The application of wavelet and time series analysis in quality monitoring of digital signal

As system clock frequency increasing, signal integrity is becoming more important, especially for digital signal. However, the signal quality always descends when they are transmitting in the circuit because of all kinds of environmental stress. In many cases, such influences are unavoidable. Monitoring the feature of signal is the key to prognostics. This paper presents an application of wavelet and time Series analysis to quality monitoring of digital signal in PHM. Different from conventional means, this method focus on the composition of signal under test, it extracts the quality feature from the signal by wavelet transform-based multi-scale analysis that decomposes the signal to different levels, then it estimates signal quality based on auto-regression moving average (ARMA) model. This context also describes how to choose an appropriate wavelet basis function depending on the characters of signal, the ability of signal reconstruction and compute speed. The method has been applied to the key AID convert module in some avionic system.