Film tension dynamic adjusting method based on multi-sensor data fusion

Abstract

The invention relates to the field of automatic control, in particular to a film tension dynamic adjustment method based on multi-sensor data fusion, and the method comprises the steps: completing initial Kalman filtering parameter setting through equipment rated parameters, and carrying out real-time data collection through sensors in a production line; analyzing the energy characteristic and change rate of the residual data of the film tension to obtain an abnormal impact intensity factor of the film tension; performing time sequence correlation evaluation on the film tension residual data to obtain a film tension abnormal event confirmation factor; self-adaptive measurement noise covariance evaluation is carried out through the film tension abnormal impact intensity factor and the film tension abnormal event confirmation factor, and film tension estimation is updated according to the self-adaptive measurement noise covariance; and closed-loop feedback is carried out on the optimal thin film tension estimation value to obtain the winding and unwinding adjusting amount of the thin film, so that the accuracy and stability of the control system are improved.

Description

Technical Field

[0001] The present invention relates to the field of automatic control, and in particular to a film tension dynamic adjustment method based on multi-sensor data fusion. Background Art

[0002] In modern industrial manufacturing, thin film materials serve as functional carriers in a wide range of high-tech applications, including flexible displays, lithium-ion battery motors, high-end optical components, and integrated circuit packaging. During the roll-to-roll production and processing of these thin film materials, such as coating, sputtering, etching, and lamination, precise and stable control of the film substrate tension is a core technology for ensuring product yield and performance. Excessive fluctuations or uneven distribution of tension can directly lead to permanent tensile deformation, wrinkling, and fracture in the film, as well as fatal defects such as uneven thickness and cracking in the functional coating.

[0003] To achieve high-precision tension control, existing technologies typically employ closed-loop feedback control systems. These systems directly measure the film's real-time tension using tension sensors installed at key locations on the production line. Encoders installed on the unwinding and rewinding motors provide indirect information such as line speed and roll diameter. This multi-sensor data, derived from diverse physical sources, enables comprehensive understanding of the system's state. To integrate this noisy and heterogeneous data and extract the tension value that best reflects the actual physical state, advanced control systems commonly employ the Kalman filter algorithm. As an optimal linear state estimation algorithm, the Kalman filter effectively filters out Gaussian white noise, a common fusion process in sensor measurements. This smooth and accurate tension estimate is then fed to a back-end PID (proportional-integral-derivative) controller, which ultimately adjusts the torque or speed of the unwinding and rewinding motors, forming a complete, high-performance tension control closed loop. This method excels in handling random noise under stable operating conditions and is currently a key technical approach for improving film tension control accuracy.

[0004] However, the standard Kalman filter algorithm used in the prior art relies strictly on a core theoretical premise to achieve optimal performance, namely that the system's process noise and measurement noise must both obey a Gaussian distribution, and their statistical characteristics must be known and fixed. In actual production scenarios, especially in high-speed sputtering coating of high-end optical films, there is a typical source of non-Gaussian disturbances that have a serious impact on product quality. Specifically, during the manufacturing process, film substrates inevitably produce tiny physical defects such as crystal points, gel points, or local thickness mutations. When these micro-defects pass through the highly sensitive tension detection roller at speeds of meters per second, they instantly generate a violent pulse-like impact on the tension sensor with an amplitude far exceeding the normal noise level and a very short duration. This signal is a typical non-Gaussian outlier in statistics, which violates the theoretical assumptions of the standard Kalman filter. If the measurement noise parameter of the filter is set to a large value in order to suppress the pulse impact, that is, the trust in the measured value is reduced, then under normal working conditions, the system's response to the actual small tension changes will become sluggish, sacrificing the system's dynamic performance; conversely, if the measured value is trusted to ensure dynamic response, the filter will mistakenly identify the pulse pseudo-signal as a real tension surge and quickly pass this erroneous estimate to the PID controller. After receiving this erroneous spike signal, the controller will immediately perform an erroneous compensation action. This compensation action itself will impose a destructive secondary tension fluctuation on the film, eventually causing wrinkles in the film near the defect point. Therefore, how to identify and effectively suppress this sudden non-Gaussian pulse measurement noise caused by micro-defects in the substrate without sacrificing the normal dynamic performance of the system, and avoid the control system being misled by it and producing destructive adjustment actions, is a technical problem that needs to be solved urgently in the current field. Summary of the Invention

[0005] In view of this, an embodiment of the present invention provides a method for dynamic adjustment of film tension based on multi-sensor data fusion to solve the problem of sudden non-Gaussian pulse measurement noise caused by micro-defects in the substrate causing wrinkles in the film near the defect point.

