Automatic roll changing method and system for tpu film

Abstract

The application provides a TPU film automatic roll changing method and system, comprising: according to the charge capture feature, using a support vector machine to classify and map the charge capture feature, and determining an electrostatic field strength threshold corresponding to the charge capture feature; according to the field strength increase instruction, adjusting the output voltage of an ion wind rod, and obtaining the adsorption force value between the film end and the surface of the roll core after adjusting the output voltage; processing the adsorption force value through a random forest algorithm, constructing the mapping relationship between the adsorption force value and the slipping risk, and obtaining the end slipping probability; if the end slipping probability is greater than a preset probability threshold, collecting the surface image of the film end, performing grayscale and edge detection processing on the surface image, and obtaining the end wrinkle feature.

Description

Technical Field

[0001] This invention relates to the field of TPU film processing technology, and in particular to an automatic TPU film roll changing method and system. Background Technology

[0002] Thermoplastic polyurethane film plays a crucial role in modern industrial manufacturing, and the roll-changing process in its continuous production directly affects the operating efficiency of the production line and the product yield. In adhesive-free tape roll-changing processes, traditional methods often rely on external mechanical pressure or simply increasing the voltage of the electrostatic generator to forcibly complete the end bonding. This approach ignores the limitations of the film material's inherent physicochemical properties, which not only easily causes surface stretching deformation of the film but also leads to winding misalignment at high speeds due to the mismatch between external force intervention and the film's motion state. This makes it difficult to meet the stringent requirements of modern production for high precision and high stability.

[0003] A deeper investigation into the failure phenomenon of adhesive-free tape rewinding reveals that the core obstacle lies in the inherent repulsion and loss characteristics of the film's internal chemical composition against electrostatic charges. Conventional thermoplastic polyurethane material formulations lack effective charge trapping mechanisms, resulting in extremely limited capacity to accept and store electrostatic charges. This inherent charge loss characteristic of the material directly leads to slow response during the rewinding process. Specifically, when the ionizer on the production line attempts to inject charge into the film tip, and a high-voltage electrostatic field has been applied to the surface of the new roll core to prepare for the film, the film tip cannot effectively lock in these injected charges, significantly weakening the attraction between opposite charges. Just as a demagnetized iron sheet cannot instantly generate a strong attraction to a magnet, the film tip, when approaching the high-speed rotating new roll core, cannot generate a sufficiently strong adsorption force in a very short time. Ultimately, this results in the tip not adhering tightly to the roll core surface, even slipping or severely wrinkling before the pressure roller intervenes.

[0004] How to overcome the bottleneck of low charge storage capacity at the film end by starting with the formulation of the material itself, so that it can quickly gather opposite charges and generate strong adsorption force when facing a high voltage electrostatic field, thus providing an extremely stable physical basis for the initial ring fixation of the subsequent pressure roller, has become the key issue for realizing an efficient adhesive-free tape rewinding process. Summary of the Invention

[0005] This invention provides an automatic TPU film roll changing method, mainly including: The film formulation data and charge loss data of thermoplastic polyurethane film are obtained, and the film formulation data and charge loss data are dimensionality reduced by principal component analysis algorithm to obtain charge trapping characteristics. Based on the charge trapping features, a support vector machine is used to classify and map the charge trapping features to determine the electrostatic field strength threshold corresponding to the charge trapping features. Acquire the opposite charge distribution data when the film end is close to the core surface, extract the current field strength value from the opposite charge distribution data, and if the current field strength value is less than the electrostatic field strength threshold, determine the field strength increase command. Adjust the output voltage of the ion wind bar according to the field strength increase command, and obtain the adsorption force value between the film end and the core surface after adjusting the output voltage; The adsorption force values ​​are processed using a random forest algorithm to construct a mapping relationship between the adsorption force values ​​and the slippage risk, thereby obtaining the end slippage probability. If the probability of the end slippage is greater than a preset probability threshold, then the surface image of the film end is acquired, and the surface image is processed by grayscale conversion and edge detection to obtain the end wrinkling features. Based on the wrinkling characteristics at the film end, the degree of deformation at the film end is calculated, an initial fixing and adjustment command is determined based on the degree of deformation, and the timing of the intervention of the pressure roller is controlled based on the initial fixing and adjustment command to complete the adhesive-free tape changeover process.

