate the mechanical-dielectric original time sequence data set; extracting stress relaxation data to calculate relaxation modulus, fitting and constructing a mechanical-dielectric correlation model, and deducing long-term creep failure life from short-time testing. Finally, the sealing strength equivalent value is calculated by combining the life and the interface state and through model mapping, and a full-process data closed loop is formed. In the application, the servo tension and dielectric homologous detection, short-time cross-scale extrapolation and block chain storage are fused to realize accurate characterization of sealing strength, full-cycle nondestructive prediction and compliance traceable closed loop, and the detection precision and efficiency are improved.

Description

Technical Field

[0001] This invention belongs to the field of sealing detection and analysis technology, specifically a method and system for detecting the sealing strength of sterile barriers based on servo tension. Background Technology

[0002] Traditional methods for testing the sealing strength of sterile barriers have significant shortcomings. They often employ destructive, single-point testing methods, only acquiring instantaneous breaking strength values. This fails to predict long-term deterioration trends and lifespan of the seal, and the tested samples are discarded, making subsequent integrity verification impossible. Furthermore, the testing dimensions are limited, focusing only on macroscopic mechanical properties and ignoring the microscopic deterioration state of the sealing interface. This makes it difficult to detect early failure precursors such as interface debonding and gap initiation, resulting in low accuracy. In addition, traditional methods lack systematic data calibration and evidence preservation mechanisms, leading to fragmented, untraceable data that is susceptible to errors caused by assembly and environmental interference. They also cannot simultaneously acquire mechanical and dielectric parameters, failing to comprehensively reflect the true state of the seal strength and thus being unsuitable for the high-precision, full-cycle, and compliantly traceable testing requirements of sterile medical devices. Summary of the Invention

[0003] To overcome the shortcomings of existing technologies, this invention proposes a method for detecting the sealing strength of sterile barriers based on servo-tension. This invention primarily addresses the problems of destructive testing, inability to predict lifespan, and lack of data traceability in traditional methods.

[0004] The present invention provides a method for detecting the sealing strength of a sterile barrier based on servo tension, comprising: S1: cutting a sterile sealing sample, attaching a flexible interdigital electrode to its sealing interface and connecting it to an impedance analyzer to form an initial test sample.

[0005] S2: Apply constant displacement constraint to the initial test sample, activate the force value acquisition state, and construct a constant mechanical test benchmark dataset by fusing sample geometric features and circuit reference parameters.

[0006] S3: Based on the constant mechanical test benchmark dataset, set the parameters in the impedance analyzer, start the temperature control module to stabilize the temperature, trigger synchronous acquisition, match the real-time servo force value with the dielectric parameters, and obtain the same source time sequence test data.

[0007] S4: Based on the same source time-series detection data, the dielectric parameters are continuously collected until they tend to stabilize. The stress attenuation changes and the evolution information of dielectric constant, loss factor and impedance modulus within the collection period are used to form the mechanical-dielectric original time-series dataset.

[0008] S5: Extract stress relaxation data from the original mechanical-dielectric time series dataset to calculate the relaxation modulus. Fit the relaxation modulus with the characteristic peak value of the dielectric loss factor to construct a mechanical-dielectric correlation model and substitute it with the preset creep failure threshold. Extrapolate the long-term creep failure life of the seal through short-term test data.

[0009] S6: Based on the interface bonding state derived from the long-term creep failure life and dielectric parameters, the measured equivalent value of the sterile barrier sealing strength is calculated through the mechanical-dielectric correlation model mapping, forming a closed-loop data system for the entire process.

[0010] According to the sterile barrier sealing strength detection method based on servo tensile force provided by the present invention, the specific steps for constructing the constant mechanical test benchmark dataset in step S2 are as follows:

[0011] S21: The assembly surface of the initial test specimen is precisely fitted with the rigid constraint fixture. The travel of the fixture is calibrated by the displacement sensor to complete the constant displacement constraint locking of the specimen with full degrees of freedom, and the specimen positioning reference data is output.

[0012] S22: Import the sample positioning reference data into the force acquisition device, activate the piezoelectric force sensor and signal conditioning circuit, acquire the initial force value timing signal under the constraint state according to the preset sampling frequency, and output the original force value.

[0013] S23: Use three-dimensional visual scanning to obtain the inner and outer contours, wall thickness, and sealing surface roughness of the sample. Based on the original force value acquisition data stream, complete the spatial matching of feature points and force value sampling points, and output the geometric-force value correlation feature matrix.

[0014] S24: Based on the circuit reference parameters, the geometry-force correlation feature matrix is ​​corrected for drift and outliers are removed. The mechanical response law under the constraint boundary is fitted by finite element simulation to generate a constant mechanical test reference dataset.

[0015] According to the sterile barrier sealing strength detection method based on servo tensile force provided by the present invention, the specific steps for generating the constant mechanical test benchmark dataset in step S24 are as follows:

[0016] Based on the geometric-force correlation feature matrix, the circuit reference parameters corresponding to the initial test sample are retrieved, the drift data in the matrix are accurately corrected, and the outliers generated during the acquisition process are removed to obtain the corrected geometric-force correlation feature matrix.

[0017] A finite element simulation model adapted to constant displacement constraint scenarios is built based on the modified geometry-force value correlation feature matrix. The constraint boundary conditions are input, and the mechanical response law of the specimen under constraint conditions is fitted through simulation iteration to form a mechanical response simulation dataset.

[0018] Data regularization, dimensional unification, and accuracy calibration are performed on the modified geometry-force correlation feature matrix and mechanical response simulation data to generate a constant mechanics test benchmark dataset.

[0019] According to the sterile barrier sealing strength detection method based on servo tension provided by the present invention, the specific steps for obtaining homologous time-series detection data in step S3 are as follows:

[0020] S31: Based on the sampling frequency, measurement point coordinates and calibration coefficients of the constant mechanical test benchmark dataset, construct a linkage mapping model of mechanical parameters-dielectric instrument configuration, and dynamically iterate and optimize the excitation voltage, test frequency band and number of sampling points through an adaptive algorithm to output the instrument configuration parameter set.

[0021] S32: Input the instrument configuration parameter set into the impedance analyzer and perform open-circuit and short-circuit load calibration. Use PID temperature control and predictive compensation algorithm to stabilize to the target detection temperature and output a constant temperature environment status indicator.

[0022] S33: Parse the temperature stability check code in the constant temperature environment status mark, trigger the dual-path trigger signal with timing pre-synchronization, drive the servo system to achieve adaptive loading according to the constraint law of the mechanical benchmark dataset, synchronously activate the force value acquisition and noise reduction module, and output the servo real-time force value timing stream.

[0023] S34: The same source anchoring algorithm is used to accurately align the servo real-time force value timing stream and the dielectric parameter timing stream with a unified timestamp. Combined with correlation analysis, timing mismatch and signal distortion data are automatically identified and removed to generate a same source timing detection dataset.

[0024] According to the sterile barrier sealing strength detection method based on servo tension provided by the present invention, the specific steps for precise timestamp alignment in step S34 are as follows:

[0025] The common clock anchors of the force value timing stream and dielectric parameter timing stream are extracted. Cubic spline interpolation resampling is performed on the two non-equal interval data streams to unify the sampling frequency and timestamp axis. The remaining timing offset is corrected by the peak location of the cross-correlation function, and the two signals are accurately digitally aligned in the time domain to generate the aligned dual-channel timing matrix.