[0006] An embodiment of the present invention provides a method for dynamically adjusting film tension based on multi-sensor data fusion, the method comprising the following steps:

[0007] Step S1: Initial Kalman filter parameter setting is completed through equipment rated parameters and real-time data collection is performed through sensors in the production line;

[0008] Step S2: Analyzing the energy characteristics and change rate of the film tension residual data to obtain the film tension abnormality impact severity factor;

[0009] Step S3: obtaining a film tension abnormal event confirmation factor by performing time correlation evaluation on the film tension residual error data;

[0010] Step S4: performing adaptive measurement noise covariance evaluation on the film tension abnormal impact intensity factor and the film tension abnormal event confirmation factor, and updating the film tension estimation according to the adaptive measurement noise covariance;

[0011] Step S5: obtaining the film winding and unwinding adjustment amount through closed-loop feedback of the optimal film tension estimation value.

[0012] Preferably, the initial Kalman filter parameter setting is completed through the equipment rated parameters, and real-time data acquisition is performed through the sensors in the production line, which includes:

[0013] The Kalman filter basic model is initialized, the state vector of the film tension is defined as the core component in the Kalman filter basic model, the state transition matrix and the control input matrix that describe the dynamic behavior of the system are determined, and the basic measurement noise covariance and the process noise covariance that represent the noise level of the system under normal working conditions are set;

[0014] The sensor sampling frequency is set, and in each control period, the actual measurement value of the film tension is collected in real time through the sensors deployed on the production line, and the control input instruction applied to the winding and unwinding servo motor is obtained through the central motion controller.

[0015] Preferably, the film tension abnormal impact intensity factor is obtained by analyzing the energy characteristics and the change rate of the film tension residual error data, which includes:

[0016] The impact factor of the film tension is obtained by performing micro-defect impact feature analysis on the residual error data of the film tension, the steepness factor of the film tension is obtained by performing impact rate analysis on the residual error data of the film tension, and the film tension abnormal impact intensity factor is obtained by fusion mapping the impact factor of the film tension and the steepness factor of the film tension.

[0017] Preferably, the impact factor of the film tension is obtained by performing micro-defect impact feature analysis on the residual error data of the film tension, and the steepness factor of the film tension is obtained by performing impact rate analysis on the residual error data of the film tension, which includes:

[0018] The film tension residual error data at the target time is calculated by the actual measurement value of the film tension at the target time and the state prediction value obtained by the Kalman filter prediction step at the target time; the residual error covariance data in the Kalman filter process at the target time is obtained; the square of the film tension residual error data is taken as the numerator, the residual error covariance data is taken as the denominator, and the resulting fraction is taken as the impact factor of the film tension;

[0019] The square of the result of subtracting the film tension residual data at the target moment from the film tension residual data at the moment before the target moment is used as the numerator, the result of multiplying the constant 2 by the residual covariance data is used as the denominator, and the resulting fraction is used as the steepness factor of the film tension.

[0020] Preferably, the step of obtaining the abnormal impact severity factor of film tension by fusing and mapping the impact factor of film tension and the steepness factor of film tension comprises:

[0021] The calculation result of adding the square of the impact factor of the film tension and the square of the steepness factor of the film tension is used as the first impact intensity assessment; the square root of the first impact intensity assessment is mapped to a power function with a natural constant as the base, and the calculation result of subtracting the corresponding mapping result from the constant 1 is used as the abnormal film tension impact intensity factor at the target moment.

[0022] Preferably, the obtaining of the confirmation factor of the abnormal film tension event by performing a time series correlation evaluation on the film tension residual data includes:

[0023] By performing time series correlation evaluation on the film tension residual data, the residual time series correlation accumulation factor of the film tension is obtained; by performing gated mapping processing on the residual time series correlation accumulation factor of the film tension, the confirmation factor of the film tension abnormal event is obtained.

[0024] Preferably, the obtaining of the residual time series correlation accumulation factor of the film tension by performing time series correlation evaluation on the film tension residual data includes:

[0025] A forgetting factor for exponential moving average is set; the film tension residual data at the target moment is multiplied by the film tension residual data at the moment before the target moment, and the result is divided by the residual covariance data at the target moment to obtain a film tension time series correlation evaluation at the target moment; the film tension time series correlation evaluation at the first moment in the time series data formed by the film tension residual data is used as the residual time series correlation accumulation factor at the first moment;

[0026] The forgetting factor is used as the weight of the film tension timing correlation evaluation at the target moment, the constant 1 is subtracted from the forgetting factor as the weight of the residual timing correlation accumulation factor at the moment before the target moment, and the calculation result of the weighted addition of the film tension timing correlation evaluation at the target moment and the residual timing correlation accumulation factor at the moment before the target moment is used as the residual timing correlation accumulation factor at the target moment.