[0006] This invention provides an automatic TPU film roll changing system, mainly comprising: The charge trapping feature extraction module is used to acquire film formulation data and charge loss data of thermoplastic polyurethane film, and to perform dimensionality reduction processing on the film formulation data and charge loss data through principal component analysis algorithm to obtain charge trapping features; The electrostatic field strength threshold determination module is used to determine the electrostatic field strength threshold corresponding to the charge trapping features by performing classification and mapping processing on the charge trapping features using a support vector machine. The field strength increase instruction determination module is used to acquire the opposite charge distribution data when the film end is close to the core surface, extract the current field strength value from the opposite charge distribution data, and determine the field strength increase instruction if the current field strength value is less than the electrostatic field strength threshold. The adsorption force acquisition module is used to adjust the output voltage of the ion wind bar according to the field strength increase command, and to obtain the adsorption force value between the end of the film and the surface of the core after adjusting the output voltage. The end slip probability calculation module is used to process the adsorption force value through the random forest algorithm, construct the mapping relationship between the adsorption force value and the slip risk, and obtain the end slip probability; The end wrinkling feature acquisition module is used to acquire a surface image of the film end if the end slippage probability is greater than a preset probability threshold, and to perform grayscale and edge detection processing on the surface image to obtain end wrinkling features. The adhesive-free tape rewinding module is used to calculate the degree of deformation of the film end based on the wrinkling characteristics of the end, determine the initial fixing adjustment command based on the degree of deformation, and control the intervention timing of the pressure roller based on the initial fixing adjustment command to complete the adhesive-free tape rewinding process.

[0007] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects: This invention discloses an automatic TPU film rewinding method. Addressing the issue of slippage and wrinkling of thermoplastic polyurethane films during adhesive-free rewinding due to uneven charge distribution and insufficient adsorption force at the film ends, the method acquires film formulation data and charge loss data. Principal component analysis is used for dimensionality reduction to obtain charge capture features, which are then classified and mapped using a support vector machine to accurately determine the electrostatic field strength threshold. When the film end approaches the core, the current field strength value in the opposite charge distribution is extracted and compared with the threshold to generate a field strength increase command, dynamically adjusting the output voltage of the ion fan to strengthen the adsorption force between the end and the core surface. Furthermore, a random forest algorithm is used to construct a mapping relationship between adsorption force and slippage risk, obtaining the end slippage probability. When the probability exceeds a preset threshold, an image of the end surface is acquired, grayscale is performed, and edge detection is conducted. Wrinkling features are extracted to calculate the degree of deformation, generating an initial fixed adjustment command to precisely control the timing of pressure roller intervention. This invention achieves synergistic closed-loop control of electrostatic adsorption and mechanical pressing, effectively improving the stability, bonding accuracy, and production yield of adhesive-free rewinding. Attached Figure Description

[0008] Figure 1 This is a flowchart of an automatic TPU film roll changing method according to the present invention.

[0009] Figure 2 This is a schematic diagram of an automatic TPU film changing method according to the present invention.

[0010] Figure 3 This is another schematic diagram of an automatic TPU film changing method according to the present invention.

[0011] Figure 4 This is a schematic diagram of the structure of an automatic TPU film changing system according to the present invention. Detailed Implementation

[0012] To further understand the content of this invention, a detailed description of the invention is provided in conjunction with the accompanying drawings and embodiments. The specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention. It should also be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.

[0013] like Figures 1-4 This embodiment of an automatic TPU film roll changing method may specifically include: S101: Obtain film formulation data and charge loss data of thermoplastic polyurethane film, and perform dimensionality reduction processing on the film formulation data and charge loss data using principal component analysis algorithm to obtain charge trapping characteristics.

[0014] Obtain the first formulation value and first loss amount of the thermoplastic polyurethane film, arrange the samples by row and the variables by column to construct an initial data matrix X; calculate the mean value vector u by column of X, and center the matrix Y by setting Y=Xu; calculate the covariance between each pair of variables based on Y to obtain the covariance matrix C; use principal component analysis to decompose C into eigenvalues ​​to obtain the eigenvalues ​​λ1, λ2, ..., λn arranged in descending order and the corresponding eigenvectors v1, v2, ..., vn; and accumulate the contribution rate rk=(λ The formula 1+λ2+...+λk) / (λ1+λ2+...+λn) is used. The process stops when rk is first greater than or equal to the preset contribution rate threshold of 0.85. The first k eigenvectors are then concatenated column-wise to form the principal component load matrix W. For example, when λ1=4.2, λ2=1.6, λ3=0.7, and the sum of the remaining eigenvalues ​​is 0.5, r2 is 0.85, i.e., k=2. The dimension-reduced projection value Z is obtained through Z=Y×W. The first column of Z is used as the charge trapping feature, denoted as the target charge trapping amount, for subsequent thin film electrical performance evaluation and formulation optimization mapping.