[0026] The Pearson correlation coefficient r is calculated window by window based on the aligned dual-channel timing matrix. A mismatch threshold is set, signal amplitude distortion and jump points are detected, an anomaly binary mask matrix is ​​generated, and the timing matrix is ​​removed according to the mask to output the cleaned dual-channel timing matrix.

[0027] The cleaned dual-channel time series matrix is ​​normalized and quantized, digital metadata is added to the matrix, and the matrix data structure of the fused metadata is organized in CSV format to generate a homogeneous time series detection dataset with digital checksum.

[0028] According to the sterile barrier sealing strength detection method based on servo tension provided by the present invention, the specific steps for forming the mechanical-dielectric raw time series dataset in step S4 are as follows:

[0029] S41: Based on the same source time-series detection data, the adaptive sliding window iterative algorithm is used to collect force value and time-series dielectric data, calculate the trend entropy and fluctuation threshold of dielectric constant, loss factor, and impedance modulus in real time, dynamically determine the steady state, and output continuously acquired time-series segments.

[0030] S42: Based on the continuously acquired time series segments and trend fitting parameters, the acquisition period is divided using a periodic adaptive segmentation algorithm. Through the time series alignment verification model, the stress decay sequence and dielectric parameter sequence within each period are accurately mapped, and the decay rate and dielectric evolution gradient features are extracted to generate a mechanical-dielectric correlated time series subset.

[0031] S43: Based on the mechanical-dielectric correlation time series subset, an anomaly detection algorithm is used to perform adaptive weighted interpolation repair on missing points, integrate full-cycle multi-dimensional feature data and complete time series consistency verification, and output the original mechanical-dielectric time series dataset.

[0032] According to the sterile barrier sealing strength detection method based on servo tension provided by the present invention, the specific steps for estimating the long-term creep failure life of the seal in step S5 are as follows:

[0033] S51: Extract the force values ​​and time series of the stress relaxation stage from the mechanical-dielectric raw time series dataset, fit the evolution of the relaxation modulus using a variable-order exponential decay model, and output the dynamic relaxation modulus sequence.

[0034] S52: Based on the dynamic relaxation modulus sequence, multi-scale wavelet transform is used to extract the time-frequency domain characteristic peak of the dielectric loss factor. The nonlinear mapping between the relaxation modulus and the dielectric loss peak is constructed by weighting through an attention mechanism, and the mechanical-dielectric correlation model is output.

[0035] S53: Substitute the mechanical-dielectric correlation model into the preset creep failure threshold, combine the transfer learning framework to transfer long-term life data of similar materials, realize cross-scale extrapolation of short-term test data, and output the prediction results of long-term creep failure life of the seal.

[0036] According to the sterile barrier sealing strength detection method based on servo tension provided by the present invention, the specific steps in step S52 for outputting the mechanical-dielectric correlation model are as follows:

[0037] The time series of dielectric loss factor is extracted from the dynamic relaxation modulus sequence. Multi-scale wavelet transform is used to decompose the dielectric loss factor in the time and frequency domain to obtain wavelet coefficient matrices at different scales. Local extrema at each scale are located as characteristic peaks in the time and frequency domain.

[0038] The time-frequency domain feature peaks are aligned with the dynamic relaxation modulus sequence. The contribution weight of each feature peak to the relaxation modulus is calculated through an attention mechanism to generate weighted feature-modulus association pairs and construct the initial nonlinear mapping relationship.

[0039] The weighted feature-modulus correlation pairs are input into a deep neural network, and the network parameters are iteratively optimized using the fitting error of the correlation pairs as the loss function. After convergence, the mechanical-dielectric correlation model with dynamic weight allocation capability is output.

[0040] According to the sterile barrier sealing strength detection method based on servo tension provided by the present invention, the specific steps for forming a closed-loop data process in step S6 are as follows:

[0041] S61: Based on the interface-combined defect distribution derived from the long-term creep failure lifetime confidence interval and dielectric time-frequency domain characteristics, a multi-dimensional intensity mapping feature tensor is constructed. The creep constitutive correction factor is embedded to perform tensor normalization processing, resulting in a normalized intensity mapping feature tensor.

[0042] S62: Input the normalized intensity mapping feature tensor into the mechanical-dielectric correlation model, solve the equivalent value of the sterile barrier sealing strength through adaptive weight inference, and generate an intensity quantization spectrum with error traceability.

[0043] S63: Based on the intensity quantization spectrum, connect the original data, model parameters and verification results of the entire process, establish a blockchain-style evidence storage link to complete data verification and form a closed data chain for the entire process.

[0044] This invention also provides a sterile barrier seal strength testing system based on servo tension, comprising:

[0045] The sample preparation module is used to cut sterile sealed samples, attach flexible interdigitated electrodes to their sealing interface and connect them to an impedance analyzer to form an initial test sample.

[0046] The benchmark construction module is used to apply constant displacement constraints to the initial test specimen, activate the force value acquisition state, and integrate the specimen's geometric features and circuit benchmark parameters to construct a constant mechanical test benchmark dataset.

[0047] The synchronous acquisition module is used to set parameters on the impedance analyzer based on the constant mechanical test benchmark dataset, start the temperature control module to stabilize the temperature, trigger synchronous acquisition, match the real-time servo force value with the dielectric parameters, and obtain the same source time sequence test data.

[0048] The dataset generation module is used to continuously collect data from the same source time-series detection data until the dielectric parameters tend to stabilize, and to generate the stress decay changes and the evolution information of dielectric constant, loss factor and impedance modulus during the collection period, forming the mechanical-dielectric original time-series dataset.

[0049] The life prediction module is used to extract stress relaxation data from the original mechanical-dielectric time series dataset, calculate the relaxation modulus, fit the relaxation modulus with the characteristic peak value of the dielectric loss factor, construct a mechanical-dielectric correlation model, and substitute it with the preset creep failure threshold to predict the long-term creep failure life of the seal through short-term test data.

[0050] The strength closed-loop module is used to calculate the measured equivalent value of the sterile barrier sealing strength based on the interface bonding state derived from the long-term creep failure life and dielectric parameters, and to form a closed-loop data system for the entire process.

[0051] The sterile barrier sealing strength detection method based on servo tension provided by this invention integrates servo tension and dielectric homology detection, short-time cross-scale extrapolation, and blockchain evidence storage to achieve accurate characterization of sealing strength, full-cycle non-destructive prediction, and compliant traceability closed loop, thereby improving detection accuracy and efficiency.

[0052] The beneficial effects of this invention are as follows:

[0053] 1. This invention deeply integrates servo-driven tensile displacement closed-loop constraint technology with flexible interdigital electrode dielectric detection technology. It simultaneously acquires macroscopic mechanical signals and microscopic dielectric parameters of sealed samples. Combined with co-source clock anchoring, precise timing alignment, and multi-dimensional data calibration, it constructs a constant mechanical testing benchmark and a co-source timing detection system, effectively eliminating systematic errors caused by assembly misalignment, circuit interference, and environmental fluctuations. Compared to traditional single mechanical or dielectric detection methods, this integrated approach can simultaneously capture stress relaxation, modulus decay, and the evolution of dielectric characteristics at the sealing interface. It accurately inverts the distribution of interface defects and the degradation law of bonding force, providing a comprehensive and reliable multi-dimensional data source for the equivalent calculation of sterile barrier sealing strength. This significantly improves the resolution, accuracy, and data traceability of sealing strength detection, solving the technical pain point of traditional detection methods that cannot accurately capture subtle interface degradation.