[0027] Preferably, the obtaining of the film tension abnormality event confirmation factor by performing gated mapping processing on the residual time series correlation accumulation factor of the film tension includes:

[0028] set a correlation accumulation activation threshold and an S-shaped function conversion scale factor for judging the event authenticity;

[0029] subtract the residual time sequence correlation accumulation factor of the target moment from the correlation accumulation activation threshold and divide the calculation result by the S-shaped function conversion scale factor to obtain an abnormal event evaluation factor of the target moment;

[0030] map the abnormal event evaluation factor of the target moment through a tanh function to obtain a first mapping result, add the first mapping result to a constant 1 and multiply the calculation result by one-half to obtain a film tension abnormal event confirmation factor of the target moment.

[0031] Preferably, the adaptive measurement noise covariance evaluation through the film tension abnormal impact intensity factor and the film tension abnormal event confirmation factor and the film tension estimation update according to the adaptive measurement noise covariance comprise:

[0032] multiply the film tension abnormal impact intensity factor of the target moment by the film tension abnormal event confirmation factor of the target moment and add the calculation result to a constant 1 to obtain a weight, weight the basic measurement noise covariance, and obtain a calculation result corresponding to the weighted basic measurement noise covariance as the adaptive measurement noise covariance;

[0033] replace the basic measurement noise covariance in the Kalman filter with the adaptive measurement noise covariance, and complete the calculation of the Kalman gain and the update of the system state optimal estimation value and the error covariance according to the process of the Kalman filter, thereby updating the film tension estimation.

[0034] Compared with the prior art, the embodiment of the application has the beneficial effects that:

[0035] The application constructs a film tension abnormal impact intensity factor on the basis of film tension measurement residual error, effectively quantifies the instantaneous abnormal intensity caused by physical disturbances such as microdefects by fusing the instantaneous energy and the change rate characteristics of the residual error, and has higher sensitivity and anti-interference ability than the single judgment mode of the residual error amplitude in the prior art; the application further constructs a film tension abnormal event confirmation factor, dynamically judges whether the abnormal signal has physical time continuity by calculating the time correlation accumulation between the residual errors at consecutive time points, and the mechanism can effectively suppress the isolated peak misjudgment caused by incidental events such as electromagnetic interference and servo equipment switching, significantly reduces the false trigger rate, and improves the response accuracy of the filter to real events; further, the application controls the release degree of the impact intensity factor through the event confirmation factor, jointly generates an adaptive measurement noise covariance, and feeds back to the Kalman filter in real time, dynamically adjusts the trust degree of the measurement value, and the double-factor gating mechanism makes the system maintain high dynamic response characteristics under normal working conditions, has rapid suppression ability when abnormal interference occurs, and realizes the compatibility improvement of response speed and robustness. BRIEF DESCRIPTION OF DRAWINGS

[0036] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0037] Figure 1 It is a method flow chart of a film tension dynamic adjustment method based on multi-sensor data fusion provided by the embodiment one of the present application. DETAILED DESCRIPTION

[0038] The embodiments of the present application will be described in detail below, and examples of the embodiments are shown in the drawings. The embodiments described below by referring to the drawings are exemplary and are intended to explain the present application, and cannot be understood as a limitation of the present application.

[0039] It should be noted that the terms "first", "second" and the like in the specification of the present application and the above drawings are used to distinguish similar objects, and do not necessarily indicate a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all the embodiments consistent with the present application. On the contrary, they are only examples of devices and methods consistent with some aspects of the present application.

[0040] In order to illustrate the technical solutions of the present application, the following will be described by specific embodiments.

[0041] See also Figure 1 , is a flow chart of a method for dynamically adjusting film tension based on multi-sensor data fusion provided by the first embodiment of the present invention, such as Figure 1 As shown, the method may include:

[0042] Step S1, complete the initial Kalman filter parameter setting through the equipment rated parameters and perform real-time data collection through the sensors in the production line.

[0043] The present invention is an optimization method implemented within the framework of a standard Kalman filter system for film tension control. First, a basic Kalman filter model must be established and initialized based on the specific characteristics of the production line being used. Establishing this model is conventional in the art and specifically includes defining a state vector whose core component in the basic Kalman filter model is film tension; determining a state transfer matrix and a control input matrix that describe the system's dynamic behavior; and setting a basic measurement noise covariance and process noise covariance that characterize the system's noise level under normal operating conditions. These matrices and parameters are set according to the technical manual provided by the equipment manufacturer. Simultaneously, the initial value of the state vector and the initial error covariance are set. The initial value of the state vector is set to the initial physical quantity measured by calibration equipment at production startup. The initial error covariance, which reflects an assessment of the reliability of the initial state value, is set to a diagonal matrix with large diagonal elements, indicating that the initial state has high uncertainty, ensuring that the filter can quickly converge with subsequent measurements. In this embodiment, the initial error covariance is set to a diagonal matrix with 100 diagonal elements and units of N / m (Newtons per meter) (corresponding to The variance estimation of ), the diagonal elements can be adjusted according to the actual scenario and are not required.