[0015] S102: Based on the charge trapping features, a support vector machine is used to classify and map the charge trapping features to determine the electrostatic field strength threshold corresponding to the charge trapping features.

[0016] The charge trapping feature is composed of a multi-dimensional feature vector formed by normalization after collaborative acquisition by the surface potential probe, the induced current acquisition module, and the temperature and humidity sensor in step S101. Specifically, it includes six components: surface potential decay time constant, trapped charge density per unit area, peak value of the retreat current, half width at half maximum (FWHM) of the retreat current, ambient temperature, and relative humidity. The sampling period is 100 milliseconds, and the average of 10 consecutive sampling points is taken as the input for each sample. The Support Vector Machine (SVM) employs a radial basis function (RBF) kernel and performs one-to-one multi-class classification training using pre-labeled charge capture feature samples. The electrostatic field strength is discretized into five levels according to hazard level, corresponding to threshold center values ​​of 2 kV / mm, 4 kV / mm, 6 kV / mm, 8 kV / mm, and 10 kV / mm, respectively. Each level serves as a class label in training. During classification mapping, the SVM first outputs decision function values ​​f1 to f5 for assigning the test sample to each level. The level corresponding to the largest decision value is selected as the primary level, and the second largest decision value is chosen. The gear corresponding to the value is taken as the neighboring gear; then, using the center thresholds Emain and Eneighbor of the main gear and the neighboring gear, and their normalized decision values ​​wmain and wneighbor, a continuous electrostatic field strength threshold E is obtained by E = wmain × Emain + wneighbor × Eneighbor, where wmain + wneighbor = 1, and wmain and wneighbor are obtained by normalizing the corresponding decision function values ​​using softmax; through the above gear classification plus neighboring gear weighted interpolation method, the discrete category output of the support vector machine is converted into a continuous electrostatic field strength threshold with a resolution better than 0.5 kV per millimeter for subsequent steps.

[0017] S103: Obtain the opposite charge distribution data when the film end is close to the core surface, extract the current field strength value from the opposite charge distribution data, and if the current field strength value is less than the electrostatic field strength threshold, determine the field strength increase command.

[0018] Data on the surface density distribution of opposite charges near the core surface at the film end is collected. A 5mm x 5mm grid is divided along the axial and circumferential directions of the bonding area on the core surface. Based on the distance between the film and the core, the electric field strength value is calculated from the surface density at each grid point according to Coulomb's law. The maximum electric field strength at all grid points within the bonding area is taken as the current field strength value Ecurrent to avoid local weak field areas being masked by the average value. Ecurrent is compared with the electrostatic field strength threshold Ethreshold. If Ecurrent is less than Ethreshold, a field strength increase command is generated. This command includes the target field strength Etarget and the field strength increment ΔEthreshold. The system includes three parameters: target voltage (E_target) and current voltage (E_current). Target voltage is 1.05 to 1.1 times the threshold voltage (E_threshold). ΔE = target voltage (E_current). ΔU is obtained by reverse lookup from ΔE based on the calibration curve of the high-voltage electrode voltage and the electric field strength on the core surface. The calibration curve is obtained by measuring the electric field strength point by point using a field strength probe before the equipment leaves the factory. For example, when the threshold voltage (E_threshold) is 3.0 x 10^6 volts per meter and the current voltage (E_current) is 2.4 x 10^6 volts per meter, the target voltage (E_target) is taken as 3.15 x 10^6 volts per meter, and ΔE is 0.75 x 10^6 volts per meter. The specific value of ΔU is then determined by the calibration curve and written into the instruction.

[0019] S104: Adjust the output voltage of the ion bar according to the field strength increase command, and obtain the adsorption force value between the end of the film and the surface of the core after adjusting the output voltage.

