Online detection method of PET film crystallinity using dielectric properties

Abstract

This invention discloses an online method for detecting the crystallinity of PET films using dielectric properties. The method comprises the following steps: acquiring raw dielectric data and real-time temperature data, performing parallel acquisition and processing to obtain spatiotemporal sampling data, performing noise identification and removal on the data, and employing contact compensation to obtain high-quality dielectric characteristic data; performing temperature-frequency coupling effect separation and crystal form feature identification on the high-quality dielectric characteristic data; establishing a multidimensional feature space through polarization mechanism quantification and microstructural feature extraction; fusing physical models and deep learning methods to predict crystallinity; and finally, obtaining the final crystallinity value through consistency testing and dynamic correction. Through multi-level data processing and model fusion strategies, the present invention achieves high-precision online detection of PET film crystallinity, demonstrating strong anti-interference capabilities and predictive accuracy.

Description

Technical Field

[0001] The invention belongs to the field of crystallinity detection of polymer materials, in particular to an online detection method for the crystallinity of a PET film using dielectric properties. Background Art

[0002] Polyethylene terephthalate (PET) film is widely used in electronics, packaging, and optics due to its excellent mechanical properties, electrical insulation, and dimensional stability. The performance of PET film is largely determined by its crystallinity, which not only affects the material's mechanical properties and thermal stability but also determines its gas barrier properties and optical transparency. Traditional offline detection methods such as differential scanning calorimetry (DSC) and X-ray diffraction (XRD), while highly accurate, cannot meet the real-time quality control requirements of modern production. Therefore, the development of reliable online detection methods is crucial for improving PET film production efficiency and product quality.

[0003] Currently, online detection methods based on dielectric properties are attracting widespread attention due to their non-contact and fast response characteristics. Existing technologies mainly use single-frequency measurement or simple frequency scanning methods to estimate the dielectric constant or loss tangent by establishing an empirical relationship between the dielectric constant and crystallinity. Some researchers use multi-frequency measurement to improve detection accuracy and introduce temperature compensation algorithms to reduce environmental impact. Other studies have attempted to establish a correlation between dielectric response and crystal structure by establishing simple physical models. However, these methods generally suffer from problems such as low detection accuracy and limited applicability.

[0004] The current technology has the following specific problems: First, in the data acquisition link, existing methods find it difficult to effectively deal with measurement errors caused by changes in contact interface impedance, especially on high-speed production lines, where the dynamic changes in the contact state between the diaphragm and the electrode seriously affect measurement stability. Second, in data processing, there is currently a lack of effective methods to separate the combined effects of temperature, orientation, and crystal effects on the dielectric response, resulting in measurement results being interfered with by multiple factors. Third, in structural characterization, existing methods are unable to accurately identify and quantitatively characterize the dynamic evolution of α-crystalline, β-crystalline, and their mixed states, especially when the crystallinity changes rapidly, and it is difficult to capture the transient characteristics of the microstructure. Fourth, in modeling and prediction, a single empirical model or physical model is difficult to accurately describe the complex relationship between dielectric response and crystal structure, especially when considering multi-scale structural features, and the prediction accuracy often cannot meet industrial needs. In addition, existing methods are not robust enough in dealing with abnormal working conditions and interference conditions, and lack a reliable prediction result evaluation and correction mechanism. Summary of the Invention

[0005] The purpose of the invention is to provide an online detection method for the crystallinity of PET film using dielectric properties, in order to solve at least one technical problem existing in the prior art.

[0006] The technical solution is an online detection method for the crystallinity of PET films using dielectric properties, comprising the following steps:

[0007] S1. Acquire raw dielectric data and real-time temperature data, obtain spatiotemporal sampling data and sampled temperature data through parallel acquisition and processing; identify and remove noise from the spatiotemporal sampling data to obtain de-noised data; use contact compensation to process the de-noised data and sampled temperature data to obtain high-quality dielectric characteristic data;

[0008] S2. Separate the temperature-frequency coupling effect and identify the crystal form characteristics of the high-quality dielectric characteristic data to obtain a decoupled characteristic vector;

[0009] S3. Based on the decoupled eigenvector, a multidimensional feature space is established through polarization mechanism quantification processing and microstructure feature extraction to obtain multidimensional feature space data;

[0010] S4. Based on the multi-dimensional feature space data, the pre-configured physical model and deep learning method are integrated to predict the crystallinity and obtain a preliminary crystallinity prediction value;

[0011] S5. Based on the preliminary crystallinity prediction value, the final crystallinity value is obtained through consistency test and dynamic correction.

[0012] According to one aspect of the present application, step S1 further comprises:

[0013] S11, obtaining raw dielectric data and real-time temperature data; constructing sampling parameters according to a preset sampling interval and sampling position; performing parallel acquisition and processing on the raw dielectric data and real-time temperature data to generate spatiotemporal sampling data and sampled temperature data;

[0014] S12. Based on the spatiotemporal sampling data, calculate the wavelet energy distribution characteristics and construct noise characteristic data; process the noise characteristic data using an adaptive threshold method to obtain processed noise characteristic data; and perform noise reduction processing on the spatiotemporal sampling data based on the processed noise characteristic data to obtain noise reduction data;

[0015] S13. Acquire contact state parameters in real time; construct a contact impedance model based on the contact state parameters and sampled temperature data, calculate the compensation coefficient and combine it with the noise reduction data to generate high-quality dielectric characteristic data.

[0016] According to one aspect of the present application, step S2 is further:

[0017] S21. Based on the high-quality dielectric characteristic data, construct a temperature-frequency relationship matrix and calculate the temperature sensitivity coefficient and frequency response function; based on the temperature sensitivity coefficient and frequency response function, perform temperature-frequency coupling effect separation processing on the high-quality dielectric characteristic data to generate decoupled dielectric data;

[0018] S22. Based on the decoupled dielectric data, construct crystal form characteristic parameters, calculate the dielectric relaxation characteristics and mixed state characteristics of the α crystal form and the β crystal form; calculate the crystal form distribution based on the dielectric relaxation characteristics and mixed state characteristics, and generate crystal form characteristic data; wherein α and β represent different crystal forms;

[0019] S23. Based on the decoupled dielectric data and crystal characteristic data, the molecular chain orientation function is calculated to construct a relationship model between orientation and dielectric properties; based on the relationship model, the orientation effect of the decoupled dielectric data is separated to generate a decoupled characteristic vector.

[0020] According to one aspect of the present application, step S3 is further:

[0021] S31. Based on the decoupled eigenvector, construct a polarization component matrix including electronic polarization, ionic polarization, and orientation polarization; calculate the weight and intensity of each polarization component in the polarization component matrix to generate polarization characteristic data;

[0022] S32. Based on the polarization characteristic data and the crystal form characteristic data, construct structural characteristic parameters, calculate crystal size characteristics, crystal integrity characteristics and interface characteristics, and generate microstructure characteristic data;

[0023] S33. Based on the microstructure feature data, a high-dimensional feature space is constructed and nonlinear dimensionality reduction processing is performed to generate multidimensional feature space data.

[0024] According to one aspect of the present application, step S4 is further:

[0025] S41. Based on the multi-dimensional feature space data, construct and solve the physical constraint equation to generate the physical model prediction value;

[0026] S42, based on multi-dimensional feature space data, data processing is performed through deep neural networks and attention mechanisms to generate deep learning prediction values;

[0027] S43. Calculate the confidence weights of the physical model prediction value and the deep learning prediction value respectively; based on the confidence weights, perform weighted fusion processing on the physical model prediction value and the deep learning prediction value to generate a preliminary crystallinity prediction value.

[0028] According to one aspect of the present application, step S5 is further:

[0029] S51. Based on the preliminary crystallinity prediction value, calculate the historical consistency score and the physical constraint score and generate consistency score data through weighted processing;

[0030] S52. Based on the preliminary crystallinity prediction value and the consistency score data, a correction function is constructed; based on the correction function, the preliminary crystallinity prediction value is dynamically corrected to generate a final crystallinity value.

[0031] According to one aspect of the present application, step S11 is further as follows:

[0032] S111, obtaining original dielectric data and real-time temperature data, calculating difference sequences, constructing probability density distribution functions, determining and optimizing sampling step lengths, and generating sampling interval parameters;

[0033] S112, dividing the measurement space based on the sampling interval parameter and the original dielectric data, calculating the signal gradient and curvature, determining the basic grid and performing local encryption, and generating sampling position data;

[0034] S113. Based on the sampling location data, a task priority queue is constructed, the acquisition tasks are grouped, and an execution time window is calculated; based on the execution time window, a clock synchronization benchmark is constructed, multi-channel data acquisition is performed, and multi-channel acquisition data is generated;

[0035] S114, extracting timestamp information based on multi-channel acquisition data, building a unified time base, calculating time deviation and generating aligned acquisition data using a cubic spline function;

[0036] S115. Calculate local statistical features of the aligned collected data, identify outliers, and construct a correlation matrix; integrate the correlation matrix with the real-time temperature data to generate spatiotemporal sampling data and sampling temperature data.

