CN-120633381-B - Calibration method and device of flexible touch sensor

CN120633381BCN 120633381 BCN120633381 BCN 120633381BCN-120633381-B

Abstract

The invention discloses a calibration method and device of a flexible touch sensor, wherein the method comprises the steps of obtaining response data of a flexible touch sensor array, and carrying out feature extraction and data preprocessing on the response data; based on standardized training data, a calibration model is generated through convolutional neural network and long-term and short-term memory network training, environmental parameters of the sensor are collected, self-adaptive adjustment is carried out through a Bayesian optimization algorithm, the sensor is conveyed to a multi-station parallel calibration unit for calibration, performance test is carried out on the calibrated sensor, and a unique identification code for quality tracing is generated. The invention solves the technical problems of strong environmental dependence, long calibration time, low efficiency and the like in the calibration process of the traditional flexible touch sensor, realizes the quick, high-precision and self-adaptive environment flexible touch sensor calibration, shortens the calibration time from a few hours to 60 seconds, controls the calibration error to be within +/-2 percent, and meets the requirement of mass production.

Inventors

CHEN XINZHUN
ZHANG BIN
CHENG YUANHONG
MA PENGFEI
LI NA
QIU GUOCAI
GUO LINLIN

Assignees

广州奥松电子股份有限公司

Dates

Publication Date: 20260512
Application Date: 20250514

Claims (8)

1. A method of calibrating a flexible tactile sensor, comprising: Obtaining response data of the flexible touch sensor array, and performing feature extraction and data preprocessing on the response data to obtain standardized training data; Based on the standardized training data, extracting sensor space characteristics through a convolutional neural network, processing the sensor space characteristics through a long-term and short-term memory network to obtain time sequence response characteristics, and taking the obtained sensor space characteristics and time sequence response characteristics as input to train and generate a calibration model, wherein the calibration model is trained and has the capability of outputting calibration parameters according to environmental parameters, and the calibration parameters comprise a temperature compensation coefficient, a humidity correction factor, a pressure excitation range and a loading rate; the method comprises the steps of collecting environment parameters of the flexible touch sensor, adaptively adjusting the calibration model by using a Bayesian optimization algorithm according to the environment parameters to obtain optimized calibration parameters, generating an optimization space of the calibration parameters based on the environment parameters, searching an optimal parameter combination in the optimization space by using the Bayesian optimization algorithm to obtain the optimization parameters, carrying out parameter adjustment on the calibration model according to the optimization parameters, and outputting the optimized calibration parameters through filtering treatment according to a mapping relation between the environment parameters and the calibration parameters formed by training of the calibration model; The flexible touch sensor is conveyed to a multi-station parallel calibration unit, the optimized calibration parameters are utilized to calibrate the flexible touch sensor in parallel, calibration data are generated, and the calibrated flexible touch sensor is obtained, and the flexible touch sensor comprises pressure, temperature and humidity excitation conditions applied to the flexible touch sensor as external stimuli according to the optimized calibration parameters are set so as to simulate working states of the flexible touch sensor in different actual environments; And performing performance test on the calibrated flexible touch sensor, recording production parameters and test results based on the calibration data, and generating a unique identification code for quality tracing.
2. The method of claim 1, wherein the performing feature extraction and data preprocessing on the response data to obtain standardized training data comprises: Extracting signal characteristics by utilizing wavelet transformation based on the response data to obtain characteristic vectors; Performing principal component analysis and dimension reduction processing on the feature vector to eliminate data redundancy and obtain dimension reduced feature data; and carrying out standardization processing on the feature data after the dimension reduction to obtain the standardized training data.
3. The method of claim 1, wherein the extracting the sensor spatial features through the convolutional neural network and the processing through the long-term memory network to obtain the time sequence response features comprises: Processing the standardized training data by utilizing a plurality of convolution layers and a pooling layer to obtain space characteristic data; Inputting the space characteristic data into a long-short-term memory network unit, extracting time sequence characteristics, and obtaining time sequence characteristic data; and carrying out feature fusion on the space feature data and the time sequence feature data to obtain the space-time feature of the sensor.
4. The method of claim 1, wherein the training generates a calibration model, comprising: dividing the standardized training data into a training set and a verification set according to a preset proportion; performing iterative training on the training set by using an optimization algorithm until a preset training precision threshold is reached, so as to obtain an initial calibration model; and performing performance evaluation on the initial calibration model by using the verification set, and obtaining the calibration model when the evaluation index meets the preset condition.
5. The method of claim 1, wherein the acquiring the environmental parameter in which the flexible tactile sensor is located comprises: setting a plurality of sensing nodes in a calibration area, and acquiring environmental data in real time by adopting a wireless transmission mode; configuring a multi-type sensor for each sensing node, wherein the multi-type sensor acquires temperature, humidity and air pressure data to obtain original environment data; and filtering the original environment data to eliminate noise interference and obtain the environment parameters.
6. The method of claim 1, wherein said transporting the flexible tactile sensor to a multi-station parallel calibration unit comprises: performing position identification on the flexible touch sensor by using a visual positioning system to obtain position information; Based on the position information, the flexible touch sensor is transmitted to a calibration station through a manipulator, wherein a closed-loop control strategy is adopted in the transmission process, and the position of the flexible touch sensor is adjusted by monitoring the position of the sensor in real time and comparing the position with a preset calibration position so that the position accuracy of the flexible touch sensor meets a preset range; And fixing the flexible touch sensor on the calibration station by adopting a clamp.
7. The method of claim 1, wherein the performance testing of the calibrated flexible tactile sensor comprises: Based on a preset performance evaluation index, measuring the performance parameters of the calibrated flexible touch sensor to obtain a parameter test result; Based on a plurality of different environmental conditions, testing the response characteristics of the calibrated flexible touch sensor to obtain a response test result; And judging the performance grade of the calibrated flexible touch sensor according to a preset performance grading standard based on the parameter test result and the response test result.
8. A calibration device for a flexible tactile sensor, comprising: the data acquisition module is used for acquiring response data of the flexible touch sensor array, and performing feature extraction and data preprocessing on the response data to obtain standardized training data; The system comprises a calibration model generation module, a calibration module and a control module, wherein the calibration model generation module is used for extracting sensor space characteristics through a convolutional neural network based on the standardized training data, obtaining time sequence response characteristics through long-term and short-term memory network processing, and taking the obtained sensor space characteristics and the time sequence response characteristics as inputs to train and generate a calibration model; The self-adaptive optimization module is used for collecting the environment parameters of the flexible touch sensor, carrying out self-adaptive adjustment on the calibration model according to the environment parameters by using a Bayesian optimization algorithm to obtain optimized calibration parameters, and comprises the steps of generating an optimization space of the calibration parameters based on the environment parameters, searching an optimal parameter combination in the optimization space by using the Bayesian optimization algorithm to obtain the optimization parameters; The calibration execution module is used for conveying the flexible touch sensor to a multi-station parallel calibration unit, carrying out parallel calibration on the flexible touch sensor by utilizing the optimized calibration parameters to generate calibration data, and obtaining the calibrated flexible touch sensor, and comprises the steps of setting pressure, temperature and humidity excitation conditions applied to the flexible touch sensor as external stimulus according to the optimized calibration parameters so as to simulate the working states of the flexible touch sensor under different actual environments; And the quality management module is used for performing performance test on the calibrated flexible touch sensor, recording production parameters and test results based on the calibration data, and generating a unique identification code for quality tracing.