[0044] After the basic model is established and initialized, the system enters real-time operation. Subsequent steps in the present invention are executed in discrete control cycles, defined by the system's sampling frequency. In this embodiment, the sampling frequency is set to 1000 Hz. During each control cycle, the Kalman filter system collects and digitizes raw data in real time from sensors deployed on the production line for subsequent use. This data includes: the actual film tension measurement at the current moment, obtained through high-sensitivity tension detection rollers based on strain gauge technology installed at key locations on the production line; and the control input commands currently being applied to the rewinding and unwinding servo motors, obtained through the central motion controller.

[0045] At this point, the initial Kalman filter parameter setting is completed through the equipment rated parameters and real-time data collection is completed through the sensors in the production line.

[0046] Step S2: Analyze the energy characteristics and change rate of the film tension residual data to obtain the impact severity factor.

[0047] The problem addressed by the present invention stems from physical micro-defects present in film substrates. When such micro-defects pass through the tension detection roller, the disturbances they cause in the film tension data collected by the sensor exhibit two physical characteristics distinct from normal Gaussian noise: first, the instantaneous energy carried by the disturbance signal far exceeds the energy statistical range of normal noise; second, the energy of the disturbance signal exhibits sudden increases, meaning that the rate of change of energy is also far greater than the random fluctuations of normal noise. The standard Kalman filter algorithm used in the prior art can theoretically only process noise whose energy distribution conforms to Gaussian characteristics. It cannot effectively process pulse shocks that exhibit both high instantaneous energy and a high rate of energy change, and can even be misleading. Therefore, in order to effectively identify and quantify this specific pulse impact, the first step of the present invention is not to simply determine the film tension residual, that is, whether the amplitude of the difference between the film tension measurement value and the film tension prediction value is too large, but to construct a measurement indicator that can simultaneously capture the above two energy characteristics. Only when a signal exhibits extremely high instantaneous energy and extremely drastic energy change rate at the same time, there is sufficient basis for preliminarily determining that it originates from the physical impact of micro-defects.

[0048] Based on this logic, the present invention first obtains the impact factor of the film tension by performing micro-defect impact characteristic analysis on the residual data of the film tension; then obtains the steepness factor of the film tension by performing impact rate analysis on the residual data of the film tension; finally, obtains the abnormal impact intensity factor of the film tension by fusing and mapping the impact factor of the film tension with the steepness factor of the film tension.

[0049] Specifically, the method first calculates the residual film tension data at the target moment by combining the actual measured film tension value at the target moment with the state prediction value obtained through the Kalman filter prediction step at the target moment; obtains the residual covariance data during the Kalman filter process at the target moment; uses the square of the residual film tension data as the numerator, the residual covariance data as the denominator, and the resulting fraction as the impact factor of the film tension; then, uses the square of the result of subtracting the residual film tension data at the target moment from the residual film tension data at the moment immediately before the target moment as the numerator, the result of multiplying the residual covariance data by a constant 2 as the denominator, and the resulting fraction as the steepness factor of the film tension. Finally, the result of adding the square of the impact factor of the film tension to the square of the steepness factor of the film tension is used as the first impact severity assessment; the square root of the first impact severity assessment is mapped to a power function with a natural constant as the base, and the result of subtracting the corresponding mapping result from the constant 1 is used as the abnormal film tension impact severity factor at the target moment.

[0050] In one embodiment, assuming that The residual data of film tension at the moment is ; in the The residual covariance of the film tension at each moment is , then in The calculation formula of the film tension abnormal impact intensity factor at a moment is:

[0051]

[0052] in, Indicates the Abnormal film tension impact intensity factor at a certain moment; represents the natural constant e; Indicates the Film tension residual data at each moment; Indicates the The covariance of the residual film tension at each moment; Indicates the The residual data of film tension at each moment.

