A STUDY ON FAILURE PREDICTION IN A PLASMA REACTOR

We use several approaches to demonstrate that neural networks can dete ct precursors to failure, That is, they can detect subtle changes in t he process signals. In some cases these subtle changes are early warni ngs that a subsystem failure Is imminent. The results on detection of precursors and faults with various types of time-delay neural networks are discussed. We also measure the noise inherent in our database and place bounds on neural network prediction in the presence of noise. W e observe that the noise level can be as high as 40% for detection of failures and can be at 30% to still detect precursors to failure. We n ote that although self-organizing networks for classification of fault s seems like a good idea, in fact they do not perform well in the pres ence of noise. Lastly, we show that neural networks can induce, or sel f-build, Markov models from process data and these models can be used to predict system state to a significant distance in the future (e,g,, 100 wafers).