[0037] According to one aspect of the present application, step S12 is further as follows:

[0038] S121, grouping the spatiotemporal sampling data according to frequency, performing multi-scale decomposition to obtain scale coefficients; calculating the energy value under each scale coefficient and generating energy distribution characteristics;

[0039] S122. Based on the energy distribution characteristics, calculate the local fluctuation index, identify the abnormal fluctuation interval, extract the time-frequency characteristic parameters, and generate noise characteristic data;

[0040] S123, performing cluster analysis on noise feature data, calculating energy concentration, constructing noise determination criteria, and generating noise determination results;

[0041] S124, performing time-frequency decomposition on the spatiotemporal sampling data based on the noise determination result, calculating the local correlation coefficient, constructing the signal reconstruction weight, and generating preliminary noise reduction data;

[0042] S125. Based on the preliminary noise reduction data, calculate the local smoothness index of the signal, identify the residual noise points and perform local correction to obtain a correction result; integrate the preliminary noise reduction data and the correction result to generate noise reduction data;

[0043] S126 , time-aligning the sampled temperature data with the noise reduction data, and outputting the aligned noise reduction data and sampled temperature data.

[0044] According to one aspect of the present application, step S22 is further as follows:

[0045] S221. Based on the decoupled dielectric data, extract characteristic frequency points of the dielectric spectrum, calculate the dielectric constant and loss value, and generate crystal parameter data; combine the data with the decoupled dielectric data to calculate the dielectric loss peak shape parameters, construct a time-temperature equivalent curve of the loss peak, and generate relaxation characteristic data;

[0046] S222. Based on the relaxation characteristic data, decompose the relaxation spectrum into basic relaxation units, calculate the orientation correlation function of the α crystal form, and generate α crystal form characteristic data;

[0047] S223. Based on the relaxation characteristic data, identify the characteristic frequency interval of the β crystal form, construct the local order parameter, and generate the β crystal form characteristic data;

[0048] S224, calculating the overlap of the characteristic data of the α and β crystal forms, identifying the mixed state region, constructing the order parameter tensor and calculating the stability index, and generating the mixed state characteristic data;

[0049] S225. Based on the characteristic data of the α-crystalline form, the β-crystalline form, and the mixed state, a multidimensional characteristic space is constructed, a distribution density function is calculated, characteristic clusters are identified, and transition probabilities are calculated to generate preliminary distribution data;

[0050] S226. Based on the preliminary distribution data, calculate the crystal form correlation, construct a free energy landscape, identify and optimize the local distribution abnormal areas, and obtain optimized local distribution data; integrate the preliminary distribution data and the optimized local distribution data to generate crystal form characteristic data.

[0051] According to one aspect of the present application, step S32 is further as follows:

[0052] S321. Based on the polarization characteristic data and the crystal form characteristic data, extract the polarization time constant sequence and calculate the polarization relaxation time distribution to generate structural parameter data;

[0053] S322. Based on the structural parameter data, calculate the characteristic length distribution at each scale, construct a scale correlation function and a hierarchical structure, and generate size characteristic data;

[0054] S323. Based on the size characteristic data and the crystal form characteristic data, the order parameter of the local structure is calculated, a spatial distribution function is constructed, the defect position and type are extracted, and the integrity characteristic data is generated;

[0055] S324. Based on the integrity characteristic data, calculate the interface energy, construct the orientation distribution function, and extract the interface curvature and roughness to generate interface characteristic data;

[0056] S325. Based on the interface characteristic data, calculate the interface stress distribution, construct the stress-orientation coupling relationship, extract the stress concentration area, and generate the interface stress data;

[0057] S326. Based on the size characteristic data, integrity characteristic data, interface characteristic data and interface stress data, a characteristic parameter correlation matrix is ​​constructed, the coupling strength is calculated, and the microstructure characteristic data is generated.

[0058] Beneficial effects: The present invention ensures the quality of original data, realizes the separation of multiple factors such as temperature, crystal form and orientation, and establishes a multidimensional characterization space including polarization characteristics and microstructural characteristics; it integrates the advantages of physical models and deep learning methods, and ensures the reliability of the final results through consistency testing and dynamic correction; it not only solves the accuracy problem of online detection of PET film crystallinity, but also provides an interpretable structure-performance relationship analysis method, which is of great value to improving the level of product quality control. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 Flowchart of the present invention.

[0060] Figure 2 This is a flow chart of step S1 of the present invention.

[0061] Figure 3 This is a flow chart of step S2 of the present invention.

[0062] Figure 4 This is a flow chart of step S3 of the present invention.

[0063] Figure 5 This is a flow chart of step S4 of the present invention.

[0064] Figure 6 This is a flow chart of step S5 of the present invention. DETAILED DESCRIPTION

[0065] The following describes the present application in more detail with reference to specific embodiments. Figure 1 As shown, the present application proposes an online detection method for the crystallinity of a PET film using dielectric properties, comprising the following steps:

[0066] S1. Acquire raw dielectric data and real-time temperature data, and obtain spatiotemporal sampling data and sampled temperature data through parallel data acquisition and processing; perform noise identification and removal on the spatiotemporal sampling data to obtain de-noised data; use contact compensation to process the de-noised data and sampled temperature data to obtain high-quality dielectric characteristic data; wherein the raw dielectric data includes dielectric constant and dielectric loss data at predetermined frequency points;

[0067] S2. Based on the high-quality dielectric characteristic data, a temperature and frequency relationship is constructed; based on the temperature and frequency relationship, coupling effects are separated on the high-quality dielectric characteristic data to obtain decoupled dielectric data; crystal form characteristics are identified on the decoupled dielectric data to obtain crystal form characteristic data; based on the decoupled dielectric data and crystal form characteristic data, orientation effects are separated to obtain a decoupled characteristic vector;

[0068] S3. Based on the decoupled characteristic vector, polarization mechanism quantification processing is performed to obtain polarization characteristic data; based on the polarization characteristic data and the crystal characteristic data, microstructure characteristics are extracted to obtain microstructure characteristic data; feature space reconstruction is performed on the microstructure characteristic data to obtain multidimensional feature space data;

[0069] S4. Based on the multidimensional feature space data, a physical model prediction value is obtained through a preconfigured physical model calculation; based on the multidimensional feature space data, a deep learning method is used to obtain a deep learning prediction value; and the physical model prediction value and the deep learning prediction value are integrated to obtain a preliminary crystallinity prediction value;

[0070] S5. Based on the preliminary crystallinity prediction value, a consistency test is performed to obtain consistency score data; based on the preliminary crystallinity prediction value and the consistency score data, a dynamic correction is performed to obtain the final crystallinity value.

[0071] like Figure 2 As shown, according to one aspect of the present application, step S1 is further:

[0072] S11. Acquire raw dielectric data and real-time temperature data; construct sampling parameters according to a preset sampling interval and sampling position; based on the sampling parameters, perform parallel acquisition and processing on the raw dielectric data and real-time temperature data to generate spatiotemporal sampling data and sampled temperature data; wherein the raw dielectric data includes dielectric constant and dielectric loss data at predetermined frequency points;

[0073] S12. Calculating wavelet energy distribution characteristics based on the spatiotemporal sampling data; constructing noise characteristic data based on the wavelet energy distribution characteristics; processing the noise characteristic data using an adaptive threshold method to obtain processed noise characteristic data; and performing noise reduction processing on the spatiotemporal sampling data using a signal transformation method based on the processed noise characteristic data to obtain noise-reduced data;

[0074] S13. Acquire contact state parameters in real time; construct a contact impedance model based on the contact state parameters and sampled temperature data, and calculate compensation coefficients; generate high-quality dielectric characteristic data based on the compensation coefficients and noise reduction data.