[0053] It should be noted that in the calculation formula of the film tension abnormal impact intensity factor, The impact factor of the film tension is represented by the square of the residual, which represents the instantaneous energy of the film tension measurement data and is compared with the expected variance. By comparing the residual energy, the standardization of residual energy is achieved. This term directly responds to the high energy characteristics of micro-defect impact. When the residual energy far exceeds its normal expectation, the value of this term will increase significantly, thus capturing the amplitude information of the impact. Represents the steepness factor of the film tension, which is used to capture the suddenness of the impact. The square of the difference between the residuals before and after the two moments represents the rate of change energy of the film tension measurement data, and the expected variance is also used to calculate it. Normalization is performed to quantify the steepness of the film tension measurement data. When the measurement data changes drastically, the steepness factor will rise sharply, thereby capturing the impact rate information. In the form of a norm, the impact factor of film tension and the steepness factor of film tension are integrated, and this comprehensive measurement value is used as the power of the natural exponent, and the constant 1 is subtracted as the film tension abnormal impact intensity factor. When the system is in normal working conditions and the value of the comprehensive measurement is close to 0, the value of the film tension abnormal impact intensity factor is also close to 0, which has no impact on the system. However, once the micro-defect impact causes the comprehensive measurement value to increase, the amplification effect of the exponential function will cause the value of the film tension abnormal impact intensity factor to grow exponentially, generating a sufficiently strong adjustment signal. The final value of the film tension abnormal impact intensity factor becomes a non-negative potential adjustment amplitude that can quantify the severity of the micro-defect impact.

[0054] At this point, the impact intensity factor is obtained by analyzing the energy characteristics and change rate of the film tension residual data.

[0055] Step S3: Obtain a confirmation factor for an abnormal film tension event by performing a time series correlation evaluation on the film tension residual data.

[0056] In step S2, the shock severity factor is obtained. However, this shock severity factor can only assess the severity of the shock from a single time slice and cannot distinguish the source of the disturbance. In complex industrial environments, in addition to real pulse shocks caused by physical micro-defects in the film, there may also be isolated or transient electromagnetic interference signals caused by equipment startup and shutdown and power grid fluctuations. If all severe shocks are indiscriminately suppressed, these non-physical pseudo-signals will be incorrectly over-adjusted. A real shock caused by a physical micro-defect must be temporally continuous in the film tension measurement data. That is, in the short period of time before and after the shock, the film tension residual data has significant time-series correlation. In contrast, the signal values ​​before and after an isolated random electromagnetic interference spike are statistically independent and do not have this correlation. Therefore, after the initial detection of a severe shock, the present invention introduces a confirmation mechanism to determine whether the shock is a real event with physical continuity by analyzing the inherent temporal correlation of the film tension residual data. Only after the authenticity of the impact event is confirmed should the system authorize the corresponding suppression adjustment. Therefore, a confirmation mechanism is required. By quantifying the temporal correlation of the film tension residual data, a gating signal is generated to determine whether the adjustment action is executed. In this step, the residual temporal correlation accumulation factor of the film tension is first obtained by performing a temporal correlation evaluation on the film tension residual data. Then, a gated mapping process is performed on the residual temporal correlation accumulation factor of the film tension to obtain the confirmation factor of the film tension abnormality event.

[0057] Specifically, first, a forgetting factor for exponential moving average is set; the film tension residual data at the target moment is multiplied by the film tension residual data at the moment before the target moment, and the result is divided by the residual covariance data at the target moment as the film tension time series correlation evaluation at the target moment; the film tension time series correlation evaluation at the first moment in the time series data formed by the residual data of the film tension is used as the residual time series correlation accumulation factor at the first moment; the forgetting factor is used as the weight of the film tension time series correlation evaluation at the target moment, a constant 1 is subtracted from the forgetting factor as the weight of the residual time series correlation accumulation factor at the moment before the target moment, and a calculation result of weighted addition of the film tension time series correlation evaluation at the target moment and the residual time series correlation accumulation factor at the moment before the target moment is used as the residual time series correlation accumulation factor at the target moment.

[0058] In one embodiment, the The calculation formula of the residual time series correlation accumulation factor at each moment is:

[0059]

[0060] in, Indicates the The cumulative factor of residual time series correlation at each moment; Indicates the The cumulative factor of residual time series correlation at each moment; Represents the forgetting factor used for the exponential moving average. In this embodiment, in order to effectively capture short-term pulse events lasting for several sampling periods, the forgetting factor is set to 0.1. The forgetting factor can be adjusted according to actual scenarios and is not required. Indicates the Film tension residual data at each moment; Indicates the Film tension residual data at each moment; Indicates the The residual covariance of the film tension at each moment.