[0054] 2. This invention extracts stress relaxation information from the original mechanical-dielectric time-series data, combines variable-order exponential fitting, multi-scale wavelet transform, and attention mechanisms to construct a high-precision mechanical-dielectric correlation model. It then introduces a transfer learning framework to transfer long-term lifetime data of similar materials, enabling cross-scale extrapolation of short-term test data to long-term creep failure lifetime. This method eliminates the need for long-term aging tests, significantly shortening the testing cycle and reducing testing costs. Furthermore, it employs a non-destructive testing mode throughout, allowing the test samples to be used for subsequent sterile barrier integrity verification. Simultaneously, through the coupled mapping of relaxation modulus and dielectric characteristics, the real-time residual value of the sealing strength can be inversely deduced, quantifying the strength degradation rate. This achieves a shift from instantaneous strength testing to predicting the strength trend throughout the entire life cycle, effectively overcoming the limitation of traditional destructive testing, which can only obtain single-point instantaneous strength and cannot predict long-term reliability.

[0055] 3. This invention uses a quantitative seal strength spectrum as the core anchor point, connecting the entire process of sample preparation, benchmark construction, synchronous acquisition, model deduction, and strength calculation. It embeds creep constitutive correction and adaptive weighted inference technologies to achieve deep linkage and precise matching of mechanical, dielectric, and strength data. Simultaneously, it employs blockchain-based multi-node hash encryption technology to construct an immutable evidence storage link, completing the consistency and integrity verification of data throughout the entire process, forming a complete closed loop of detection-modeling-deduction-verification-evidence storage. This closed-loop system not only solves the problems of fragmented and untraceable traditional testing data but also enables full verification of the strength calculation process through error source tracing. The retained non-destructive samples can be further linked to sterile barrier integrity verification, achieving linked verification of seal strength and barrier performance. This fully meets the compliance requirements of sterile medical device quality control, providing standardized and quantifiable technical support for product quality supervision and fault tracing. Attached Figure Description

[0056] The invention will now be further described with reference to the accompanying drawings.

[0057] Figure 1 This is a flowchart illustrating the steps of the sterile barrier sealing strength detection method based on servo tensile force provided in this embodiment of the invention.

[0058] Figure 2 This is a flowchart of the sterile barrier sealing strength detection method based on servo tensile force provided in the embodiments of the present invention;

[0059] Figure 3 This is a block diagram of a sterile barrier sealing strength detection system based on servo tension provided in an embodiment of the present invention. Detailed Implementation

[0060] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below according to specific embodiments.

[0061] like Figures 1 to 3 As shown in the embodiment of the present invention, the method for detecting the sealing strength of a sterile barrier based on servo tension includes:

[0062] S1: Cut the sterile sealed sample and attach the flexible interdigitated electrode to the sealing interface. Connect the electrode leads to the impedance analyzer to complete the insulation check to ensure the stability of the test circuit and form the initial test sample.

[0063] Cut the sterile sealed sample, remove impurities and excess edges from the sample surface, and form a sterile sealed base sample.

[0064] A flexible interdigitated electrode is precisely bonded to the complete sealed interface of the sterile sealed base sample. After bonding, the electrode leads are arranged to form a pretreated sample.

[0065] Connect the electrode leads of the pretreated sample to the impedance analyzer to avoid poor contact affecting the test stability. After the lead connection is completed, the sample to be tested for insulation is obtained.

[0066] Insulation testing is performed on the test samples that have been connected to the impedance analyzer to check for potential problems such as short circuits and poor contact. The lead connection status is adjusted in a timely manner until the test is qualified to ensure the stability of the test circuit and form an initial test sample suitable for subsequent tests.

[0067] S2: Clamp the initial test sample onto the insulating fixture of the servo tensile testing machine, switch the servo tensile testing machine to the displacement closed-loop control mode, set the preload displacement and lock the displacement value, open and wait for the force value acquisition channel, and construct the mechanical test benchmark conditions with constant constraints based on the initial state of the circuit.

[0068] S21: The assembly surface of the initial test specimen is precisely fitted with the rigid constraint fixture. The travel of the fixture is calibrated by the displacement sensor to complete the constant displacement constraint locking of the specimen with full degrees of freedom, and the specimen positioning reference data is output.

[0069] The initial test specimen assembly surface is aligned edge by edge with the positioning reference surface of the rigid constraint fixture, and the gap is checked with a feeler gauge. ,like > (Allowable gap threshold), attitude correction is performed through a three-axis fine-tuning mechanism, with a correction amount of... This eliminates assembly misalignment and gap errors.

[0070] Start the three-axis feed stroke of the displacement sensor closed-loop calibration fixture, collect the displacement feedback value of each feed stage, collect the feed displacement feedback xi, yi, zi, calculate and compensate for the cumulative deviation of the stroke.

[0071] The formula for calculating cumulative travel deviation is:

[0072]

[0073] In the formula, The mean displacement To set the displacement.

[0074] The formula for calculating the cumulative travel deviation compensation is as follows:

[0075]

[0076] In the formula, The sum of cumulative deviations across multiple strokes refers to the algebraic sum of the cumulative deviations ΔX generated in each feed action from the start of the tooling to the current moment. The cumulative deviation of the j-th stage of travel refers to the average displacement at the j-th sampling time. With the corresponding level of set displacement The difference, where m is the cumulative number of feeds.

[0077] Based on the calibrated precise displacement parameters, the tooling locking mechanism is driven to lock the six degrees of freedom of the specimen, and outputs specimen positioning reference data including three-dimensional coordinates, constraint stiffness, and stroke compensation values. .

[0078] S22: Import the sample positioning reference data into the force acquisition device, activate the piezoelectric force sensor and signal conditioning circuit, acquire the initial force value timing signal under the constraint state according to the preset sampling frequency, and output the original force value.

[0079] The sample positioning reference data is imported into the main control unit of the force acquisition device to complete the one-to-one mapping configuration between the tooling coordinates and the sensor acquisition channels, and to verify the matching accuracy of the coordinate mapping. Power is supplied to the piezoelectric force sensor and signal conditioning circuit, and zero-point clearing, gain calibration, and noise floor testing are performed to ensure that the circuit is in a stable acquisition state.

[0080] The formula for zeroing out is expressed as follows:

[0081]

[0082] In the formula, Where n is the no-load voltage, and n is the number of sampling points. The zero-point reset compensation voltage is denoted by k, which is the sample number and represents the kth no-load voltage sampling.

[0083] Continuous acquisition is triggered by a preset high-frequency sampling frequency, and the sampling timestamp and corresponding constraint coordinates are marked synchronously. The continuous force value time sequence signal under constraint state is stored, and the original force value dataset with timestamp and coordinate association is output.

[0084] S23: Use three-dimensional visual scanning to obtain the inner and outer contours, wall thickness, and sealing surface roughness of the sample. Based on the original force value acquisition data stream, complete the spatial matching of feature points and force value sampling points, and output the geometric-force value correlation feature matrix.

[0085] The focal length, scanning accuracy, and field of view of the 3D vision scanning equipment were adjusted to perform non-contact scanning of the entire sample. Internal and external contour coordinates, wall thickness distribution values, and micro-roughness parameters of the sealing surface were extracted through point cloud reconstruction. The timestamps and sampling coordinate sequences of the original force value dataset were analyzed, and geometric feature points were matched with force value sampling points according to the principle of spatial coordinate consistency, eliminating invalid data with mismatched coordinates. The one-to-one corresponding geometric feature parameters and force value data were integrated along a unified dimension to construct a geometric-force correlation feature matrix containing spatial coordinates, geometric parameters, and force value signals.