[0075] In one embodiment of the present application, an original dielectric data sequence D(t) is obtained, comprising {f[i], ε'[i], tanΔ[i]}, i=1...n, and a sampling matrix H(t) is constructed, comprising the following elements: a frequency sampling interval vector h1, a time sampling interval vector h2, and a spatial sampling position vector h3, to obtain a spatiotemporal sampling data matrix M(t). Based on the spatiotemporal sampling data matrix M(t), the wavelet energy entropy E = -∑(p[i]log(p[i])) is calculated, where p[i] is the energy probability at the i-th scale; a noise feature map N(f, t) is constructed, an adaptive threshold method is applied to identify noise, and the Hilbert-Huang transform is used to remove noise, to obtain a denoised data matrix Mc(t). Based on the denoised data matrix Mc(t) and the contact state parameter λ(t), the contact impedance model Z(t) = R(t) + jX(t) is established, the compensation coefficient K(t) = f(Z(t), λ(t)) is calculated, and the compensation is applied: Mf(t) = Mc(t) × K(t), outputting high-quality dielectric characteristic data Mf(t). Where f[i] is the frequency point, ε'[i] is the real part of the dielectric constant corresponding to the frequency point, tanΔ[i] is the dielectric loss tangent corresponding to the frequency point, R(t) is the resistance component in the contact impedance model, j is the imaginary unit, X(t) is the reactance component in the contact impedance model, and f( ) is a function of the compensation coefficient.

[0076] In another embodiment of the present application, the denoised data is Y(t, s) = X(t, s) - N(t, s) M(t, s), where Y(t, s) is the denoised data, N(t, s) is the noise characteristic function, and M(t, s) is the noise mask function.

[0077] The contact impedance compensation coefficient is C(x, y, t) = α(T)·R(p)·Z(f); where C(x, y, t) is the compensation coefficient at position (x, y) and time t; α(T) = exp(-E_a / kT) is the temperature correction term; R(p) = (1 - exp(-βp)) / (1+ γp) is the pressure response function; Z(f) = Z_0 / (1 + jωτ) is the frequency characteristic function; E_a is the contact activation energy; k is the Boltzmann constant; T is the absolute temperature; p is the contact pressure; β and γ are pressure coefficients; Z_0 is the standard contact impedance; τ is the characteristic relaxation time; ω is the angular frequency.

[0078] Contact compensation processing Z(t, s) = Y(t, s) C(p, T, f); where Z(t, s) is the high-quality dielectric characteristic data after compensation; C(p, T, f) is the compensation coefficient function; p is the contact pressure; T is the temperature; and f is the measurement frequency.

[0079] This embodiment achieves the acquisition of high-quality dielectric characteristic data through parallel data acquisition and a multi-level data processing strategy. Specifically, by first parallelizing the acquisition and processing of raw dielectric and temperature data, the temporal and spatial synchronization and integrity of the data are ensured. Then, adaptive noise identification and removal are performed using wavelet energy distribution characteristics, improving the signal-to-noise ratio. Finally, compensation processing is performed using a contact impedance model, effectively eliminating measurement errors caused by changes in contact state. This embodiment not only improves the reliability of data acquisition but also enhances measurement accuracy through compensation processing. In particular, in the face of various interference factors in the dynamic production process of PET film, the processing strategy can be adaptively adjusted to ensure the quality of the output data, providing a reliable data foundation for subsequent crystallinity analysis.

[0080] According to one aspect of the present application, step S11 is further as follows:

[0081] S111, obtaining original dielectric data and real-time temperature data, calculating a difference sequence of data at adjacent time points; constructing a probability density distribution function of the difference sequence; determining an initial sampling step length based on the variance of the probability density distribution function; adjusting the initial sampling step length through an iterative optimization method until the correlation coefficient of adjacent sampling points is less than a preset threshold, thereby generating a sampling interval parameter;

[0082] S112, based on the sampling interval parameter and the original dielectric data, the measurement space is divided into a predetermined number of sub-regions; the gradient value and curvature value of the signal in each sub-region are calculated; based on the gradient value, the base grid size is determined to obtain a preliminary grid; the preliminary grid is locally encrypted and thinned using the curvature value to obtain a processed grid; based on the processed grid, the encrypted area is secondary divided to obtain a subdivided grid; the information of the preliminary grid, the processed grid, and the subdivided grid is integrated to generate sampling position data;

[0083] S113: Constructing a task priority queue based on the sampling position data and the original dielectric data; grouping the acquisition tasks in the task priority queue according to the spatial proximity of the sampling position data to obtain grouped acquisition tasks; calculating the execution time window of each group of tasks based on the grouped acquisition tasks to generate a task scheduling sequence; establishing a clock synchronization benchmark between multiple channels, and executing data acquisition based on the clock synchronization benchmark and the task scheduling sequence to generate multi-channel acquisition data;

[0084] S114. Based on the multi-channel collected data and the real-time temperature data, extract the timestamp information of each data point and construct a unified time base; based on the unified time base, calculate the time deviation between the data of each channel and construct a piecewise continuous cubic spline function; use the cubic spline function to calculate the data value at the unified time point to generate aligned collected data;

[0085] S115. Calculate local statistical features of the aligned collected data, including mean and standard deviation; identify outliers based on the local statistical features, and construct a correlation matrix of time and space dimensions; integrate the correlation matrix with the real-time temperature data to generate spatiotemporal sampling data and sampling temperature data.

[0086] In one embodiment of the present application, the spatiotemporal sampling data is X(t, s) = ∑[D(t_i, s_j)·W(t-t_i, s-s_j)]; where D(t_i, s_j) is the original dielectric data; W(t, s) is the sampling weight function; t_i is the time sampling point; and s_j is the spatial sampling point.

[0087] This embodiment achieves efficient and reliable data acquisition. Specifically, the optimal sampling interval is determined by analyzing the probability density distribution of the data difference sequence. Then, spatial grid division is performed based on gradient and curvature values ​​to achieve optimal configuration of sampling locations. Task priority queues and multi-channel clock synchronization are then used to ensure the timing accuracy of data acquisition. Finally, spline function interpolation and outlier identification are used to generate high-quality spatiotemporal sampling data. This embodiment not only improves data acquisition efficiency but also ensures the representativeness and reliability of the sampled data through multiple optimizations.

[0088] According to one aspect of the present application, step S12 is further as follows:

[0089] S121. Grouping the spatiotemporal sampling data according to frequency to obtain grouped data; performing multi-scale decomposition on each group of grouped data to obtain different scale coefficients; calculating the energy value under each scale coefficient to construct an energy distribution matrix; based on the energy distribution matrix, calculating the energy proportion of each scale coefficient to generate an energy distribution feature;

[0090] S122. Based on the energy distribution characteristics, calculate the local fluctuation index of each scale coefficient; construct a cumulative distribution function of the local fluctuation index to identify the abnormal fluctuation interval; extract the time-frequency characteristic parameters of the abnormal fluctuation interval, construct a mapping relationship between the time-frequency characteristic parameters and the preset background noise, and generate noise characteristic data;

[0091] S123, performing cluster analysis on the noise feature data to obtain noise categories; based on the noise categories, calculating the energy concentration of each type of noise; constructing a noise determination criterion based on the energy concentration, calculating the noise probability of each frequency point according to the noise determination criterion, and generating a noise determination result;

[0092] S124. Based on the noise determination result, perform time-frequency decomposition on the spatiotemporal sampling data to obtain a time-frequency decomposition result; based on the time-frequency decomposition result, calculate the local correlation coefficient of each time-frequency point; construct a signal reconstruction weight based on the local correlation coefficient; apply the signal reconstruction weight to the time-frequency decomposition result, perform an inverse transform to reconstruct the signal, and generate preliminary noise reduction data;

[0093] S125. Based on the preliminary noise reduction data, calculate the local smoothness index of the signal and identify the residual noise points; construct a correction function containing neighborhood information to perform local correction on the residual noise points to obtain a correction result; integrate the preliminary noise reduction data and the correction result to generate the noise reduction data;

[0094] S126 , time-aligning the sampled temperature data with the noise reduction data, keeping the sampled temperature data unchanged, and outputting the aligned noise reduction data and sampled temperature data.

[0095] This embodiment achieves efficient noise identification and removal through multi-scale energy analysis and adaptive noise reduction processing. First, energy distribution characteristics are calculated through multi-scale decomposition; noise characteristics are then identified based on fluctuation indicators; cluster analysis is then used to determine noise categories and judgment criteria; signal reconstruction is then performed through time-frequency decomposition and local correlation coefficient calculation; and finally, residual noise is eliminated through local correction. This embodiment not only effectively identifies and removes different types of noise, but also maintains the integrity of the valid signal through a multi-level processing strategy, improving the accuracy of subsequent analysis.

