基于聚苯胺/聚乙烯醇水凝胶的柔性传感器信号解耦算法研究

结合双网络聚苯胺/聚乙烯醇复合水凝胶(PANI/PVA)各向异性的特性，设计了一种“三明治”夹层式结构的柔性传感器，主要用于管道内复杂受力情况的复合测量。并利用有限元软件验证了传感器对三维空间受力的可测性，搭建了标定试验台。因柔性传感器存在个性化参数差异及受力维间耦合等特性，在标定平台上采集了500组数据，涵盖了多方向、多尺度的重复试验，并设计了一种麻雀搜索算法优化的BP神经网络算法，对测试结果进行解耦，成功将传感器4个电压信号解耦为传感器矢量受力情况。结果表明，相对于单一的BP神经网络，所提算法结果稳定且收敛速度更快，不易陷入局部极值，能更好地完成三维力检测。

Study on Signal Decoupling Algorithm of Flexible Sensor Based onPolyaniline/Polyvinyl Alcohol Hydrogel

A sandwhich-typed flexible sensor has been designed based on the combination of the anisotropic properties of the double network polyaniline/polyvinyl alcohol composite hydrogel (PANI/PVA), which is mainly used for a composite measurement of complex stress in pipelines. Firstly, a finite element software is used to verify the measurability of the sensor for three-dimensional spatial forces, with a calibration platform established as well. Due to such characteristics of flexible sensors as individualized parameter difference and force dimension coupling, 500 groups of data are collected on the calibration platform, covering repeated experiments in multiple directions and scales. A BP neural network algorithm optimized by sparrow search algorithm has been designed to decouple the test results, with four voltage signals of the sensor successfully decoupled into the force condition of the sensor vector. The results show that compared with a single BP neural network algorithm, the proposed algorithm is characterized with an improved stability and a faster convergence speed, not easy to fall into local extremum with an improved performance in the detection of three-dimensional force.