FILTERING,PREDICTION AND SMOOTHING METHODS FOR CORRELATION COEFFICIENT STATIONARY SERIES
Chengrui Liu · Jixie qiangdu · 2003
The correlation coefficient stationary series that is one kind of non-stationary series is familiar in engineering. The prediction formulas of correlation coefficient AR(p), MA(q) and ARMA(p, q) series are given while these series can be directly measured. Under the condition that correlation coefficient series is intermixed with noise, the filtering, prediction and smoothing methods for correlation coefficient AR(p), MA(q) and ARMA(p, q) series are discussed in two parts: one is the independent noise, the other is the coherent noise . The presented methods can analyze the correlation coefficient stationary series accurately in which the mean and variance vary with time, and can be widely used in communication, automatic control, structure response analysis and fault diagnosis et al.