[0096] According to one aspect of the present application, step S13 is further as follows:

[0097] S131. Receive contact state parameters and sampled temperature data, calculate the spatial uniformity index of pressure distribution, establish a corresponding relationship between pressure distribution and contact area, discretize the contact area into multiple micro-area units, calculate the effective contact rate of each micro-area, and generate contact feature data.

[0098] S132, receiving contact characteristic data and sampled temperature data, calculating equivalent circuit parameters of each micro-area, establishing a parallel equivalent circuit including contact resistance and contact capacitance, calculating the influence coefficient of temperature on circuit parameters, and integrating all parameters to generate impedance model parameters.

[0099] S133. Receive impedance model parameters and noise reduction data, calculate the standard response curve under the ideal contact state, extract the deviation characteristics between the actual measurement data and the standard curve, construct a mapping relationship between the deviation characteristics and the compensation factor, and generate a preliminary compensation coefficient.

[0100] S134, receiving preliminary compensation coefficients, calculating spatial distribution characteristics of the compensation coefficients, identifying compensation abnormality areas, constructing a correction function based on physical constraints, correcting the compensation coefficients of the abnormal areas, and generating optimized compensation coefficients.

[0101] S135, receiving the optimized compensation coefficient and noise reduction data, processing the data by frequency segment, calculating the compensation weight of each frequency band, performing segmented compensation operation, performing continuity smoothing processing on the compensation result, and generating compensation processing data.

[0102] S136. Receive the compensated processed data, calculate the quality evaluation indicators of the data, including signal-to-noise ratio, consistency coefficient and physical rationality parameters, screen valid data according to the evaluation indicators, integrate all processing results, and generate high-quality dielectric characteristic data.

[0103] This embodiment effectively corrects measurement errors through precise contact feature modeling and compensation. First, the pressure distribution is analyzed to establish a micro-area contact model. Next, a parallel equivalent circuit is constructed that accounts for temperature effects. Compensation coefficients are calculated using standard response curves. Compensation optimization is performed using physical constraints. Finally, high-quality data is obtained through segmented compensation and smoothing. This embodiment not only considers the spatial distribution of contact states but also improves measurement accuracy through temperature compensation, effectively addressing measurement errors caused by varying contact states.

[0104] like Figure 3 As shown, according to one aspect of the present application, step S2 is further:

[0105] S21. Based on the high-quality dielectric characteristic data, construct a temperature-frequency relationship matrix and calculate the temperature sensitivity coefficient and frequency response function; based on the temperature sensitivity coefficient and frequency response function, perform temperature-frequency coupling effect separation processing on the high-quality dielectric characteristic data to generate decoupled dielectric data;

[0106] S22. Based on the decoupled dielectric data, construct crystal form characteristic parameters, calculate the dielectric relaxation characteristics and mixed state characteristics of the α crystal form and the β crystal form; calculate the crystal form distribution based on the dielectric relaxation characteristics and mixed state characteristics, and generate crystal form characteristic data; wherein α and β represent different crystal forms;

[0107] S23. Based on the decoupled dielectric data and crystal characteristic data, the molecular chain orientation function is calculated to construct a relationship model between orientation and dielectric properties; based on the relationship model, the orientation effect of the decoupled dielectric data is separated to generate a decoupled characteristic vector.

[0108] In one embodiment of the present application, high-quality dielectric characteristic data Mf(t) and temperature data T(t) are obtained, a temperature-frequency coupling matrix C(f, T) = {ε'(f, T), tanΔ(f, T)} is constructed, and the temperature sensitivity coefficient is calculated: ST = ΨC / ΨT, where Ψ is the partial derivative; the frequency response function is calculated: RF(f) = C(f, T) / C(f0, T); a decoupling calculation is performed: Cd(f, T) = C(f, T) - ST·ΔT - RF(f), and the decoupled dielectric data Cd(f, T) is output. Based on the decoupled dielectric data Cd(f, T), the crystal characteristic vector is constructed: V(α, β) = {v1, v2, v3}, where v1 is the dielectric relaxation characteristic of the α crystal; v2 is the dielectric relaxation characteristic of the β crystal; and v3 is the mixed state characteristic. The crystal distribution function is calculated: P(α, β) = g(V(α, β)), and the crystal characteristic data P(α, β) is output. Based on the decoupled dielectric data Cd(f, T) and the crystal characteristic data P(α, β), the molecular chain orientation function is calculated: O(θ) = ∫P(θ)cos 2 θdθ; construct the orientation-dielectric relationship: D(O) = h(Cd, O(θ)), separate the orientation effect: Cf(f, T) = Cd(f, T) / D(O), and output the decoupled eigenvector F(t). Where ε'(f, T) is the real part of the dielectric constant at frequency f and temperature T; tanΔ(f, T) is the dielectric loss tangent at frequency f and temperature T; C(f0, T) is the temperature-frequency coupling matrix at reference frequency f0 and temperature T, S is the temperature sensitivity coefficient, ΔT is the temperature change, and g( ) is the function used to calculate the crystal distribution function; P(θ) is the probability density function of the molecular chain orientation function, θ is the orientation angle of the molecular chain, dθ is the small change in the orientation angle, h( ) is the function used to construct the orientation-dielectric relationship, and Cf(f, T) is the decoupled dielectric data at frequency f and temperature T.

[0109] In another embodiment of the present application, the temperature-frequency coupling effect separation process is performed as follows: ε(T, f) = ε_0(f)·exp(-E(f) / kT); wherein ε(T, f) is the dielectric constant, ε_0(f) is the reference dielectric constant, E(f) is the frequency-related activation energy, k is the Boltzmann constant, and T is the absolute temperature.

[0110] The specific crystal characteristic parameters are constructed as follows: χ(α, β) = K_α·F_α(f) + K_β·F_β(f); where χ(α, β) is the crystal characteristic function; K_α and K_β are the crystal weight coefficients; F_α(f) and F_β(f) are the α and β crystal characteristic spectra.

[0111] This embodiment achieves the decoupling of various influencing factors in dielectric data by establishing a temperature-frequency coupling effect separation model and a crystal form feature identification method. First, a temperature-frequency relationship matrix is ​​constructed, and the temperature-frequency coupling effect is effectively separated by calculating the temperature sensitivity coefficient and frequency response function. Then, the dielectric relaxation characteristics and mixed state characteristics of the α-crystal and β-crystal forms are identified to accurately obtain crystal form distribution information. Finally, by establishing a relationship model between molecular chain orientation and dielectric properties, the orientation effect is separated. This embodiment not only improves the accuracy of crystallinity measurement, but also can separately obtain the independent effects of factors such as temperature, crystal form and orientation, providing important information for a deeper understanding of the crystallization behavior of PET films.

[0112] According to one aspect of the present application, step S22 is further as follows:

[0113] S221. Based on the decoupled dielectric data, extract characteristic frequency points of the dielectric spectrum; calculate the dielectric constant and loss value at the characteristic frequency points, construct a parameter matrix containing a crystal form discrimination threshold, and generate crystal form parameter data; based on the crystal form parameter data and the decoupled dielectric data, calculate the dielectric loss peak position at different frequencies; extract the shape parameters of the dielectric loss peak position, including peak width and peak height, construct a time-temperature equivalent curve of the loss peak, and generate relaxation characteristic data;

[0114] S222. Based on the relaxation characteristic data, decompose the relaxation spectrum into a predetermined number of basic relaxation units; calculate the characteristic time and intensity of each basic relaxation unit, and extract the response data of the characteristic relaxation time interval of the α crystal form; based on the response data, calculate the orientation correlation function of the α crystal form to generate the α crystal form characteristic data;

[0115] S223. Based on the relaxation characteristic data, identify the characteristic frequency interval of the β crystal form, calculate the relaxation intensity distribution within the characteristic frequency interval of the β crystal form; based on the relaxation intensity distribution, construct the local order parameter of the β crystal form, extract the conformational characteristic parameters of the β crystal form, and generate the β crystal form characteristic data;

[0116] S224. Calculate the degree of overlap between the α-crystal characteristic data and the β-crystal characteristic data, and identify the mixed state region based on the overlap; construct an order parameter tensor for the mixed state region; calculate the stability index of the mixed state based on the order parameter tensor, and generate mixed state characteristic data;

[0117] S225. Constructing a multidimensional feature space based on the α-crystal characteristic data, the β-crystal characteristic data, and the mixed state characteristic data; calculating a distribution density function in the multidimensional feature space to identify characteristic clusters of each crystal form; calculating transition probabilities between characteristic clusters to generate preliminary distribution data;

[0118] S226. Based on the preliminary distribution data, calculate the crystal form correlation of adjacent areas and construct the free energy landscape of crystal form transformation; based on the free energy landscape, identify and optimize the local distribution abnormal areas to obtain optimized local distribution data; integrate the preliminary distribution data and the optimized local distribution data to generate crystal form characteristic data.