[0086] S24: Based on the circuit reference parameters, the geometry-force correlation feature matrix is ​​corrected for drift and outliers are removed. The mechanical response law under the constraint boundary is fitted by finite element simulation to generate a constant mechanical test reference dataset.

[0087] Based on the geometric-force correlation feature matrix, the circuit reference parameters corresponding to the initial test sample are retrieved, the drift data in the matrix are accurately corrected, and the outliers generated during the acquisition process are removed to obtain the corrected geometric-force correlation feature matrix.

[0088] A finite element simulation model adapted to constant displacement constraint scenarios is built based on the modified geometry-force value correlation feature matrix. The constraint boundary conditions are input, and the mechanical response law of the specimen under constraint conditions is fitted through simulation iteration to form a mechanical response simulation dataset.

[0089] Data regularization, dimensional unification, and accuracy calibration are performed on the modified geometry-force correlation feature matrix and mechanical response simulation data to generate a constant mechanics test benchmark dataset.

[0090] S3: Based on the mechanical testing benchmark conditions, set the test frequency band and low-voltage AC test voltage of the impedance analyzer to be compatible with the polymer sealing material, start the temperature control module to stabilize the test environment temperature, trigger the servo tensile tester and the impedance analyzer to collect data synchronously, and collect the force value output by the servo tensile tester and the dielectric parameters collected by the impedance analyzer on the time axis to obtain macroscopic mechanical and microscopic dielectric co-source time sequence test data.

[0091] S31: Based on the sampling frequency, measurement point coordinates and calibration coefficients of the constant mechanical test benchmark dataset, construct a linkage mapping model of mechanical parameters-dielectric instrument configuration, and dynamically iterate and optimize the excitation voltage, test frequency band and number of sampling points through an adaptive algorithm to output the instrument configuration parameter set.

[0092] A time-domain sampling frequency Fourier spectrum analysis was performed on the mechanical test benchmark dataset to extract the dominant frequency components and aliasing threshold. The Cartesian coordinates of the measurement points were converted into spatially normalized digital coordinate vectors (mapped to [0,1]). Least square error fitting was performed on the calibration coefficients to generate the calibration coefficient error covariance matrix. After removing outlier coefficients, a standardized digital feature library was constructed.

[0093] The numerical modeling of the linkage mapping model uses mechanical parameters (stress, strain, loading rate) as the input dimension and dielectric instrument configuration (excitation voltage, test frequency band, number of sampling points) as the output dimension to construct a multi-input multi-output (MIMO) nonlinear digital mapping model. Tensor decomposition is used to reduce dimensional redundancy, and the mechanical-dielectric coupling transfer function matrix is ​​obtained by solving it to complete the initial numerical calibration of the model.

[0094] The adaptive iterative optimization digital computation sets the fitness function as the dielectric signal-to-noise ratio (SNR) plus the mechanical response matching degree (RMSE). Through iterative adaptive particle swarm optimization (PSO) algorithm, a candidate set of parameters is generated in each round, substituted into the mapping model for forward numerical calculation, and the fitness value is calculated. Particle velocities and positions are updated, and convergence is determined. The iteration terminates when the fitness change rate is less than 1e-6, and the quantized digital parameter set of the instrument configuration (including voltage quantization step size, frequency band discrete point sequence, and integer value of the number of sampling points) is output.

[0095] S32: Input the instrument configuration parameter set into the impedance analyzer and perform open-circuit and short-circuit load calibration. Use PID temperature control and predictive compensation algorithm to stabilize to the target detection temperature and output a constant temperature environment status indicator.

[0096] The configuration parameter set is encoded into instrument bus digital command frames, and frequency domain data of open-circuit parasitic impedance, short-circuit reference impedance, and standard load impedance are collected sequentially. S-parameters are solved through vector network analysis to calculate the complex impedance compensation factor of the fixture cable, generating a full-range frequency domain calibration compensation digital matrix. Complex frequency domain digital compensation is then applied to the original dielectric test channel to eliminate system parasitic errors.

[0097] The PID discrete digital temperature control algorithm samples the analog signal from the temperature sensor using an ADC, quantizes it into a digital temperature value, and calculates the discrete deviation e(k) from the target temperature. The digital control quantity is then calculated using the digital PID formula. The digital PID formula is expressed as:

[0098]

[0099] In the formula, Let k be the digital control quantity at the k-th sampling time, where k is the sampling time number, representing the k-th sampling (k=0,1,2,...). The proportional coefficient is used to adjust the response speed of the deviation. The larger the proportional coefficient, the faster the response, but too large a coefficient can easily lead to system oscillation. Let be the deviation value at the k-th sampling time. The integral coefficient is used to eliminate the steady-state error of the system. T is the sampling period. These are differential coefficients used to suppress dynamic overshoot of the system. They predict future trends based on the rate of change of the deviation and apply damping in advance. The deviation value at the (k-1)th sampling time is used to calculate the rate of change of the deviation. The deviation change rate represents the amount of change in deviation per unit time. This is the cumulative sum of the biases from the 0th to the kth sampling.

[0100] The control quantity is quantized into a PWM duty cycle digital code and output to the temperature control actuator to form a closed-loop discrete regulation sequence.

[0101] Predictive compensation and constant temperature state digital coding use an exponential smoothing time series prediction model to extrapolate the temperature deviation at the next moment, perform feedforward compensation on the PID output to correct the lag error, and calculate the variance and peak value of the temperature digital sequence in real time. When the fluctuation is less than or equal to the set digital threshold, a constant temperature state identifier is generated: a binary digital code containing the temperature mean, variance check code, and stable duration, which serves as the subsequent trigger enable signal.

[0102] S33: Parse the temperature stability check code in the constant temperature environment status mark, trigger the dual-path trigger signal with timing pre-synchronization, drive the servo system to achieve adaptive loading according to the constraint law of the mechanical benchmark dataset, synchronously activate the force value acquisition and noise reduction module, and output the servo real-time force value timing stream.

[0103] The constant temperature status indicator is checked using a CRC digital verification. After successful verification, the trigger module is unlocked. Based on the system's own digital clock, clock frequency division and phase synchronization are performed to generate a dual-channel in-phase digital trigger pulse sequence. One channel drives the servo controller, and the other triggers the dielectric acquisition card, ensuring that the timing deviation between the two trigger channels is ≤1 clock cycle.

[0104] The servo adaptive loading digital solution analyzes the loading patterns of the mechanical benchmark dataset and converts them into a discrete force-displacement time-series digital table. The servo controller executes closed-loop digital control: it acquires feedback force values ​​in real time, calculates the deviation from the target time-series table, and outputs the drive digital quantity through a PID / fuzzy control algorithm, which is then converted into motor drive pulses. During the loading process, the loading rate digital parameter is dynamically corrected to adapt to changes in the material's mechanical response.

[0105] Force signal digital noise reduction and timing stream encapsulation: After quantization by ADC, the force acquisition signal is matched to the applied bandwidth using an IIR digital low-pass filter cutoff frequency. Zero-point drift is eliminated using a least-squares detrending term, and outlier detection and interpolation are performed using the 3σ criterion. Each valid force value data is stamped with a high-precision digital timestamp and encapsulated into a continuous servo real-time force digital timing stream.

[0106] S34: The same source anchoring algorithm is used to accurately align the servo real-time force value timing stream and the dielectric parameter timing stream with a unified timestamp. Combined with correlation analysis, timing mismatch and signal distortion data are automatically identified and removed to generate a same source timing detection dataset.

