Artificial Neural Network Based Approach for Control Points Detection in Smart Textile Signals

Abstract

This article proposes a method to study the signal coming from the sensors of the smart fabric. The initial stage of the research aims recognition of so-called control points in the signal, which indicates parameters of an athlete running. This research is part of a larger study which solves the issues of developing a measuring system based on the smart fabric and signal processing research. The article presents the experimental results of applying the method to detect the control points using artificial neural network (ANN). Within the framework of the method, a conversion of dynamic data into static data is carried out through the fragmentation of the smart fabric signal into training patterns, which are then fed to the ANN. Finally, the trained ANN is able to classify the training patterns, allowing one to identify the control points in the operation mode. This research is interdisciplinary: it combines textile engineering, data mining and information processes. The developed methods and results can also be used to study other signals, e.g., those received from sensors of smart medical fabric during the patient rehabilitation period.