[0061] It should be noted that in order to quantify the time series correlation of the film tension residual data, the present invention introduces the residual time series correlation accumulation amount , the core of the calculation formula is , that is, the product of the residual at the current moment and the residual at the previous moment. For a real pulse shock caused by a physical micro-defect, the sign of the residual has a high probability of being consistent during the occurrence and duration of the shock, which will lead to The residual time series correlation accumulation factor is continuously positive, so that the residual time series correlation accumulation factor is steadily accumulated and increased by the exponential moving average. For normal Gaussian white noise, the residual values ​​before and after the moment are statistically unrelated, and the signs of the film tension residual data will appear alternately positive and negative. The expectation is 0, and the residual time series correlation accumulation factor will fluctuate in a small range around 0. Divide by the residual covariance Normalization can eliminate the influence of the film tension measurement data itself, making it a dimensionless statistical correlation measure, so that the value of the residual time series correlation accumulation factor becomes an effective indicator to measure whether there is a continuous physical impact time in the recent period.

[0062] After obtaining the residual time series correlation accumulation factor of the target moment, set the correlation accumulation activation threshold and the S-type function conversion scale factor for judging the authenticity of the event; subtract the residual time series correlation accumulation factor of the target moment from the correlation accumulation activation threshold and divide the result by the S-type function conversion scale factor as the abnormal event assessment factor of the target moment; map the abnormal event assessment factor of the target moment through the tanh function, and obtain the first mapping result accordingly, add the first mapping result to the constant 1 and multiply by half to obtain the calculated result as the film tension abnormal event confirmation factor of the target moment.

[0063] In one embodiment, the The calculation formula for the confirmation factor of the abnormal film tension event at a moment is:

[0064]

[0065] in, Indicates the Confirmation factor of abnormal film tension event at a certain moment; Indicates the The cumulative factor of residual time series correlation at each moment; represents the cumulative activation threshold of correlation used to judge the authenticity of an event; represents the S-type function conversion scale factor; Represents the tanh function.

[0066] It should be noted that the determination of the correlation cumulative activation threshold used to judge the authenticity of the event in the calculation formula of the confirmation factor of the abnormal film tension event follows the standard engineering calibration method, that is, it is obtained by offline statistical analysis of a large amount of historical sensor data collected under normal and stable operating conditions of a specific production line. The basic principle of its setting is to maximize the ability to suppress random noise while ensuring sufficient sensitivity to real physical impact events, thereby achieving an extremely low false trigger rate. In this embodiment, the application scenario is an optical thin film magnetron sputtering production line for processing PET substrates, the film running line speed is 1.5 meters / second, and the tension sensor sampling frequency is 1000Hz. Under this specific working condition, the threshold is determined by the above-mentioned statistical calibration method. The threshold is set to under normal working conditions. The mean of the value distribution plus a safety margin sufficient to cover more than 99.9% of normal fluctuations. In this embodiment, the specific value is set to , this value can be adjusted according to different working conditions and is not required. , which is used to control the smoothness of the transition of the gating signal from 0 to 1. This value determines the decisiveness of the confirmation process. In the embodiment of the present invention, in order to achieve a relatively fast and clear gating switch, the specific value is set to For the obtained residual time series correlation accumulation factor, the role of the film tension abnormal event confirmation factor is to convert the residual time series correlation accumulation measurement value into a standardized gating signal. The present invention uses the hyperbolic tangent function tanh to construct an S-type conversion function. When Far less than the preset threshold When , it indicates that the system is in normal working condition or encounters an isolated noise point. At this time, the input of the tanh function is a large negative number, and its output value approaches negative one, which eventually makes the value of the confirmation factor of the abnormal film tension event approach 0, indicating that the gate is closed. When a real pulse impact causes Continuous accumulation and significantly exceeding the threshold When , the input of the tanh function becomes a large positive number, and its output value approaches 1, which eventually makes the value of the film tension abnormality event confirmation factor approach 1, indicating that the gate is fully open. By embedding the tanh function into The output value range is changed from Map to interval, so that the final value of the film tension event confirmation factor becomes a gating signal between 0 and 1, which can accurately quantify the confidence level of the authenticity of the impact event.

[0067] Step S4, the film tension abnormal impact intensity factor and the film tension abnormal event confirmation factor are used to adaptively measure the noise covariance and update the film tension estimation according to the adaptive measurement noise covariance.

[0068] After the film tension abnormal impact intensity factor at the target time and the film tension abnormal event confirmation factor at the target time are obtained, this step continues to construct the adaptive measurement noise covariance, and uses the adaptive measurement noise covariance to complete the updating process of the standard Kalman filter. Specifically, the calculation result of multiplying the film tension abnormal impact intensity factor at the target time and the film tension abnormal event confirmation factor at the target time and adding a constant 1 is used as a weight to weight the basic measurement noise covariance, and the calculation result corresponding to the weighted basic measurement noise covariance is used as the adaptive measurement noise covariance.