[0119] In one embodiment of the present application, the crystal form characteristic data χ(f, T) = w_α·S_α(f, T) + w_β·S_β(f, T); wherein χ(f, T) is the comprehensive crystal form characteristic index; S_α(f, T) = A_α·exp[-(f-f_α) 2 / 2σ_α 2 ] is the α crystal spectrum characteristic; S_β(f, T) = A_β·exp[-(f-f_β) 2 / 2σ_β 2 ] is the β crystal spectrum characteristic; w_α, w_β are weight coefficients; f_α, f_β are characteristic frequencies; σ_α, σ_β are spectrum width parameters; A_α, A_β are amplitude coefficients.

[0120] This embodiment achieves precise analysis of different crystal components in PET film. First, a crystal discrimination parameter system is established through dielectric spectrum characteristic frequency analysis; then, characteristic information of the α crystal and β crystal is extracted separately through relaxation spectrum decomposition; then, the overlapping area of ​​the two crystals is analyzed to determine the mixed state characteristics; finally, a crystal distribution model is established through multidimensional feature space analysis. This embodiment achieves quantitative characterization of the orientation correlation of the α crystal through relaxation time distribution, characterizes the conformational characteristics of the β crystal through local order parameters, describes the stability of the mixed state through order parameter tensors, and predicts crystal transformation through free energy landscape analysis. This not only provides accurate quantification of crystal components, but also reveals the kinetic mechanism of crystal transformation, providing theoretical guidance for process optimization.

[0121] like Figure 4 As shown, according to one aspect of the present application, step S3 is further:

[0122] S31. Based on the decoupled eigenvector, construct a polarization component matrix including electronic polarization, ionic polarization, and orientation polarization; calculate the weight and intensity of each polarization component in the polarization component matrix to generate polarization characteristic data;

[0123] S32. Based on the polarization characteristic data and the crystal form characteristic data, construct structural characteristic parameters, calculate crystal size characteristics, crystal integrity characteristics and interface characteristics, and generate microstructure characteristic data;

[0124] S33. Based on the microstructure feature data, a high-dimensional feature space is constructed; based on the data of the high-dimensional feature space, nonlinear dimensionality reduction processing is performed to generate multi-dimensional feature space data.

[0125] In one embodiment of the present application, based on the decoupled eigenvector F(t), a polarization component matrix is ​​constructed: P(t) = {P1(t), P2(t), P3(t)}, where P1 is the electronic polarization component, P2 is the ionic polarization component, and P3 is the orientation polarization component; the polarization intensity is calculated: I(P) = ∑wi Pi(t), and the polarization characteristic data I(P) is output. Based on the polarization characteristic data I(P) and the crystal form characteristic data P(α, β), a structural characteristic tensor is constructed: S = {s1, s2, s3}, where s1 is the crystal size characteristic, s2 is the crystal integrity characteristic, and s3 is the interface characteristic; the topological feature is calculated: T(S) = k(S, I(P)), and the microstructure feature T(S) is output. Based on the microstructural features T(S), a high-dimensional feature space is constructed: H = {F(t), I(P), T(S)}, and nonlinear dimensionality reduction is applied: L = PCA(H) + t-SNE(H), outputting the multidimensional feature space data L. Here, wi is the weight coefficient, Pi(t) is the polarization component, k( ) is the topological characteristic function, PCA( ) is principal component analysis, and t-SNE( ) is t-distributed stochastic neighbor embedding.

[0126] In another embodiment of the present application, the polarization feature quantization P(t) = P_e + P_i + P_o; wherein P(t) is the total polarization intensity, P_e is the electronic polarization component, P_i is the ionic polarization component, and P_o is the orientation polarization component; the microstructure feature S(r) =∫[G(r, r')·ρ(r')]dr'; wherein S(r) is the structural characteristic function, G(r, r') is the structural correlation function, ρ(r') is the local density distribution at position r', and dr' is the integral variable at position r'.

[0127] This example establishes a multidimensional feature space characterization method by quantifying polarization mechanisms and extracting microstructural features. First, by constructing a polarization component matrix, quantitative characterization of electronic, ionic, and orientational polarization is achieved. Then, combined with crystal form characteristic data, microstructural features such as crystal size, integrity, and interfaces are extracted. Finally, through feature space reconstruction, a multidimensional feature space is established that comprehensively characterizes the crystalline state of PET films. This example not only provides complete information about the crystallization process but also retains key feature information through nonlinear dimensionality reduction, providing reliable feature input for subsequent crystallinity prediction.

[0128] According to one aspect of the present application, step S32 is further as follows:

[0129] S321. Extracting a polarization time constant sequence based on the polarization characteristic data and the crystal form characteristic data; calculating a polarization relaxation time distribution of the polarization time constant sequence; constructing a correspondence between relaxation time and structural scale based on the polarization relaxation time distribution, and generating structural parameter data;

[0130] S322. Based on the structural parameter data, calculate the characteristic length distribution at each scale and construct a scale correlation function; based on the scale correlation function, extract the hierarchical structure of the characteristic scale; establish the topological connection relationship of the hierarchical structure and generate size characteristic data;

[0131] S323. Based on the size characteristic data and the crystal form characteristic data, the order parameter of the local structure is calculated and a spatial distribution function of the order parameter is constructed; based on the spatial distribution function, the location and type information of the structural defects are extracted, a defect density distribution map is constructed, and integrity characteristic data is generated;

[0132] S324. Based on the integrity characteristic data, calculate the interface energy between the structural units and construct an interface orientation distribution function; based on the interface orientation distribution function, extract the curvature and roughness parameters of the interface, construct an interface connectivity network, and generate interface characteristic data;

[0133] S325. Based on the interface characteristic data, calculate the stress distribution state of the interface and construct a stress-orientation coupling relationship; based on the stress-orientation coupling relationship, extract the interface stress concentration area, establish stress relaxation parameters, and generate interface stress data;

[0134] S326. Based on the size characteristic data, integrity characteristic data, interface characteristic data and interface stress data, a correlation matrix of characteristic parameters is constructed to calculate the coupling strength between the parameters; based on the coupling strength, a multi-scale structural characteristic space is constructed to generate microstructure characteristic data.

[0135] In one embodiment of the present application, the polarization relaxation time distribution G(τ) = ∑[H_i·exp(-(ln(τ / τ_i)) 2 / 2B_i 2 )]; where G(τ) is the relaxation time distribution function, τ is the relaxation time, τ_i is the characteristic time of the i-th relaxation process, H_i is the intensity of the i-th relaxation process, and B_i is the distribution width parameter.

[0136] The integrity characteristic data I(r) =∫[ρ(r)·Φ(r)]dr; where I(r) is the structural integrity index, r is the spatial position vector, ρ(r) is the local density distribution function, and Φ(r) is the order parameter field.

[0137] This embodiment establishes a complete microstructure characterization system through multi-scale structural feature analysis and parameter coupling characterization. First, the structural scale correlation is established through polarization relaxation time distribution analysis; then the characteristic size distribution is determined through hierarchical structure analysis; then the structural integrity is characterized through order parameter analysis; then the interface characteristics are described through interface energy and stress analysis; finally, a comprehensive characterization model is established through multi-parameter coupling. This embodiment establishes a quantitative correspondence between polarization characteristics and structural characteristics, reveals the structural evolution law through multi-scale analysis, and clarifies the structural stability mechanism through interface feature analysis; it not only provides a comprehensive characterization of structural characteristics, but also reveals the structure-performance relationship through parameter coupling analysis, providing new ideas for material design.

[0138] like Figure 5 As shown, according to one aspect of the present application, step S4 is further:

[0139] S41. Constructing a physical constraint equation based on the multidimensional feature space data; solving a nonlinear system of the physical constraint equation to generate a physical model prediction value;

[0140] S42, based on multi-dimensional feature space data, data processing is performed through deep neural networks and attention mechanisms to generate deep learning prediction values;

[0141] S43. Calculate the confidence weights of the physical model prediction value and the deep learning prediction value respectively; based on the confidence weights, perform weighted fusion processing on the physical model prediction value and the deep learning prediction value to generate a preliminary crystallinity prediction value.