[0107] Extract the common clock anchor points of the force value timing stream and dielectric parameter timing stream, perform cubic spline interpolation resampling on the two non-equal interval data streams, and unify the sampling frequency and timestamp axis. Correct the remaining timing offset by locating the peak of the cross-correlation function, complete the precise digital alignment of the two signals in the time domain, and output the aligned dual-channel timing matrix.

[0108] The Pearson correlation coefficient r is calculated window-by-window for the aligned dual-channel timing matrix, with r < 0.9 set as the mismatch threshold. Simultaneously, signal amplitude distortion and abrupt changes in the matrix are detected, generating an outlier binary mask matrix. Row / column removal operations are performed on the dual-channel timing matrix according to the mask, retaining valid data segments. Short missing match points are repaired using linear interpolation, and the cleaned and normalized dual-channel timing matrix is ​​output.

[0109] The cleaned and standardized dual-channel time series matrix is ​​normalized and quantized, and digital metadata (configuration parameter ID, temperature check code, measurement point coordinate encoding, and time stamp precision) is added to the matrix. The matrix data structure of the fused metadata is organized in CSV format to generate a homogeneous time series detection dataset with digital checksums, and the final standardized homogeneous time series detection dataset (CSV format) is output.

[0110] S4: Based on the same source time-series detection data, the dielectric parameters are continuously collected until they tend to stabilize. The stress attenuation changes and the evolution information of dielectric constant, loss factor and impedance modulus within the collection period are used to form the mechanical-dielectric original time-series dataset.

[0111] S41: Based on the same source time-series detection data, the adaptive sliding window iterative algorithm is used to collect force value and time-series dielectric data, calculate the trend entropy and fluctuation threshold of dielectric constant, loss factor, and impedance modulus in real time, dynamically determine the steady state, and output continuously acquired time-series segments.

[0112] Using the unified timestamps, force value time series, and dielectric parameter time series of the same-source time series detection dataset as inputs, the basic width W0 and iteration step size S0 of the sliding window are first initialized, and an adaptive window scale adjustment mechanism is constructed: the first-order difference amplitude of the dielectric parameter in the time domain is calculated in real time; when the difference exceeds the threshold, the window shrinks to 0.5W0 to improve transient capture accuracy; when the difference is below the threshold, the window expands to 1.5W0 to enhance the statistical robustness of the stationary segment. Force value digital sequences and dielectric constant, loss factor, and impedance modulus digital sequences are collected iteratively window by window. Information trend entropy is calculated for the three types of dielectric parameters within a single window: the parameter sequences are first normalized to a probability distribution, and then... Solve for the entropy value, where H is the information trend entropy. This represents the probability value corresponding to the i-th state in the time series after normalization. Let be the probability value for the i-th state. The natural logarithm is used to quantify the information content of the probability distribution; the smaller the entropy value, the more regular the parameter trend. Simultaneously, based on the dynamic quantiles of the data within the window, the upper and lower limits of fluctuation are iteratively updated, and the parameter fluctuation residual is calculated after removing instantaneous impulse interference. A dual convergence criterion is set: the entropy of the dielectric parameter trend is less than the threshold H for three consecutive iteration windows. thFurthermore, if the residual fluctuations of the three types of parameters all fall within the dynamic threshold range, it is determined that the dielectric parameter has entered a steady state. The data stream is then truncated, and a continuous acquisition time sequence segment carrying the window number, trend entropy sequence, fluctuation threshold sequence, and timestamp is output.

[0113] S42: Based on the continuously acquired time series segments and trend fitting parameters, the acquisition period is divided using a periodic adaptive segmentation algorithm. Through the time series alignment verification model, the stress decay sequence and dielectric parameter sequence within each period are accurately mapped, and the decay rate and dielectric evolution gradient features are extracted to generate a mechanical-dielectric correlated time series subset.

[0114] Using continuously acquired time series segments and embedded trend entropy and fluctuation fitting parameters as input, an extreme point coupled periodic adaptive segmentation algorithm is employed. First, the inflection points of the stress sequence within each time series segment are solved using second-order difference. Combined with dielectric evolution gradient abrupt change points, the long time series is divided into independent acquisition periods such as initial loading, steady-state decay, and stable holding. The segmentation boundary is jointly calibrated by stress inflection points and dielectric abrupt change points. A time series alignment verification model is constructed, using the start and end timestamps of each period as anchor points. The temporal synchronization between the stress decay sequence and the three types of dielectric parameter sequences is verified through a cross-correlation function. After correcting period boundary deviations, precise point-to-point mapping is achieved. Linear fitting is performed on the stress sequence within each period to solve for the stress decay rate. Sliding difference is used to solve for the dielectric evolution gradient of the dielectric constant, loss factor, and impedance modulus sequences. The period number, decay rate, dielectric gradient, and original time series data are bound together to generate a multi-dimensional feature-annotated mechanical-dielectric correlated time series subset.

[0115] S43: Based on the mechanical-dielectric correlation time series subset, an anomaly detection algorithm is used to perform adaptive weighted interpolation repair on missing points, integrate full-cycle multi-dimensional feature data and complete time series consistency verification, and output the original mechanical-dielectric time series dataset.

[0116] Using a subset of mechanical-dielectric correlation time series data as input, a multi-dimensional fusion anomaly detection algorithm is employed: simultaneously detecting missing time series points, stress-dielectric correlation mismatch points, and parameter amplitude jump points. All anomalous data points are marked by setting correlation coefficient thresholds and gradient distortion thresholds. Adaptive weighted interpolation is then performed to repair missing data points; the weight calculation formula is as follows:

[0117]

[0118] In the formula, The stress-dielectric correlation coefficient at adjacent points. This represents the time-series distance. , To normalize the weighting coefficients, missing values ​​are fitted and filled using a weighted sum. After repair, a three-layer temporal consistency check is performed: verifying the strict increment of timestamps, the continuity of characteristic values ​​for each period, and the coupling of stress and dielectric parameter evolution trends. Period segments that do not meet the checks are re-segmented and repaired iteratively. After passing the check, the original time series data, derived feature data, period labels, check results, and metadata information are integrated to output a structured, traceable mechanical-dielectric original time series dataset.

[0119] S5: Extract stress relaxation data from the original mechanical-dielectric time series dataset to calculate the relaxation modulus, perform cross-boundary fitting with the characteristic peak value of the dielectric loss factor, construct a mechanical-dielectric correlation model, and substitute it with the preset creep failure threshold. Then, extrapolate the long-term creep failure life of the seal through short-term test data.

[0120] S51: Extract the force values ​​and time series of the stress relaxation stage from the mechanical-dielectric raw time series dataset, fit the evolution of the relaxation modulus using a variable-order exponential decay model, and output the dynamic relaxation modulus sequence.

[0121] From the original mechanical-dielectric time-series dataset, complete data segments of the stress relaxation stage were selected by using a first-order stress difference threshold. The force, time, and strain digital sequences of this stage were extracted, and strain was digitally normalized according to the gauge length of the seal to eliminate the influence of gauge length differences. A basic array for relaxation calculation was constructed, and the data sampling frequency and the number of valid data points were recorded.

[0122] Using the normalized force and strain numerical arrays as input, the relaxation modulus calculation formula is expressed as follows: In the formula, For relaxation modulus, This represents the real-time force value during the stress relaxation phase. This represents the initial cross-sectional area of ​​the seal. To obtain the normalized real-time strain, an initial relaxation modulus sequence was initially calculated. Iterative fitting was performed using exponential decay functions of orders 1 to 4. The formula for the exponential decay function is as follows:

[0123]

[0124] In the formula, y is the target variable obtained by fitting. Let be the amplitude coefficient of the i-th exponential component. Let t be the relaxation time of the i-th exponential component, and t be the time variable.