[0020] The output voltage of the ion bar is adjusted according to the field strength increase command. The adjustment is performed according to the linear relationship U=U0+k×ΔE, where U is the adjusted output voltage, U0 is the current output voltage, ΔE is the difference between the target field strength and the current field strength, and k is the voltage field strength conversion coefficient, with a value range of 50 to 80 volts per kilovolt per meter. The single voltage adjustment step size does not exceed 200 volts, and the adjustment range is limited to 3 kilovolts to 15 kilovolts. After each adjustment, the stability time is not less than 500 milliseconds before entering the next round of adjustment until the field strength reaches the set target value. After the voltage adjustment is completed, the normal adhesion force of the contact area between the film end and the surface of the core is synchronously collected by a film tension sensor pre-arranged under the core and a micro-force measurement module attached to the core shaft end. The sampling frequency is not less than 100 Hz, and the continuous acquisition time is 1 second. The collected raw data is subjected to mean filtering to obtain the adsorption force value in millinewtons, and this value is uploaded to the control unit for subsequent judgment and feedback.

[0021] S105: The adsorption force value is processed by the random forest algorithm to construct a mapping relationship between the adsorption force value and the slippage risk, and the end slippage probability is obtained.

[0022] The adsorption force value is fed into a pre-trained random forest model, which outputs the end slip probability, which is a quantitative representation of the slip risk. The random forest model is trained based on historical inspection data before use. The training samples are adsorption force records and corresponding slip events collected from field operations over the past three years. Positive samples are records of working conditions where end slip has occurred, and negative samples are records of normal fit conditions. The ratio of positive to negative samples is controlled at approximately 1:3 to avoid class imbalance. Each sample extracts features including the mean adsorption force, variance of adsorption force, minimum adsorption force, slope of force decrease, and duration of adsorption force below the safety threshold of 80 N, totaling five dimensions, forming a feature vector x = [x1, x2, x3, x4, x5]. The model consists of 10... The system consists of 0 CART decision trees, with a maximum depth of 8 layers per tree. Node splits are performed using the Gini coefficient criterion, and two features are randomly selected for comparison in each split. During training, the sample set is sampled with replacement to enhance generalization ability. During online inference, the currently collected adsorption force values ​​are extracted according to the above 5 dimensions and input into the model. Each decision tree independently gives a slip or normal judgment. The final end slip probability P is calculated according to the voting ratio, i.e., P=N1 / N, where N1 is the number of trees judged as slipping and N is the total number of decision trees (100). When P is not less than 0.6, it is considered high slip risk; between 0.3 and 0.6, it is considered medium slip risk; and below 0.3, it is considered low slip risk. This establishes a mapping relationship between the adsorption force value and the end slip probability.

[0023] S106: If the end slippage probability is greater than a preset probability threshold, then the surface image of the film end is acquired, and the surface image is processed by grayscale conversion and edge detection to obtain the end wrinkling feature.

[0024] The end slippage probability is compared with a preset probability threshold, which is 0.6. This threshold is determined by the inflection point of the statistical distribution of slippage events and normal bonding samples in previous process tests. The value can be finely adjusted between 0.55 and 0.7 according to the winding speed and film thickness to maintain the same dimension as the normalized slippage probability output by the previous module. If the end slippage probability is greater than the preset probability threshold, the linear scan camera is triggered to acquire a surface image of the film end. First, the color image is converted into a grayscale image I using a weighting method of 0.299×R+0.587×G+0.114×B. Then, Gaussian filtering is applied to I for smoothing, and the Canny operator is used to complete edge detection. The measurement yields a binary edge map E, with high and low thresholds set to 0.2 and 0.08, respectively. Based on edge map E, end-wrinkle features are extracted, specifically including: dividing the end region into several 5mm wide sliding windows along the winding direction; performing Hough line detection on the edge pixels within each window to obtain the main wrinkle direction angle θ; calculating the ratio of the number of edge pixels within the window to the window area to obtain the wrinkle density ρ; merging adjacent edge segments with an angle less than 10 degrees into connected components, taking the pixel length of the longest connected component and converting it to the actual wrinkle length L according to the camera calibration coefficient; finally, the four-dimensional vector composed of θ, ρ, L, and the number of wrinkles N is used as the end-wrinkle feature and output to the subsequent judgment module.

[0025] S107: Based on the wrinkling characteristics at the end, calculate the degree of deformation at the film end, determine the initial fixing and adjustment command based on the degree of deformation, control the intervention timing of the pressure roller based on the initial fixing and adjustment command, and complete the adhesive-free tape rewinding process.