[0069] It should be noted that the gating modulation mechanism is realized, that is, the film tension abnormal event confirmation factor determines to what extent the adjustment amplitude calculated by the film tension abnormal impact intensity factor can be actually executed. Only when the film tension abnormal event confirmation factor confirms the authenticity of the event (the value of the film tension abnormal event confirmation factor tends to 1), the adjustment effect of the film tension abnormal impact intensity factor is completely released. On the contrary, if the film tension abnormal event confirmation factor determines that the current impact is isolated noise (the value of the film tension abnormal event confirmation factor tends to 0), the adjustment effect of the film tension abnormal impact intensity factor will be effectively inhibited, thereby avoiding unnecessary disturbance to the system.

[0070] After the adaptive measurement noise covariance is obtained, the adaptive measurement noise covariance is used to replace the basic measurement noise covariance in the Kalman filter, and the calculation of the Kalman gain and the updating of the system state estimation value and the error covariance are completed according to the process of the Kalman filter, thereby updating the film tension estimation.

[0071] Step S5, the film winding and unwinding adjustment amount is obtained through closed loop feedback of the optimal film tension estimation value.

[0072] In step S2, the tension estimation value that can accurately reflect the real physical state of the film is obtained by the proposed double-factor gating adjustment mechanism. In the tension estimation value, the non-Gaussian impulse noise interference caused by the micro-defects has been effectively suppressed. The purpose of this step is to use the optimized estimation value to complete the closed-loop feedback control of the film tension, thereby finally solving the technical problems proposed in the background art. The system sends the tension estimation value output by step S2 as a closed-loop feedback quantity into the PID controller, which compares the feedback quantity with the preset tension target value and calculates the deviation between the two. Then the PID controller calculates the control adjustment quantity according to the proportional, integral, and derivative parameters set in it. The control adjustment quantity is then packaged into a digital target torque instruction and sent to the servo driver that controls the take-up and pay-off motor. After receiving the instruction, the servo driver controls and adjusts the output torque of the servo motor according to the target torque instruction. Since the tension estimation value input into the PID controller reflects the real physical state and avoids the misleading of false peak signals, the adjustment output by the PID controller will no longer produce destructive compensation actions. The actual tension of the film is stably maintained near the target value, and even in the case of micro-defects in the film substrate, the tension can be uniform and stable throughout the production process.

[0073] The above examples are only used to illustrate the technical solutions of the present application, and not to limit it; although the present application has been described in detail with reference to the foregoing examples, those skilled in the art should understand that they can still modify the technical solutions recorded in the foregoing examples, or make equivalent replacements for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application.

Claims

1. A method for dynamic film tension adjustment based on multi-sensor data fusion, characterized in that: The method for dynamically adjusting film tension based on multi-sensor data fusion includes: Step S1: Initial Kalman filter parameter setting is completed through equipment rated parameters and real-time data collection is performed through sensors in the production line; Step S2: Analyzing the energy characteristics and change rate of the film tension residual data to obtain the film tension abnormality impact severity factor; Step S3: Obtaining a confirmation factor for an abnormal film tension event by performing a time series correlation evaluation on the film tension residual data; Step S4: performing adaptive measurement noise covariance evaluation based on the film tension abnormality impact severity factor and the film tension abnormality event confirmation factor and updating the film tension estimation based on the adaptive measurement noise covariance; Step S5: Obtaining the film winding and unwinding adjustment amount through closed-loop feedback of the optimal film tension estimation value.

2. The method for dynamic film tension adjustment based on multi-sensor data fusion according to claim 1, characterized in that: The initial Kalman filter parameter setting is completed by using the equipment rated parameters and real-time data collection is performed by using sensors in the production line, including: Initialize the Kalman filter basic model, define the state vector of the membrane tension as the core component in the Kalman filter basic model, determine the state transfer matrix and control input matrix that describe the dynamic behavior of the system, and set the basic measurement noise covariance and process noise covariance that characterize the noise level of the system under normal operating conditions; The sensor sampling frequency is set. In each control cycle, the actual measured value of the film tension is collected in real time through the sensors deployed on the production line, and the control input instructions applied to the rewinding and unwinding servo motor are obtained through the central motion controller.

3. The method for dynamic film tension adjustment based on multi-sensor data fusion according to claim 1, characterized in that: The analysis of the energy characteristics and change rate of the film tension residual data to obtain the film tension abnormal impact severity factor includes: The impact factor of film tension is obtained by performing micro-defect impact characteristic analysis on the residual data of film tension; the steepness factor of film tension is obtained by performing impact rate analysis on the residual data of film tension; and the abnormal impact intensity factor of film tension is obtained by fusion mapping the impact factor of film tension and the steepness factor of film tension.