[0125] After each fitting order, the Akaike Information Criterion and Root Mean Square Error (RMSE) are calculated. The optimal fitting order is selected based on the dual screening criteria of minimum AIC and RMSE < 50 MPa. The model parameters at the optimal order are solved, and the interference of instantaneous elastic deformation of the material is eliminated through iterative correction to obtain a continuous fitting curve that closely matches the measured data.

[0126] The precise value of the relaxation modulus E(t) is calculated step-by-step based on the optimal order exponential decay model. The relaxation rate is calculated using a first-order difference algorithm to quantify the decay trend of the modulus over time. The calculated relaxation modulus sequence is then subjected to outlier detection using the 3σ criterion and linear interpolation for repair, generating a dynamic relaxation modulus sequence containing timestamps, relaxation modulus values, fitting residuals, and relaxation rates, which is then saved as a structured array format.

[0127] S52: Based on the dynamic relaxation modulus sequence, multi-scale wavelet transform is used to extract the time-frequency domain characteristic peak of the dielectric loss factor. The nonlinear mapping between the relaxation modulus and the dielectric loss peak is constructed by weighting through an attention mechanism, and the mechanical-dielectric correlation model is output.

[0128] The dielectric loss factor digital time series corresponding to the timestamps of the dynamic relaxation modulus sequence is extracted synchronously to ensure complete alignment of the two timelines. Daubechies wavelet decomposition is performed on the dielectric loss factor time series using a four-level multi-scale wavelet decomposition to obtain high-frequency detail coefficient matrices and low-frequency approximation coefficient matrices at each scale. A soft thresholding function is used to denoise the coefficient matrices at each scale, generating a denoised three-dimensional digital matrix of wavelet coefficients. A local extremum detection algorithm is used to locate local maxima points in the denoised coefficient matrices at each scale, and the amplitude, timestamp, corresponding scale, and frequency range of each extremum point are extracted as time-frequency domain feature peaks. The time-frequency domain feature peaks are time-aligned with the dynamic relaxation modulus sequence, and an attention mechanism is used to calculate the contribution weight of each feature peak to the relaxation modulus, generating weighted feature-modulus association pairs and constructing an initial nonlinear mapping relationship.

[0129] The weighted feature-modulus correlation pairs are input into a deep neural network. The mean square error between the predicted and measured dynamic relaxation modulus values ​​is used as the loss function. The Adam optimizer is used to iteratively optimize the network parameters. The iteration stops when the loss function converges to below 1e-6. The output is a mechanical-dielectric correlation model with dynamic weight allocation capability. The model includes a feature weight matrix, a nonlinear mapping function, and error correction coefficients.

[0130] S53: Substitute the mechanical-dielectric correlation model into the preset creep failure threshold, combine the transfer learning framework to transfer long-term life data of similar materials, realize cross-scale extrapolation of short-term test data, and output the prediction results of long-term creep failure life of the seal.

[0131] A long-term creep failure dataset of similar sealing materials is loaded, and a domain distribution difference analysis is performed with the current short-term test dataset to calculate the feature distribution distance between the two datasets. Domain adversarial network (DAN) training is used to minimize the feature distribution difference between the source and target domains, achieving domain adaptation. The weights of the mature long-term life prediction model trained in the source domain are transferred to the mechanical-dielectric correlation model. By fine-tuning the parameters of the fully connected layers of the model, the transferability of long-term failure features from the source domain is preserved, resulting in a transfer-enhanced mechanical-dielectric correlation model adapted to the current sealing component, improving the reliability of model extrapolation.

[0132] The preset creep failure modulus threshold of the seal is substituted into the migration-enhanced mechanical-dielectric correlation model. The critical peak value of the dielectric loss factor under the corresponding failure state is obtained by inverse model solving, and the frequency scale and characteristic weights corresponding to the critical peak value are labeled. A power-law time-scale extension function is used, combined with the critical peak value of the dielectric loss factor, to complete the cross-scale extrapolation calculation from short-term test data to long-term creep failure lifetime. Error coefficients are calculated simultaneously during the extrapolation process to correct the extrapolation results.

[0133] The KS test was used to verify the reasonableness of the distribution of the cross-scale extrapolation results, and the goodness of fit R of the extrapolation results was calculated. 2 The confidence interval is used to eliminate outlier extrapolated values ​​outside the confidence interval. The failure risk level of the predicted results is labeled by combining the source domain migration weights and the short-term data fitting error of the target domain. A structured prediction result for the long-term creep failure life of the seal is generated, including: the time it takes for the seal strength to drop to the failure threshold, the confidence interval, the goodness of fit, the failure risk level, the peak value of the critical dielectric loss, and the error coefficient, ensuring the accuracy and traceability of the prediction results.

[0134] The essence of a sterile barrier seal strength lies in the bonding force at the sealing interface and the material's resistance to creep deformation. The stress relaxation and relaxation modulus decay captured by your process are the direct mechanical root cause of the seal strength deterioration over time.

[0135] The continuous decrease in relaxation modulus leads to the loss of sealing interface bonding stiffness, which in turn leads to the attenuation of sealing clamping force, and the attenuation of sealing clamping force leads to a simultaneous decrease in sealing strength.

[0136] The peak value of the dielectric loss factor and abrupt changes in its time-frequency domain characteristics lead to debonding and gap initiation at the sealing interface, which in turn are precursor signals of sealing strength failure.

[0137] S6: Based on the long-term creep failure life results and the sealing interface bonding state reflected by the dielectric parameters, determine the creep failure risk level, retain the sample after the whole process non-destructive testing, connect to the subsequent sterile barrier integrity verification, and form a closed loop of whole process data.

[0138] S61: Based on the interface defect distribution obtained from the long-term creep failure lifetime confidence interval and dielectric time-frequency domain feature inversion, a multi-dimensional strength mapping feature tensor is constructed. A creep constitutive correction factor is embedded for tensor normalization to obtain a normalized strength mapping feature tensor. The upper and lower confidence limits and mean lifetime of the long-term creep failure lifetime, as well as the interface defect area, distribution uniformity, and debonding probability obtained from the dielectric time-frequency domain feature inversion, are extracted to construct a four-dimensional basic feature array, completing the feature tensor initialization. A material temperature-creep coupling constitutive correction factor is introduced to correct feature deviations under different operating conditions. The correction factor is multiplied dimension-by-dimensionally with the basic feature tensor to generate the original strength mapping feature tensor adapted to the operating conditions. The max-min normalization algorithm is used to scale the corrected tensor within the [0,1] interval to eliminate dimensional differences, outputting a normalized strength mapping feature tensor with uniform dimensions and regular distribution.

[0139] S62: Input the normalized intensity mapping feature tensor into the mechanics-dielectric correlation model, and solve for the equivalent value of the sterile barrier sealing strength through adaptive weight inference, generating an intensity quantization map with error tracing. Input the normalized intensity mapping feature tensor into the pre-trained mechanics-dielectric correlation model, complete the forward propagation through multi-layer nonlinear mapping, and output the initial equivalent predicted value of the sterile barrier sealing strength. Based on the contribution of each feature dimension to the strength, dynamically allocate adaptive inference weights to correct the model prediction bias and obtain a high-precision equivalent value of the sterile barrier sealing strength. Use the correlated equivalent strength value, the feature contribution weights of each dimension, the inference error, and the tracing node information to generate an intensity quantization map with error tracing, including numerical cloud maps, error curves, and tracing paths.