[0142] In one embodiment of the present application, based on the multidimensional feature space data L, a physical constraint equation is constructed: Φ(x) = {φ1(x), φ2(x), φ3(x)}, the nonlinear equation system is solved: x* = arg min||Φ(x)||, and the physical model prediction value x* is output. Based on the multidimensional feature space data L, a deep neural network is constructed: DNN(L)→y, an attention mechanism is applied: A(L) = softmax(W·L), the prediction value is calculated: y = DNN(A(L)), and the deep learning prediction value y is output. Based on the physical model prediction value x* and the deep learning prediction value y, the confidence weights are calculated: w1 = f1(x*), w2 = f2(y); the fusion prediction: z = w1·x*+ w2·y, and the preliminary crystallinity prediction value z is output. Where φ1(x), φ2(x), φ3(x) are different components in the physical constraint equation Φ(x); W is the weight matrix; DNN( ) is a deep neural network.

[0143] In another embodiment of the present application, the physical model predicts X_p = f(S, P, T) · [1 + Δ(t)]; where X_p is the physical model prediction value, f(S, P, T) is the physical model function, and Δ(t) is the correction term.

[0144] Deep learning prediction: X_d = ∑[wi h_i(F)]; where X_d is the deep learning prediction value, w_i is the neural network weight, h_i is the hidden layer activation function, and F is the feature input vector.

[0145] According to another aspect of the present application, the physical constraint equation is specifically constructed as follows: the basic equilibrium equation of crystallinity: X(t) = X_a(t) + X_b(t) + X_m(t); wherein X(t) is the total crystallinity, X_a(t) is the α-crystal crystallinity, X_b(t) is the β-crystal crystallinity, and X_m(t) is the amorphous region content, satisfying the constraint condition: X_a(t) + X_b(t) + X_m(t) = 1. The crystallization kinetic equation is: dX_a / dt = k_a(T)·[1 - X_a(t)]·f_a(σ) dX_b / dt = k_b(T)·[1 -X_b(t)]·f_b(σ); wherein k_a(T) and k_b(T) are temperature-dependent crystallization rate constants; f_a(σ) and f_b(σ) are stress-dependent functions; and σ is the local stress field. Dielectric response equation: ε(ω) = ε_∞ + (ε_s - ε_∞)·[X_a·φ_a(ω) + X_b·φ_b(ω)]; where ε(ω) is the complex dielectric constant; ε_∞ is the high-frequency dielectric constant; ε_s is the static dielectric constant; φ_a(ω) and φ_b(ω) are the crystal characteristic response functions.

[0146] The nonlinear system of equations used to solve the physical constraint equations is specifically constructed as follows: The residual function is constructed as follows: R_1(X) = |X_a + X_b + X_m - 1| R_2(X) = |ε_measured - ε_calculated| R_3(X) = |dX / dt - F(X, T, σ)|; where R_1 is the mass conservation residual, R_2 is the dielectric response residual, R_3 is the dynamic residual, ε_measured is the measured dielectric constant, ε_calculated is the calculated dielectric constant, and F(X, T, σ) is the dynamic function representing the dynamic behavior of the system under state X, temperature T, and stress σ. The objective function is optimized as follows: J(X) = w_1·R_1(X) + w_2·R_2(X) + w_3·R_3(X); where w_1, w_2, and w_3 are weight coefficients. The solution strategy is to use the improved Levenberg-Marquardt algorithm: X_(k+1) = X_k - [H + λI] -1 ▽J(X_k); where H is the Hessian matrix, λ is the damping factor, ▽J is the gradient vector, and I is the identity matrix.

[0147] The physical model predictions are generated using the crystallinity prediction equation: X_predict = g(X_a, X_b, S); where g is the prediction function and S is the structural characteristic parameter. The predictions are corrected using the equation: X_final = X_predict [1 + γ(T) Δ(σ)]; where γ(T) is the temperature correction function and Δ(σ) is the stress correction function.

[0148] This example achieves high-precision crystallinity prediction by integrating physical models and deep learning methods. First, the physical model prediction is obtained by solving the physical constraint equations, ensuring the physical rationality of the prediction results. Then, deep learning predictions are obtained through processing with a deep neural network and an attention mechanism, fully leveraging the implicit patterns in the data. Finally, confidence weights are used to fuse the results, combining the advantages of both methods. This example not only improves the accuracy of the predictions, but also ensures the interpretability and reliability of the prediction results through the constraints of the physical model.

[0149] According to one aspect of the present application, step S42 is further as follows:

[0150] S421. Receive multi-dimensional feature space data, calculate the correlation matrix between feature vectors, extract the temporal dependency of features, construct a multi-scale feature pyramid, calculate feature weights for each scale level, and generate hierarchical feature data.

[0151] S422. Receive hierarchical feature data, construct long-range and short-range dependency graphs, calculate the information flow intensity between nodes, extract key information transmission paths, establish a feature propagation network, and generate feature dependency data.

[0152] S423. Receive feature dependency data, calculate the importance score of each feature node, construct an adaptive attention weight matrix, extract the correlation pattern of cross-scale features, establish a dynamic attention map, and generate attention feature data.

[0153] S424. Receive attention feature data, construct a multi-layer perception scene graph, calculate the activation state of scene nodes, extract conversion rules between scenes, establish a scene evolution model, and generate scene representation data.

[0154] S425: Receive scene characterization data, establish a multi-channel predictor group, calculate the prediction results of each channel, extract the uncertainty distribution of the prediction results, build a confidence assessment model, and generate multi-channel prediction data.

[0155] S426. Receive multi-channel prediction data, calculate the consistency index of each channel prediction, construct prediction fusion weights, extract the optimal prediction path, establish a prediction calibration model, and generate deep learning prediction values.

[0156] This embodiment achieves high-precision crystallinity prediction. First, a hierarchical feature representation is established through a multi-scale feature pyramid; then a feature propagation network is established through dependency graph analysis; then feature weight optimization is achieved through an adaptive attention mechanism; a dynamic prediction model is established through scene graph analysis; and finally, the final prediction result is obtained through multi-channel fusion. This embodiment achieves effective integration of multi-scale information through a feature pyramid, captures long- and short-range correlations between features through a dependency graph, increases the weight of key features through a dynamic attention mechanism, and achieves dynamic updates of predictions through a scene evolution model; not only does it improve the accuracy of predictions, but it also improves the reliability of predictions through multi-channel design, providing a practical solution for industrial applications.

[0157] like Figure 6 As shown, according to one aspect of the present application, step S5 is further:

[0158] S51. Calculate a historical consistency score and a physical constraint score based on the preliminary crystallinity prediction value; generate consistency score data through weighted processing based on the historical consistency score and the physical constraint score;

[0159] S52. Based on the preliminary crystallinity prediction value and the consistency score data, a correction function is constructed; based on the correction function, the preliminary crystallinity prediction value is dynamically corrected to generate a final crystallinity value.

[0160] In one embodiment of the present application, based on the preliminary crystallinity prediction value z, the historical consistency score is calculated: H(z) = d(z, z_hist); the physical constraint score is calculated: P(z) = c(z, Φ); the comprehensive score is calculated: S(z) = w1·H(z) +w2·P(z), and the consistency score S(z) is output. Based on the preliminary crystallinity prediction value z and the consistency score S(z), a correction function is constructed: R(z) = g(z, S(z)), and the correction is applied: zf = R(z), and the final crystallinity value zf is output. Wherein z_hist is the historical crystallinity data, d( ) is the distance function, c( ) is the physical constraint function, Φ is the physical constraint condition, w1 and w2 are different weight coefficients, and g( ) is the correction function.

[0161] In another embodiment of the present application, the consistency score Q(t) = α·H(t) + β·V(t); where Q(t) is the consistency score, H(t) is the historical consistency component, and V(t) is the physical constraint component; α and β are weight coefficients. The final crystallinity is calculated as: X_f = γ·X_p + (1-γ)·X_d + K·Q(t); where X_f is the final crystallinity value, γ is the model fusion coefficient, K is the correction coefficient, X_p is the physical model prediction, and X_d is the deep learning prediction.

[0162] According to another aspect of the present application, the historical consistency score is calculated as follows: Time series consistency function: H_t(X) = exp[-α·(X(t) - X_pred(t)) 2 / σ_t 2 ]; where H_t(X) is the temporal consistency score, X(t) is the current predicted value, X_pred(t) is the predicted value based on historical data, α is the decay coefficient, and σ_t is the standard deviation of temporal fluctuations. Spatial consistency function: H_s(X) = exp[-β·∑(X(r) - X(r+Δr)) 2 / σ_s 2 ]; where H_s(X) is the spatial consistency score, X(r) is the predicted value at position r, Δr is the spatial displacement vector, β is the spatial weight coefficient, and σ_s is the standard deviation of spatial fluctuation. Comprehensive historical consistency score: H(X) = w_t·H_t(X) + w_s·H_s(X); where w_t and w_s are spatiotemporal weight coefficients.