[0026] The calculation process for the degree of deformation is as follows: extract the average amplitude h of the ripples, the number of ripples per unit length n, and the maximum lateral offset d of the end relative to the baseline from the wrinkling characteristics at the end. These are then weighted and summed using D = 0.5 × h + 0.3 × n + 0.2 × d to obtain the degree of deformation D. Before calculation, h, n, and d are all normalized to the range of 0 to 1 according to their respective ranges. The value of the degree of deformation D falls between 0 and 1; 0 to 0.3 is considered slight deformation, 0.3 to 0.6 is considered moderate deformation, and values ​​above 0.6 are considered severe deformation. For example, if a roll change yields h of 0.4, n of 0.5, and d of 0.3, then D = 0.5 × 0.4 + 0.3 × 0.5 + 0.2 × 0.3 = 0.41, which is determined to be moderate deformation.

[0027] The initial fixed adjustment command is a set of structured control parameters, including three fields: pressure F of the pressure roller, displacement S of the pressure roller, and tension threshold T. These parameters are generated by the degree of deformation D in a graded manner: slight deformation corresponds to F of 80 N, S of 5 mm, and T of 12 N / m; moderate deformation corresponds to F of 120 N, S of 8 mm, and T of 15 N / m; and severe deformation corresponds to F of 160 N, S of 12 mm, and T of 18 N / m.

[0028] The intervention timing is determined by S and T in the initial fixed adjustment command. When the end of the new roll reaches the displacement point defined by S before the overlap reference position, and the measured value of the winding and unwinding tension enters the allowable range of ±1 N / m for T, the pressure roller pressing actuator will press F to complete the bonding and pressing of the film end. After the pressure roller is in place, the cutter is triggered to perform a transverse cut on the tail section of the old roll. The cut point is located 20 mm behind the pressure roller indentation. Then, the winding motor presses T to maintain the tension for 2 to 3 seconds, so that the end of the new roll and the tail of the old roll complete the glueless overlap under the pressure of the pressure roller. Finally, the old roll path is disconnected and switched to the new roll path to complete the glueless tape changing process.

[0029] This invention provides an automatic TPU film roll changing system, mainly comprising: The charge trapping feature extraction module is used to acquire film formulation data and charge loss data of thermoplastic polyurethane film, and to perform dimensionality reduction processing on the film formulation data and charge loss data through principal component analysis algorithm to obtain charge trapping features; The electrostatic field strength threshold determination module is used to determine the electrostatic field strength threshold corresponding to the charge trapping features by performing classification and mapping processing on the charge trapping features using a support vector machine. The field strength increase instruction determination module is used to acquire the opposite charge distribution data when the film end is close to the core surface, extract the current field strength value from the opposite charge distribution data, and determine the field strength increase instruction if the current field strength value is less than the electrostatic field strength threshold. The adsorption force acquisition module is used to adjust the output voltage of the ion wind bar according to the field strength increase command, and to obtain the adsorption force value between the end of the film and the surface of the core after adjusting the output voltage. The end slip probability calculation module is used to process the adsorption force value through the random forest algorithm, construct the mapping relationship between the adsorption force value and the slip risk, and obtain the end slip probability; The end wrinkling feature acquisition module is used to acquire a surface image of the film end if the end slippage probability is greater than a preset probability threshold, and to perform grayscale and edge detection processing on the surface image to obtain end wrinkling features. The adhesive-free tape rewinding module is used to calculate the degree of deformation of the film end based on the wrinkling characteristics of the end, determine the initial fixing adjustment command based on the degree of deformation, and control the intervention timing of the pressure roller based on the initial fixing adjustment command to complete the adhesive-free tape rewinding process.

[0030] The above description is merely a preferred embodiment of this application and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in this application is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the concept of this application. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this application.

Claims

1. An automatic TPU film roll changing method, characterized in that, The method includes: S101: Obtain film formulation data and charge loss data of thermoplastic polyurethane film, and perform dimensionality reduction processing on the film formulation data and charge loss data through principal component analysis algorithm to obtain charge trapping characteristics; S102: Based on the charge trapping features, a support vector machine is used to classify and map the charge trapping features to determine the electrostatic field strength threshold corresponding to the charge trapping features; S103: Obtain the opposite charge distribution data when the film end is close to the core surface, extract the current field strength value from the opposite charge distribution data, and if the current field strength value is less than the electrostatic field strength threshold, determine the field strength increase command. S104: Adjust the output voltage of the ion bar according to the field strength increase command, and obtain the adsorption force value between the end of the film and the surface of the core after adjusting the output voltage. S105: The adsorption force value is processed by the random forest algorithm to construct a mapping relationship between the adsorption force value and the slippage risk, and the end slippage probability is obtained; S106: If the end slippage probability is greater than a preset probability threshold, then the surface image of the film end is acquired, and the surface image is processed by grayscale conversion and edge detection to obtain the end wrinkling features. S107: Based on the wrinkling characteristics at the end, calculate the degree of deformation at the film end, determine the initial fixing and adjustment command based on the degree of deformation, control the intervention timing of the pressure roller based on the initial fixing and adjustment command, and complete the adhesive-free tape rewinding process.