[0140] S63: By connecting the original data, model parameters, and verification results of the entire process using the intensity quantification spectrum, a blockchain-based evidence storage link is established to complete data verification and form a closed-loop data chain for the entire process. Using the intensity quantification spectrum as the core anchor point, a unified dataset for the entire process is formed based on sample information, benchmark datasets, source-specific time-series data, and associated model parameters. The unified dataset is hash-encrypted at each node, and chained evidence storage blocks are generated according to the detection time sequence to ensure that each segment of data is tamper-proof and reversibly traceable, thus completing the construction of the blockchain evidence storage link. Consistency and integrity checks are performed on the evidence storage data, abnormal nodes are corrected, and a compliant closed-loop data chain is output, achieving a complete closed loop of detection-modeling-deduction-verification-evidence storage.

[0141] like Figure 3 As shown, the present invention also provides a sterile barrier sealing strength testing system based on servo tension, comprising:

[0142] The sample preparation module is used to cut sterile sealed samples, attach flexible interdigitated electrodes to their sealing interface and connect them to an impedance analyzer to form an initial test sample.

[0143] The benchmark construction module is used to apply constant displacement constraints to the initial test specimen, activate the force value acquisition state, and integrate the specimen's geometric features and circuit benchmark parameters to construct a constant mechanical test benchmark dataset.

[0144] The synchronous acquisition module is used to set parameters on the impedance analyzer based on the constant mechanical test benchmark dataset, start the temperature control module to stabilize the temperature, trigger synchronous acquisition, match the real-time servo force value with the dielectric parameters, and obtain the same source time sequence test data.

[0145] The dataset generation module is used to continuously collect data from the same source time-series detection data until the dielectric parameters tend to stabilize, and to generate the stress decay changes and the evolution information of dielectric constant, loss factor and impedance modulus during the collection period, forming the mechanical-dielectric original time-series dataset.

[0146] The life prediction module is used to extract stress relaxation data from the original mechanical-dielectric time series dataset, calculate the relaxation modulus, fit the relaxation modulus with the characteristic peak value of the dielectric loss factor, construct a mechanical-dielectric correlation model, and substitute it with the preset creep failure threshold to predict the long-term creep failure life of the seal through short-term test data.

[0147] The strength closed-loop module is used to calculate the measured equivalent value of the sterile barrier sealing strength based on the interface bonding state derived from the long-term creep failure life and dielectric parameters, and to form a closed-loop data system for the entire process.

[0148] In summary, this embodiment provides a method and system for testing the sealing strength of sterile barriers based on servo-tension force. By deeply integrating servo-tension displacement closed-loop constraint technology with flexible interdigital electrode dielectric testing technology, it simultaneously acquires macroscopic mechanical signals and microscopic dielectric parameters of the sealed sample. Combined with co-source clock anchoring, precise timing alignment, and multi-dimensional data calibration, it constructs a constant mechanical testing benchmark and a co-source timing testing system, effectively eliminating systematic errors caused by assembly misalignment, circuit interference, and environmental fluctuations. Compared to traditional single mechanical or dielectric testing methods, this integrated mode can simultaneously capture stress relaxation, modulus decay, and the evolution of dielectric characteristics of the sealing interface, accurately inverting the distribution of interface defects and the degradation law of bonding force. It provides a comprehensive and reliable multi-dimensional data source for the equivalent calculation of sterile barrier sealing strength, significantly improving the resolution, accuracy, and data traceability of sealing strength testing, and solving the technical pain point of traditional testing methods that cannot accurately capture subtle interface degradation.

[0149] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of various embodiments or some parts of embodiments.

[0150] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. These modifications or substitutions do not cause the essence of the corresponding technical solutions to depart from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for detecting the sealing strength of a sterile barrier based on servo-tension force, characterized in that, include: S1: Cut the sterile sealed sample, attach the flexible interdigital electrode to its sealing interface and connect it to the impedance analyzer to form the initial test sample; S2: Apply a constant displacement constraint to the initial test sample, activate the force value acquisition state, and construct a constant mechanical test benchmark dataset by fusing the sample's geometric features and circuit reference parameters. S3: Based on the constant mechanical test benchmark dataset, set parameters in the impedance analyzer, start the temperature control module to stabilize the temperature, trigger synchronous acquisition, match the real-time servo force value with the dielectric parameter, and obtain the same source time sequence detection data; S4: Based on the same source time series detection data, the dielectric parameters are continuously collected until they tend to stabilize. The stress attenuation changes and the evolution information of dielectric constant, loss factor and impedance modulus within the collection period are used to form the mechanical-dielectric original time series dataset. S5: Extract stress relaxation data from the mechanical-dielectric original time series dataset to calculate the relaxation modulus, fit the relaxation modulus with the characteristic peak value of the dielectric loss factor, construct a mechanical-dielectric correlation model and substitute it with the preset creep failure threshold, and extrapolate the long-term creep failure life of the seal through short-time test data. S6: Based on the interface bonding state obtained from the long-term creep failure lifetime and dielectric parameter inversion, the measured equivalent value of the sterile barrier sealing strength is calculated through the mechanical-dielectric correlation model mapping, forming a closed loop of data for the entire process.

2. The method for detecting the sealing strength of a sterile barrier based on servo-tension according to claim 1, characterized in that: In step S2, the specific steps for constructing the constant mechanics test benchmark dataset are as follows: S21: Precisely fit the assembly surface of the initial test specimen with the rigid constraint fixture, calibrate the fixture stroke through the displacement sensor, complete the constant displacement constraint locking of the specimen with full degrees of freedom, and output the specimen positioning reference data; S22: Import the sample positioning reference data into the force acquisition device, activate the piezoelectric force sensor and signal conditioning circuit, acquire the initial force value timing signal under the constraint state according to the preset sampling frequency, and output the original force value; S23: Use three-dimensional visual scanning to obtain the inner and outer contours, wall thickness, and sealing surface roughness of the sample. Based on the original force value acquisition data stream, complete the spatial matching of feature points and force value sampling points, and output the geometric-force value correlation feature matrix. S24: Based on the circuit reference parameters, the geometry-force correlation feature matrix is ​​corrected for drift and outliers are removed. The mechanical response law under the constraint boundary is fitted by finite element simulation to generate a constant mechanical test reference dataset.

3. The method for detecting the sealing strength of a sterile barrier based on servo-tension according to claim 2, characterized in that: In step S24, the specific steps for generating the constant mechanics test benchmark dataset are as follows: Based on the geometric-force correlation feature matrix, the circuit reference parameters corresponding to the initial test sample are retrieved, the drift data in the matrix is ​​accurately corrected, and the abnormal values ​​generated during the acquisition process are removed to obtain the corrected geometric-force correlation feature matrix. Based on the modified geometry-force value correlation feature matrix, a finite element simulation model adapted to the constant displacement constraint scenario is built. The constraint boundary conditions are input, and the mechanical response law of the specimen under the constraint state is fitted through simulation iteration to form a mechanical response simulation dataset. The modified geometry-force correlation feature matrix and mechanical response simulation data are subjected to data normalization, dimensional unification and accuracy calibration to generate a constant mechanics test benchmark dataset.

