Anomaly Detection in Yarn Tension Signal Using Independent Component Analysis

Abstract

Finding patterns in data that defy expected behavior is what anomaly detection entails. In many application fields, these incorrect patterns are referred to as contaminants, abnormalities, exceptions, or outliers. The significance of anomaly detection is that it helps to identify irregularities in data across a range of application domains and turns them into valuable information. When the yarn tension signals are inspected, anomaly states in the signals are seen in situations where it malfunctions for whatever reason. This distinction makes it possible to predict whether the twister is malfunctioning. So, a bigger issue is avoided. The employment of Cluster-Based Algorithms, Statistical Method Algorithms, and other techniques to identify anomalies is common in the literature. The yarn tension signals in the twisting machines have been analyzed in this work using independent component analysis, and the problematic signal locations have been identified. The proposed method has been contrasted with other ways, and it has produced the highest success rate.