[0163] The physical constraint score is calculated as follows: Thermodynamic constraint function: P_t(X) = exp[-γ·|ΔG(X)| / RT]; where P_t(X) is the thermodynamic score, ΔG(X) is the change in crystallization free energy, R is the gas constant, T is the absolute temperature, and γ is the scaling factor. Kinetic constraint function: P_k(X) = exp[-λ·|dX / dt - f(X, T)| 2 ]; where P_k(X) is the kinetic score, f(X, T) is the crystallization kinetics function, and λ is the kinetic weight coefficient. The comprehensive physical constraint score is: P(X) = w_th·P_t(X) + w_k·P_k(X); where w_th and w_k are the physical constraint weights. The consistency score data is specifically: Q(X) = μ_h·H(X) + μ_p·P(X); where Q(X) is the final consistency score; μ_h and μ_p are the overall weight coefficients.

[0164] The process of generating the final crystallinity value is as follows: Constructing a correction function: The basic correction function is C(X, Q) = X [1 + η(Q) Δ(T)]; where C(X, Q) is the correction function, η(Q) is the score-dependent function, and Δ(T) is the temperature correction term; the adaptive weight function is W(Q) = (1 + exp[-k(Q - Q_0)]) -1 Where W(Q) is the weight function, k is the steepness coefficient, and Q_0 is the threshold parameter. Dynamic correction processing is performed to construct a time smoothing function: S(t) = ∫[X(τ)·K(t-τ)]dτ; where S(t) is the smoothing result, K(t) is the kernel function, and τ is the integral variable. The final crystallinity is calculated: X_final = W(Q)·C(X,Q) + [1-W(Q)]·S(t); where X_final is the final crystallinity value, W(Q) is the adaptive weight, C(X,Q) is the correction result, and S(t) is the smoothing result.

[0165] This example optimizes and corrects crystallinity predictions through consistency checking and dynamic correction. First, historical consistency scores and physical constraint scores are calculated to assess the reliability of the predictions. Dynamic correction is then performed based on these scores, improving the stability and accuracy of the predictions. This example not only identifies and corrects anomalous predictions but also enhances the robustness of the predictions through analysis of historical data, ensuring the reliability of the final crystallinity values.

[0166] According to one aspect of the present application, an online device for detecting the crystallinity of a PET film using dielectric properties comprises a dielectric property measurement unit, a temperature measurement unit, a film surface condition monitoring unit, a synchronization control unit, a data acquisition and preprocessing unit, and an auxiliary measurement unit. The dielectric property measurement unit includes a multi-frequency array electrode system consisting of multiple sets of parallel plate electrodes arranged along the direction of motion of the PET film for multi-point parallel measurement; a high-precision impedance analyzer capable of simultaneously measuring the dielectric constant and dielectric loss at multiple frequencies (e.g., 100 Hz-1 MHz); an electrode contact pressure control device comprising a precision pressure sensor and a pressure adjustment mechanism for monitoring and controlling the contact state between the electrode and the PET film; and a signal conditioning circuit for amplifying, filtering, and preprocessing the raw dielectric signal. The temperature measurement unit includes a non-contact infrared temperature sensor array arranged along the width of the film surface for measuring temperature field distribution; contact thermocouples placed at key locations for infrared temperature measurement calibration; and a temperature signal acquisition module for synchronously acquiring multiple temperature signals. The film surface condition monitoring unit includes a high-speed industrial camera for monitoring the surface condition and motion characteristics of the PET film; a laser displacement sensor for measuring film surface flatness and thickness fluctuations; and a tension sensor for monitoring the tension distribution on the film surface. The synchronization control unit includes a high-precision encoder for accurately measuring the film's speed and position; a clock synchronization module for ensuring timing consistency during multi-channel data acquisition; and a trigger control circuit for synchronized sampling of multiple sensors. The data acquisition and preprocessing unit includes a high-speed multi-channel data acquisition card that supports parallel acquisition of multiple signals, including dielectric and temperature; a real-time data buffer module for temporary storage of high-speed raw data; and a fieldbus communication module for data exchange between each measurement unit and the control system. Auxiliary measurement units include an environmental parameter monitoring device for measuring parameters such as ambient temperature and humidity; a production parameter acquisition interface for acquiring process parameters such as line speed and stretch ratio; and a marking system for automatically marking abnormal or characteristic points.

[0167] Dielectric measurement accuracy requirements are: relative error of dielectric constant ≤ 0.1%, loss tangent accuracy ≤ 0.0001; temperature measurement accuracy requirements are: ±0.1°C; sampling frequency requirements are: ≥1kHz for dielectric measurement, ≥100Hz for temperature measurement; spatial resolution requirements are: measurement point spacing ≤50mm across the film width. System integration requirements for modular design are: independent installation and commissioning of each functional unit; protection level requirements are: key measurement units meet IP65 protection rating; interference immunity requirements are: electromagnetic shielding and anti-interference design; calibration requirements are: support for online calibration and compensation. Time synchronization accuracy requirements are: better than 1ms; spatial positioning accuracy requirements are: better than ±1mm; and data time-stamping requirements are: use of GPS or industrial Ethernet clock synchronization.

[0168] This invention addresses the issue of varying contact interface impedance by proposing a contact compensation method. Specifically, a spatial uniformity index for pressure distribution is first established, enabling precise characterization of micro-region contact characteristics. A parallel equivalent circuit comprising contact resistance and capacitance is then constructed, and the temperature influence coefficient is introduced to accurately model the contact impedance. Finally, dynamic compensation based on physical constraints is implemented, effectively eliminating measurement errors caused by changes in contact state through segmented compensation calculations and continuous smoothing. This multi-level compensation strategy improves measurement stability on high-speed production lines. Addressing the issue of multiple effect coupling and separation, a systematic separation of temperature, orientation, and crystal effects is achieved. Specifically, by constructing a temperature-frequency relationship matrix, calculating the temperature sensitivity coefficient and frequency response function, precise separation of the temperature-frequency coupling effect is achieved. Accurate identification of α and β crystal forms is achieved by analyzing dielectric relaxation characteristics. Finally, by establishing a relationship model between orientation and dielectric properties, orientation effects are separated. This systematic decoupling approach effectively addresses the issue of interference from multiple factors. A complete structural characterization system is constructed to address the issue of characterizing the dynamic evolution of structures. Specifically, by analyzing the dielectric spectrum's characteristic frequencies, relaxation time distributions, and order parameter tensors, a dynamic characterization of the α-, β-, and mixed forms is achieved. By establishing a multiscale structural feature space and combining polarization and crystal form characteristics, precise characterization of microstructural evolution is achieved. In particular, by constructing a free energy landscape for crystal form transitions, dynamic prediction of structural evolution is achieved. To address model prediction accuracy, a fusion strategy combining physical models and deep learning is proposed. Specifically, physical constraint equations are used to ensure the physical plausibility of the prediction results. The prediction accuracy of the deep learning model is improved through a multiscale feature pyramid, dependency graph, and a dynamic attention mechanism. Dynamic optimization of confidence weights achieves optimal fusion of the two approaches. This hybrid prediction strategy improves prediction accuracy and reliability. To address the robustness of prediction results, a comprehensive result evaluation and correction mechanism is established. Specifically, the reliability of the prediction results is assessed by calculating historical consistency scores and physical constraint scores. A dynamic correction function is constructed to enable real-time correction of anomalous results. This post-processing strategy improves the robustness of the system under complex operating conditions.

[0169] This invention discloses an online method for detecting the crystallinity of PET films using dielectric properties. The method comprises: acquiring raw dielectric data and real-time temperature data, obtaining spatiotemporal sampling data through parallel acquisition and processing, performing noise identification and removal on the data, and obtaining high-quality dielectric characteristic data using contact compensation; separating the temperature-frequency coupling effect and identifying crystal form characteristics from the high-quality dielectric characteristic data; establishing a multidimensional feature space through polarization mechanism quantification and microstructural feature extraction; predicting crystallinity through the integration of physical models and deep learning methods; and finally obtaining the final crystallinity value through consistency testing and dynamic correction. This invention establishes a comprehensive online method for detecting the crystallinity of PET films through a systematic data processing, feature extraction, and model fusion strategy. It employs a five-layer architecture: "data acquisition - feature decoupling - feature reconstruction - model prediction - result optimization," with each layer providing solutions to specific technical challenges. At the data acquisition layer, parallel acquisition and multiple processing ensure the quality of raw data. At the feature decoupling layer, multiple factors such as temperature, crystal form, and orientation are separated. At the feature reconstruction layer, a multidimensional representation space is established that encompasses polarization and microstructural features. At the model prediction layer, the advantages of physical models and deep learning methods are integrated. At the result optimization layer, consistency testing and dynamic correction ensure the reliability of the final results. This invention not only solves the accuracy problem of online detection of PET film crystallinity but also provides an interpretable structure-performance relationship analysis method, which is of great value in improving product quality control.

