Analysis of yarn unevenness by using a digital-signal-processing technique

This paper describes a microcomputer system using digital-signal processing for yarn-unevenness analysis. The system converts the analogue signal from an Uster Tester into digital form and then analyzes the signal by using digital-signal-processing techniques. The yarn unevenness is described by four statistical parameters (statistical average, probability-density function, correlation functions, and power-spectral-density function) as usually adopted for a steady stochastic process. The probability-density function, known to be closely correlated with fabric-appearance quality, is derived in detail, and a new scheme for numerical approximation of a shorter-term signal from a longer-term one is proposed. The system can be used to expand the use of the Uster Tester, providing insight for the analysis of yarn unevenness.