2. The automatic TPU film roll changing method according to claim 1, characterized in that, The process involves acquiring film formulation data and charge loss data for thermoplastic polyurethane films, and then performing dimensionality reduction on the film formulation data and charge loss data using principal component analysis to obtain charge trapping features, including: The first formulation value and first loss amount of the thermoplastic polyurethane film are obtained. Samples are arranged by row and variables by column to construct an initial data matrix X. The mean vector u is obtained by calculating the mean of X column by column. Y = Xu is then used to center the data, resulting in a centered matrix Y. Based on Y, the covariance between each pair of variables is calculated to obtain the covariance matrix C. Principal component analysis is used to decompose C into eigenvalues, yielding eigenvalues ​​arranged in descending order. and the corresponding feature vectors ;Accumulate contribution rate one by one ,when The process stops when the first contribution rate is greater than or equal to the preset contribution rate threshold of 0.

85. The first k feature vectors are concatenated column by column to form the principal component load matrix W. The dimension-reduced projection value Z is obtained by Z=Y×W. The first column of Z is taken as the charge trapping feature and recorded as the target charge trapping amount, which is used for subsequent thin film electrical performance evaluation and formulation optimization mapping.

3. The automatic TPU film roll changing method according to claim 1, characterized in that, The step of using a support vector machine to classify and map the charge trapping features based on the charge trapping features, and determining the electrostatic field strength threshold corresponding to the charge trapping features, includes: The charge trapping feature is composed of a multi-dimensional feature vector formed by normalization after collaborative acquisition by the surface potential probe, the induced current acquisition module, and the temperature and humidity sensor in step S101. Specifically, it includes six components: surface potential decay time constant, trapped charge density per unit area, peak value of the retreat current, half width at half maximum (FWHM) of the retreat current, ambient temperature, and relative humidity. The sampling period is 100 milliseconds, and the average of 10 consecutive sampling points is taken as the input for each sample. The Support Vector Machine (SVM) employs a radial basis function (RBF) kernel and performs one-to-one multi-class classification training using pre-labeled charge capture feature samples. The electrostatic field strength is discretized into five levels according to hazard level, corresponding to threshold center values ​​of 2 kV / mm, 4 kV / mm, 6 kV / mm, 8 kV / mm, and 10 kV / mm, respectively. Each level serves as a class label in training. During classification mapping, the SVM first outputs decision function values ​​f1 to f5 for assigning the test sample to each level. The level corresponding to the largest decision value is selected as the primary level, and the level corresponding to the second largest decision value is selected as the neighboring level. Subsequently, the center threshold E between the primary and neighboring levels is used... 主 E 邻 and its normalized decision value w 主 w 邻 According to E=w 主 ×E 主 +w 邻 ×E 邻 The continuous electrostatic field strength threshold E is obtained, where w 主 +w 邻 =1, w 主 w 邻 The value is obtained by normalizing the corresponding decision function value using softmax. By using the above-mentioned gear classification plus neighbor gear weighted interpolation method, the discrete class output of the support vector machine is converted into a continuous electrostatic field strength threshold with a resolution better than 0.5 kV per millimeter for subsequent steps.

4. The automatic TPU film changing method according to claim 1, characterized in that, The process involves acquiring data on the distribution of opposite charges when the film end is close to the core surface, extracting the current field strength value from the opposite charge distribution data, and determining a field strength increase instruction if the current field strength value is less than the electrostatic field strength threshold. This includes: Data on the surface density distribution of opposite charges at the end of the film near the surface of the core were collected. A 5mm x 5mm grid was then divided along the axial and circumferential directions of the bonding area on the core surface. Based on the distance between the film and the core, the electric field strength value was calculated from the surface density at each grid point using Coulomb's law. The maximum electric field strength value across all grid points within the bonding area was taken as the current electric field strength value E. 当前 To avoid local weak field regions being masked by the average value; E 当前 With electrostatic field strength threshold E 阈 Compare, if E 当前 Less than E 阈 Then, a field strength increase command is generated, which includes the target field strength E. 目标 The three parameters are: field strength increment ΔE, and the corresponding high-voltage electrode voltage adjustment ΔU, where E 目标 Take E 阈 1.05 to 1.1 times, ΔE = E 目标 -E 当前 ΔU is obtained by reverse lookup from ΔE based on the calibration curve of high voltage electrode voltage and magnetic field strength on the core surface. The calibration curve is obtained by measuring point by point with the magnetic field probe before the equipment leaves the factory; then the specific value of ΔU is determined by the calibration curve and written into the instruction.

