VALIDATION AND RECOVERY OF VIBRATION DATA IN ELECTROMACHINE SYSTEMS USING NEURAL-NETWORK SOFTWARE

Analysis of vibration data from rotating electromachine systems is an important means of machine diagnostics and preventive maintenance. Vib ration signals, which represent the lateral displacement of a shaft, a re obtained from proximity probes and transmitted to a data acquisitio n system. Invariably, data distortions frequently occur because of the high sensitivity of the probes and transmission channels, imperfectio ns of the shaft surface under the probes, and several other factors. T o screen out corrupt data, a tedious and time-consuming visual inspect ion must precede further analysis. This paper describes a successful e ndeavor of using neural network software for validation and recovery o f distorted vibration data. Structure of the developed networks, prepa ratory and training procedures, and experimental results are presented .