Abstract
This paper is concerned with the change detection of the statistics for the stochastic process whose probability density belongs to the exponential families including the gaussian distribution, the exponential distribution, the gamma distribution and others. Combining the change mode with the occurrence time of the statistics change, the countable and exclusive events may be hypothesized. It is possible to detect the change of the statistics on basis of the Bayesian theorem. It is shown from the asymptotical property of a posteriori probabilities that the change of the statistics may be detect from online computations.
The validity of the proposed method has been confirmed from the digital simulation and we applied this method to the cement plant to detect the change of the mean values in the ingredient ratios of raw materials. The results of the application to the actual cement plant have shown that the proposed method is useful.