4. The method for detecting the sealing strength of a sterile barrier based on servo-tension according to claim 1, characterized in that: In step S3, the specific steps for obtaining the same-source time series detection data are as follows: S31: Based on the sampling frequency, measurement point coordinates and calibration coefficients of the constant mechanical test benchmark dataset, construct a linkage mapping model of mechanical parameters-dielectric instrument configuration, and dynamically iterate and optimize the excitation voltage, test frequency band and number of sampling points through an adaptive algorithm to output the instrument configuration parameter set; S32: Input the instrument configuration parameter set into the impedance analyzer and perform open-circuit and short-circuit load calibration. Use PID temperature control and predictive compensation algorithm to stabilize to the target detection temperature and output constant temperature environment status indicator. S33: Parse the temperature stability check code in the constant temperature environment status mark, trigger the dual trigger signal with timing pre-synchronization, drive the servo system to achieve adaptive loading according to the constraint law of the mechanical benchmark dataset, synchronously activate the force value acquisition and noise reduction module, and output the servo real-time force value timing stream; S34: The same source anchoring algorithm is used to accurately align the servo real-time force value timing stream and the dielectric parameter timing stream with a unified timestamp. Combined with correlation analysis, timing mismatch and signal distortion data are automatically identified and removed to generate a same source timing detection dataset.

5. The method for detecting the sealing strength of a sterile barrier based on servo-tension according to claim 4, characterized in that: In step S34, the specific steps for precise timestamp alignment are as follows: Extract the common clock anchor points of the force value timing stream and dielectric parameter timing stream, perform cubic spline interpolation resampling on the two non-equal interval data, and unify the sampling frequency and timestamp axis; correct the remaining timing offset by the peak location of the cross-correlation function, complete the precise digital alignment of the two signals in the time domain, and generate the aligned dual-channel timing matrix; Based on the aligned dual-channel timing matrix, the Pearson correlation coefficient r is calculated window by window, a mismatch threshold is set, signal amplitude distortion and jump points are detected, an anomaly binary mask matrix is ​​generated, the timing matrix is ​​removed according to the mask, and the cleaned dual-channel timing matrix is ​​output. The cleaned dual-channel time series matrix is ​​normalized and quantized to add digital metadata to the matrix. The matrix data structure of the fused metadata is organized in CSV format to generate a homogeneous time series detection dataset with digital checksum.

6. The method for detecting the sealing strength of a sterile barrier based on servo-tension according to claim 1, characterized in that: In step S4, the specific steps for forming the mechanical-dielectric raw time series dataset are as follows: S41: Based on the same source time-series detection data, use an adaptive sliding window iterative algorithm to collect force values ​​and time-series dielectric data, calculate the trend entropy and fluctuation threshold of dielectric constant, loss factor, and impedance modulus in real time, dynamically determine the stable state, and output continuously collected time-series segments. S42: Based on the continuously acquired time series segments and trend fitting parameters, the acquisition period is divided using a periodic adaptive segmentation algorithm. Through the time series alignment verification model, the stress decay sequence and dielectric parameter sequence in each period are accurately mapped, and the decay rate and dielectric evolution gradient features are extracted to generate a mechanical-dielectric correlated time series subset. S43: Based on the aforementioned mechanical-dielectric correlated time series subset, an anomaly detection algorithm is used to perform adaptive weighted interpolation repair on missing points, integrate full-cycle multi-dimensional feature data and complete time series consistency verification, and output the original mechanical-dielectric time series dataset.

7. The method for detecting the sealing strength of a sterile barrier based on servo-tension according to claim 1, characterized in that: In step S5, the specific steps for estimating the long-term creep failure life of the seal are as follows: S51: Extract the force values ​​and time series of the stress relaxation stage from the mechanical-dielectric raw time series dataset, fit the evolution of the relaxation modulus using a variable-order exponential decay model, and output the dynamic relaxation modulus sequence. S52: Based on the dynamic relaxation modulus sequence, multi-scale wavelet transform is used to extract the time-frequency domain characteristic peak of the dielectric loss factor. The nonlinear mapping between the relaxation modulus and the dielectric loss peak is constructed by weighting through an attention mechanism, and the mechanical-dielectric correlation model is output. S53: Substitute the mechanical-dielectric correlation model into the preset creep failure threshold, combine the transfer learning framework to transfer long-term life data of similar materials, realize cross-scale extrapolation of short-term test data, and output the prediction result of long-term creep failure life of the seal.

8. The method for detecting the sealing strength of a sterile barrier based on servo tension according to claim 7, characterized in that: In step S52, the specific steps for outputting the mechanical-dielectric correlation model are as follows: The time series of dielectric loss factor is extracted from the dynamic relaxation modulus sequence. Multi-scale wavelet transform is used to decompose the dielectric loss factor in the time and frequency domain to obtain the wavelet coefficient matrix at different scales. The local extrema at each scale are located as the characteristic peaks in the time and frequency domain. Based on the time-frequency domain feature peaks and the dynamic relaxation modulus sequence, time-series alignment is performed, and the contribution weight of each feature peak to the relaxation modulus is calculated through an attention mechanism to generate weighted feature-modulus association pairs and construct an initial nonlinear mapping relationship. The weighted feature-modulus correlation pairs are input into a deep neural network, and the network parameters are iteratively optimized using the fitting error of the correlation pairs as the loss function. After convergence, a mechanical-dielectric correlation model with dynamic weight allocation capability is output.

9. The method for detecting the sealing strength of a sterile barrier based on servo tensile force according to claim 1, characterized in that: In step S6, the specific steps to form a closed loop of data throughout the entire process are as follows: S61: Based on the interface-combined defect distribution derived from the long-term creep failure lifetime confidence interval and dielectric time-frequency domain characteristics, a multi-dimensional intensity mapping feature tensor is constructed, and a creep constitutive correction factor is embedded for tensor normalization to obtain the normalized intensity mapping feature tensor. S62: Input the normalized intensity mapping feature tensor into the mechanical-dielectric correlation model, solve the equivalent value of the sterile barrier sealing strength through adaptive weight inference, and generate an intensity quantization spectrum with error traceability; S63: Based on the intensity quantization spectrum, connect the original data, model parameters and verification results of the entire process, establish a blockchain-style evidence storage link to complete data verification and form a closed data chain for the entire process.

10. A servo-tension-based sterile barrier seal strength testing system, which employs the servo-tension-based sterile barrier seal strength testing method as described in any one of claims 1 to 9, characterized in that, The detection system includes: The sample preparation module is used to cut sterile sealed samples, attach flexible interdigitated electrodes to their sealing interface and connect them to an impedance analyzer to form an initial test sample. The benchmark construction module is used to apply a constant displacement constraint to the initial test sample, activate the force value acquisition state, and integrate the sample geometric features and circuit benchmark parameters to construct a constant mechanical test benchmark dataset. The synchronous acquisition module is used to set parameters in the impedance analyzer according to the constant mechanical test benchmark dataset, start the temperature control module to stabilize the temperature, trigger synchronous acquisition, match the real-time servo force value with the dielectric parameter, and obtain the same source time sequence detection data. The dataset generation module is used to continuously collect the same source time-series detection data until the dielectric parameters tend to stabilize, and to generate the stress decay changes and the evolution information of dielectric constant, loss factor and impedance modulus within the collection period to form a mechanical-dielectric original time-series dataset. The life prediction module is used to extract stress relaxation data from the mechanical-dielectric original time series dataset, calculate the relaxation modulus, fit the relaxation modulus with the characteristic peak value of the dielectric loss factor, construct a mechanical-dielectric correlation model and substitute it with a preset creep failure threshold, and predict the long-term creep failure life of the seal through short-term test data. The strength closed-loop module is used to calculate the measured equivalent value of the sterile barrier sealing strength based on the interface bonding state derived from the long-term creep failure life and dielectric parameters, through a mechanical-dielectric correlation model mapping, thus forming a closed-loop data system for the entire process.

Images