4. The method for dynamic film tension adjustment based on multi-sensor data fusion according to claim 3, characterized in that: The impact factor of the film tension is obtained by performing micro-defect impact characteristic analysis on the residual data of the film tension; By performing impact rate analysis on the residual data of film tension, the steepness factor of film tension is obtained, including: Calculating residual data of the film tension at the target moment by using the actual measured value of the film tension at the target moment and the state prediction value obtained through the Kalman filter prediction step at the target moment; obtaining residual covariance data during the Kalman filter process at the target moment; using the square of the film tension residual data as the numerator and the residual covariance data as the denominator, and using the resulting fraction as the impact factor of the film tension; The square of the result of subtracting the film tension residual data at the target moment from the film tension residual data at the moment before the target moment is used as the numerator, the result of multiplying the constant 2 by the residual covariance data is used as the denominator, and the resulting fraction is used as the steepness factor of the film tension.

5. The method for dynamic film tension adjustment based on multi-sensor data fusion according to claim 3, characterized in that: The method of obtaining the abnormal impact severity factor of film tension by fusing and mapping the impact factor of film tension and the steepness factor of film tension includes: The calculation result of adding the square of the impact factor of the film tension and the square of the steepness factor of the film tension is used as the first impact intensity assessment; the square root of the first impact intensity assessment is mapped to a power function with a natural constant as the base, and the calculation result of subtracting the corresponding mapping result from the constant 1 is used as the abnormal film tension impact intensity factor at the target moment.

6. The method for dynamic film tension adjustment based on multi-sensor data fusion according to claim 1, characterized in that: The method of obtaining the confirmation factor of the abnormal film tension event by performing a time series correlation evaluation on the film tension residual data includes: By performing time series correlation evaluation on the film tension residual data, the residual time series correlation accumulation factor of the film tension is obtained; by performing gated mapping processing on the residual time series correlation accumulation factor of the film tension, the confirmation factor of the film tension abnormal event is obtained.

7. The method for dynamic film tension adjustment based on multi-sensor data fusion according to claim 6, characterized in that: The method of performing time series correlation evaluation on the film tension residual data to obtain the residual time series correlation accumulation factor of the film tension includes: A forgetting factor for exponential moving average is set; the film tension residual data at the target moment is multiplied by the film tension residual data at the moment before the target moment, and the result is divided by the residual covariance data at the target moment to obtain a film tension time series correlation evaluation at the target moment; the film tension time series correlation evaluation at the first moment in the time series data formed by the film tension residual data is used as the residual time series correlation accumulation factor at the first moment; The forgetting factor is used as the weight of the film tension timing correlation evaluation at the target moment, the constant 1 is subtracted from the forgetting factor as the weight of the residual timing correlation accumulation factor at the moment before the target moment, and the calculation result of the weighted addition of the film tension timing correlation evaluation at the target moment and the residual timing correlation accumulation factor at the moment before the target moment is used as the residual timing correlation accumulation factor at the target moment.

8. The method for dynamic film tension adjustment based on multi-sensor data fusion according to claim 6, characterized in that: The method of obtaining a confirmation factor of an abnormal film tension event by performing gated mapping processing on the residual time series correlation accumulation factor of the film tension includes: Setting the correlation cumulative activation threshold and the S-type function conversion scale factor for judging the authenticity of the event; The residual time series correlation accumulation factor at the target moment is subtracted from the correlation accumulation activation threshold and the resultant is divided by the S-type function conversion scale factor to obtain a calculation result as the abnormal event assessment factor at the target moment; The abnormal event assessment factor at the target moment is mapped through the tanh function to obtain a first mapping result. The first mapping result is added to a constant 1 and then multiplied by half to obtain a calculation result as the film tension abnormal event confirmation factor at the target moment.

9. The method for dynamic film tension adjustment based on multi-sensor data fusion according to claim 1, characterized in that: The method of performing adaptive measurement noise covariance evaluation based on the film tension abnormality impact severity factor and the film tension abnormality event confirmation factor and updating the film tension estimation based on the adaptive measurement noise covariance includes: The film tension abnormality impact severity factor at the target moment is multiplied by the film tension abnormality event confirmation factor at the target moment and the result of the addition of the result and the constant 1 is used as a weight to weight the basic measurement noise covariance, and the calculation result of the corresponding weighted basic measurement noise covariance is used as the adaptive measurement noise covariance; The adaptive measurement noise covariance replaces the basic measurement noise covariance in the Kalman filter, and the calculation of the Kalman gain is completed according to the process of the Kalman filter, and the optimal estimate of the system state and the error covariance are updated, thereby updating the film tension estimation.

Images