[0170] The preferred embodiments of the present invention are described in detail above. However, the present invention is not limited to the specific details in the above embodiments. Within the technical concept of the present invention, various equivalent transformations can be made to the technical solutions of the present invention, and these equivalent transformations all fall within the scope of protection of the present invention.

Claims

1. An online detection method for PET film crystallinity using dielectric properties, characterized in that: The steps include: S1. Obtain raw dielectric data and real-time temperature data, and obtain high-quality dielectric characteristic data through parallel acquisition processing, noise identification and removal, and contact compensation processing; S2. Separate the temperature-frequency coupling effect and identify the crystal form characteristics of the high-quality dielectric characteristic data to obtain a decoupled characteristic vector; S3. Based on the decoupled eigenvector, a multidimensional feature space is established through polarization mechanism quantification processing and microstructure feature extraction to obtain multidimensional feature space data; S4. Based on the multi-dimensional feature space data, the pre-configured physical model and deep learning method are integrated to predict the crystallinity and obtain a preliminary crystallinity prediction value; S5. Based on the preliminary crystallinity prediction value, consistency check and dynamic correction are performed to obtain the final crystallinity value; Step S2 is further as follows: S21. Constructing a temperature-frequency relationship matrix for the high-quality dielectric characteristic data, calculating the temperature sensitivity coefficient and the frequency response function, and performing temperature-frequency coupling effect separation processing on the high-quality dielectric characteristic data accordingly to generate decoupled dielectric data; S22. Based on the decoupled dielectric data, the dielectric relaxation characteristics and mixed state characteristics of the α-crystalline form and the β-crystalline form are calculated and the crystal form distribution is calculated accordingly to generate crystal form characteristic data; S23. Based on the decoupled dielectric data and crystal characteristic data, calculate the molecular chain orientation function and construct a relationship model between orientation and dielectric properties; based on the relationship model, perform orientation effect separation processing on the decoupled dielectric data to generate a decoupled characteristic vector; Step S3 is further as follows: S31. Based on the decoupled eigenvector, construct a polarization component matrix including electronic polarization, ionic polarization, and orientation polarization; calculate the weight and intensity of each polarization component in the polarization component matrix to generate polarization characteristic data; S32. Based on the polarization characteristic data and the crystal form characteristic data, construct structural characteristic parameters, calculate crystal size characteristics, crystal integrity characteristics and interface characteristics, and generate microstructure characteristic data; S33. Based on the microstructure feature data, a high-dimensional feature space is constructed and nonlinear dimensionality reduction processing is performed to generate multi-dimensional feature space data.

2. The method for online detection of PET film crystallinity using dielectric properties according to claim 1, wherein: Step S1 is further as follows: S11, acquiring raw dielectric data and real-time temperature data; and performing parallel acquisition processing on them according to pre-established sampling parameters to generate spatiotemporal sampling data and sampling temperature data; S12. Based on the spatiotemporal sampling data, calculate the wavelet energy distribution characteristics and construct noise characteristic data; Adopting an adaptive threshold method to process noise characteristic data to obtain processed noise characteristic data; Based on the processed noise characteristic data, the spatiotemporal sampling data is subjected to noise reduction processing to obtain noise-reduced data; S13, obtaining contact status parameters in real time; Based on the contact state parameters and sampled temperature data, a contact impedance model is constructed, the compensation coefficient is calculated and combined with the noise reduction data to generate high-quality dielectric characteristic data.

3. The method for online detection of PET film crystallinity using dielectric properties according to claim 2, wherein: Step S4 is further as follows: S41. Based on the multi-dimensional feature space data, construct and solve the physical constraint equation to generate the physical model prediction value; S42, based on multi-dimensional feature space data, data processing is performed through deep neural networks and attention mechanisms to generate deep learning prediction values; S43. Calculate the confidence weights of the physical model prediction value and the deep learning prediction value respectively; based on the confidence weights, perform weighted fusion processing on the physical model prediction value and the deep learning prediction value to generate a preliminary crystallinity prediction value.

4. The method for online detection of crystallinity of a PET film using dielectric properties according to claim 3, wherein: Step S5 is further as follows: S51. Based on the preliminary crystallinity prediction value, calculate the historical consistency score and the physical constraint score and generate consistency score data through weighted processing; S52. Construct a correction function by combining the preliminary crystallinity prediction value and the consistency score data; based on the correction function, dynamically correct the preliminary crystallinity prediction value to generate a final crystallinity value.

5. The method for online detection of PET film crystallinity using dielectric properties according to claim 4, wherein: Step S11 is further as follows: S111, acquiring original dielectric data and real-time temperature data, and generating sampling interval parameters; S112, determining a basic grid based on the sampling interval parameter and generating sampling position data; S113, performing multi-channel data acquisition based on the sampling position data to generate multi-channel acquisition data; S114, calculating the time deviation based on the multi-channel acquisition data and generating aligned acquisition data using a cubic spline function; S115, calculating local statistical features of the aligned collected data, identifying outliers, and constructing a correlation matrix; The correlation matrix is ​​correspondingly integrated with the real-time temperature data to generate spatiotemporal sampling data and sampling temperature data.

6. The method for online detection of PET film crystallinity using dielectric properties according to claim 4, wherein: Step S12 is further as follows: S121, grouping the spatiotemporal sampling data according to frequency, performing multi-scale decomposition, calculating the energy value at each scale, and generating energy distribution features; S122. Based on the energy distribution characteristics, identify the abnormal fluctuation interval, extract the time-frequency characteristic parameters, and generate noise characteristic data; S123, performing cluster analysis on the noise characteristic data, calculating energy concentration, and generating a noise determination result; S124, performing time-frequency decomposition on the spatiotemporal sampling data based on the noise determination result, calculating the local correlation coefficient, constructing the signal reconstruction weight, and generating preliminary noise reduction data; S125. Based on the preliminary noise reduction data, calculate the local smoothness index of the signal, identify the residual noise points and perform local correction to obtain a correction result; Integrate preliminary noise reduction data and correction results to generate noise reduction data; S126 , time-aligning the sampled temperature data with the noise reduction data, and outputting the aligned noise reduction data and sampled temperature data.

7. The method for online detection of PET film crystallinity using dielectric properties according to claim 4, wherein: Step S22 is further as follows: S221. Calculate the dielectric loss peak shape parameters based on the decoupled dielectric data, construct a time-temperature equivalent curve of the loss peak, and generate relaxation characteristic data; S222. Calculating the orientation correlation function of the α crystal form based on the relaxation characteristic data to generate α crystal form characteristic data; S223. Based on the relaxation characteristic data, construct local order parameters and generate β crystal characteristic data; S224, calculating the overlap of the characteristic data of the α and β crystal forms, identifying the mixed state region, constructing the order parameter tensor and calculating the stability index, and generating the mixed state characteristic data; S225. Construct a multidimensional feature space for the characteristic data of the α-crystalline form, the β-crystalline form, and the mixed state, calculate the distribution density function, identify the characteristic clusters, calculate the transition probability, and generate preliminary distribution data; S226. Based on the preliminary distribution data, calculate the crystal form correlation, identify and optimize the local distribution abnormality area, and generate crystal form characteristic data.

8. The method for online detection of PET film crystallinity using dielectric properties according to claim 4, wherein: Step S32 is further as follows: S321. Based on the polarization characteristic data and the crystal form characteristic data, extract the polarization time constant sequence and calculate the polarization relaxation time distribution to generate structural parameter data; S322, calculating the characteristic length distribution at each scale to obtain dimension characteristic data; S323. Based on the size characteristic data and the crystal form characteristic data, the order parameter of the local structure is calculated, the defect position and type are extracted, and the integrity characteristic data is generated; S324, calculating the interface energy, extracting the interface curvature and roughness, and obtaining interface characteristic data; S325, calculating the interface stress distribution, extracting the stress concentration area, and generating interface stress data; S326. Based on the size characteristic data, integrity characteristic data, interface characteristic data and interface stress data, a characteristic parameter correlation matrix is ​​constructed, the coupling strength is calculated, and the microstructure characteristic data is generated.

Images