Description

Calibration method and device of flexible touch sensor Technical Field The invention relates to the technical field of flexible electronics, in particular to a calibration method and device of a flexible touch sensor. Background The flexible touch sensor is used as a core element in the emerging fields of wearable equipment, soft robots, man-machine interaction systems and the like, has the characteristics of light weight, flexibility, high sensitivity and the like, and has wide application prospects in the fields of medical health, industrial automation, consumer electronics and the like. At present, flexible touch sensors are mainly designed based on different working principles of resistance type, capacitance type, piezoelectric type and the like, and are generally composed of a multi-layer composite structure of conductive materials, base materials, functional materials and the like. Conventional flexible tactile sensor production techniques rely primarily on screen printing, ink jet printing, micro-nano machining, and the like to produce the sensing element. For example, screen printing technology forms a conductive pattern by transferring conductive ink onto a flexible substrate through a screen, while micro-nano processing technology builds a microstructure on the flexible substrate by using processes such as photolithography and etching, etc., to realize a specific tactile sensing function. The existing flexible touch sensor array batch production technology mainly adopts a roll-to-roll continuous printing manufacturing process, combines an automatic assembly and packaging system, and sets a sensor performance detection and calibration link at the tail end of a production line. The technology prints conductive patterns on a flexible substrate through a full-automatic production line, stacks functional materials, and finally performs cutting, packaging and performance testing to form a complete production flow. However, the prior art has obvious defects in the sensor calibration link, and is mainly characterized in that a strict constant temperature and humidity environment is required in the calibration process, the requirement on production conditions is high, the calibration time is long, the production efficiency is severely restricted, the calibration parameters depend on fixed environment conditions, the sensor has poor performance consistency in actual variable environments, and the calibration device has large volume, fixed design, lacks flexibility and is difficult to adapt to the requirement of large-scale mass production. These problems directly affect the yield, yield and performance stability of the flexible tactile sensor, limiting its scale application. Disclosure of Invention In order to solve the technical problems, the invention provides the calibration method and the device for the flexible touch sensor, which realize the calibration of the flexible touch sensor in a rapid, high-precision and self-adaptive environment and meet the requirement of mass production. The technical scheme of the invention is as follows: a method of calibrating a flexible tactile sensor, comprising: Obtaining response data of the flexible touch sensor array, and performing feature extraction and data preprocessing on the response data to obtain standardized training data; Based on the standardized training data, the sensor space characteristics are extracted through a convolutional neural network, the time sequence response characteristics are obtained through long-term memory network processing, the obtained sensor space characteristics and the time sequence response characteristics are used as input, and a calibration model is generated through training; collecting environment parameters of the flexible touch sensor, and carrying out self-adaptive adjustment on the calibration model according to the environment parameters by using a Bayesian optimization algorithm to obtain optimized calibration parameters; conveying the flexible touch sensor to a multi-station parallel calibration unit, and calibrating the flexible touch sensor in parallel by using the optimized calibration parameters to generate calibration data so as to obtain a calibrated flexible touch sensor; And performing performance test on the calibrated flexible touch sensor, recording production parameters and test results based on the calibration data, and generating a unique identification code for quality tracing. Preferably, the feature extraction and data preprocessing are performed on the response data to obtain standardized training data, including: Extracting signal characteristics by utilizing wavelet transformation based on the response data to obtain characteristic vectors; Performing principal component analysis and dimension reduction processing on the feature vector to eliminate data redundancy and obtain dimension reduced feature data; and carrying out standardization processing on the feature data after the dimension reductio