5. The automatic TPU film roll changing method according to claim 1, characterized in that, The step of adjusting the output voltage of the ion bar according to the field strength increase command, and obtaining the adsorption force value between the film end and the core surface after adjusting the output voltage, includes: The output voltage of the ion bar is adjusted according to the field strength increase command. The adjustment is performed according to the linear relationship U=U0+k×ΔE, where U is the adjusted output voltage, U0 is the current output voltage, ΔE is the difference between the target field strength and the current field strength, and k is the voltage field strength conversion coefficient, with a value range of 50 to 80 volts per kilovolt per meter. The single voltage adjustment step size does not exceed 200 volts, and the adjustment range is limited to 3 kilovolts to 15 kilovolts. After each adjustment, the stability time is not less than 500 milliseconds before entering the next round of adjustment until the field strength reaches the set target value. After the voltage adjustment is completed, the normal adhesion force of the contact area between the film end and the surface of the core is synchronously collected by a film tension sensor pre-arranged under the core and a micro-force measurement module attached to the core shaft end. The sampling frequency is not less than 100 Hz, and the continuous acquisition time is 1 second. The collected raw data is subjected to mean filtering to obtain the adsorption force value in millinewtons, and this value is uploaded to the control unit for subsequent judgment and feedback.

6. The automatic TPU film roll changing method according to claim 1, characterized in that, The process of processing the adsorption force value using a random forest algorithm to construct a mapping relationship between the adsorption force value and the slippage risk, and obtaining the end slippage probability, includes: The adsorption force value is fed into a pre-trained random forest model, which outputs the end slip probability, which is a quantitative representation of the slip risk. The random forest model is trained based on historical inspection data before use. The training samples are adsorption force records and corresponding slip events collected from field operations over the past three years. Positive samples are records of working conditions where end slip has occurred, and negative samples are records of normal bonding conditions. The ratio of positive to negative samples is controlled at 1:3 to avoid class imbalance. Each sample extracts features including the mean adsorption force, variance of adsorption force, minimum adsorption force, slope of force decrease, and duration of adsorption force below the safety threshold of 80 N, forming a feature vector x = [x1, x2, x3, x4, x5]. The model consists of 100... The system consists of CART decision trees, with a maximum depth of 8 layers per tree. Node splits are performed using the Gini coefficient criterion, and two features are randomly selected for comparison during each split. During training, the sample set is sampled with replacement to enhance generalization ability. During online inference, the currently collected adsorption force values ​​are extracted according to the above five dimensions and input into the model. Each decision tree independently determines whether the slippage is normal. The final end slippage probability P is calculated according to the voting ratio, i.e., P = N1 / N, where N1 is the number of trees that determine slippage, and N is the total number of decision trees (100). When P is not less than 0.6, it is considered a high slippage risk; between 0.3 and 0.6, it is considered a medium slippage risk; and below 0.3, it is considered a low slippage risk. This establishes a mapping relationship between the adsorption force value and the end slippage probability.

7. The automatic TPU film roll changing method according to claim 1, characterized in that, If the probability of the end slippage is greater than a preset probability threshold, then a surface image of the film end is acquired, and the surface image is processed by grayscale conversion and edge detection to obtain end wrinkling features, including: The end slippage probability is compared with a preset probability threshold, which is 0.

6. This threshold is determined by the inflection point of the statistical distribution of slippage events and normal bonding samples in previous process tests, and its value ranges from 0.55 to 0.

