Real-time monitoring of machine tools via Walsh-Hadamard tranform

Real-time machine monitoring is an open problem in machine tool industry. The presence of unmanned production systems increases the demands on such equipments. Success can be expected only by joint application of signal processing and pattern recognition methods. In real-time applications the speed of the system is of vital importance. In this respect the use of Walsh-Hadamard transform in preprocessing stage of the recognition system is investigated. Fourier and walsh transform of measured vibration data is compared. The paper gives an overview of the applicable methods of classification and feature selection . A multiprocessor-based real-time monitoring system is outlined, which system is under development.