7. It is finely adjusted according to the winding speed and film thickness to maintain the same dimension as the normalized slippage probability output by the previous module. If the end slippage probability is greater than the preset probability threshold, the linear scan camera is triggered to acquire a surface image of the film end. First, the color image is converted into a grayscale image I using a weighting method of 0.299×R+0.587×G+0.114×B. Then, Gaussian filtering is applied to I for smoothing, and edge detection is performed using the Canny operator. A binary edge map E is obtained, with high and low thresholds set to 0.2 and 0.08, respectively. Based on the edge map E, end wrinkling features are extracted, specifically including: dividing the end region into several sliding windows with a width of 5 mm along the winding direction, performing Hough line detection on the edge pixels in each window to obtain the main direction angle of the wrinkle θ; calculating the ratio of the number of edge pixels in the window to the window area to obtain the wrinkle density ρ; merging adjacent edge segments with a direction angle less than 10 degrees into connected components, taking the pixel length of the longest connected segment and converting it into the actual wrinkle length L according to the camera calibration coefficient; finally, the four-dimensional vector composed of θ, ρ, L and the number of wrinkles N is used as the end wrinkling feature and output to the subsequent judgment module.

8. The automatic TPU film roll changing method according to claim 1, characterized in that, The process of calculating the degree of deformation of the film end based on the wrinkling characteristics at the end, determining the initial fixing and adjustment command based on the degree of deformation, and controlling the intervention timing of the pressure roller according to the initial fixing and adjustment command to complete the adhesive-free tape rewinding process includes: The calculation process for the degree of deformation is as follows: The average amplitude h of the ripples, the number of ripples per unit length n, and the maximum lateral offset d of the end relative to the baseline are extracted from the wrinkling characteristics at the end. These are then weighted and summed using the formula D = 0.5 × h + 0.3 × n + 0.2 × d to obtain the degree of deformation D. Before calculation, h, n, and d are all normalized to the range of 0 to 1 according to their respective ranges. The value of the degree of deformation D falls between 0 and 1; 0 to 0.3 is considered slight deformation, 0.3 to 0.6 is considered moderate deformation, and values ​​above 0.6 are considered severe deformation. The initial fixed adjustment command is a set of structured control parameters, including three fields: pressure F of the pressure roller, displacement S of the pressure roller, and tension threshold T. These parameters are generated by classifying the degree of deformation D: slight deformation corresponds to F = 80 N, S = 5 mm, and T = 12 N / m; moderate deformation corresponds to F = 120 N, S = 8 mm, and T = 15 N / m; severe deformation corresponds to F = 160 N, S = 12 mm, and T = 18 N / m. The intervention timing is determined by S and T in the initial fixed adjustment command. When the end of the new roll reaches the displacement point defined by S before the overlap reference position, and the measured value of the winding and unwinding tension enters the allowable range of ±1 N / m for T, the pressure roller pressing actuator will press F to complete the bonding and pressing of the film end. After the pressure roller is in place, the cutter is triggered to perform a transverse cut on the tail section of the old roll. The cut point is located 20 mm behind the pressure roller indentation. Then, the winding motor presses T to maintain the tension for 2 to 3 seconds, so that the end of the new roll and the tail of the old roll complete the glueless overlap under the pressure of the pressure roller. Finally, the old roll path is disconnected and switched to the new roll path to complete the glueless tape changing process.

9. An automatic TPU film roll changing system, characterized in that, The system includes: The charge trapping feature extraction module is used to acquire film formulation data and charge loss data of thermoplastic polyurethane film, and to perform dimensionality reduction processing on the film formulation data and charge loss data through principal component analysis algorithm to obtain charge trapping features; The electrostatic field strength threshold determination module is used to determine the electrostatic field strength threshold corresponding to the charge trapping features by performing classification and mapping processing on the charge trapping features using a support vector machine. The field strength increase instruction determination module is used to acquire the opposite charge distribution data when the film end is close to the core surface, extract the current field strength value from the opposite charge distribution data, and determine the field strength increase instruction if the current field strength value is less than the electrostatic field strength threshold. The adsorption force acquisition module is used to adjust the output voltage of the ion wind bar according to the field strength increase command, and obtain the adsorption force value between the end of the film and the surface of the core after adjusting the output voltage. The end slip probability calculation module is used to process the adsorption force value through the random forest algorithm, construct the mapping relationship between the adsorption force value and the slip risk, and obtain the end slip probability; The end wrinkling feature acquisition module is used to acquire a surface image of the film end if the end slippage probability is greater than a preset probability threshold, and to perform grayscale and edge detection processing on the surface image to obtain end wrinkling features. The adhesive-free tape rewinding module is used to calculate the degree of deformation of the film end based on the wrinkling characteristics of the end, determine the initial fixing adjustment command based on the degree of deformation, and control the intervention timing of the pressure roller based on the initial fixing adjustment command to complete the adhesive-free tape rewinding process.

Images