The subject of this paper is multiresolution analysis by wavelet representation of actual running data of a glass melting furnace. The results of this analysis are further analyzed through statistical analysis such as auto regressive model-building, and finally the process behavior is identified. The variables of particular frequency ranges can be chosen and the original data of variables can be reconstructed from decomposed data in each range. Then the dynamics between the explicit and implicit variables can be investigated. The use of wavelet decomposition and reconstruction is thought to be an efficient and instructive approach for identification and control in such environments as glass melting furnaces, in which the characteristics are combined with high and low frequency ranges, and include long, variable time lags and high noise levels. In short, judging from the case study, it is our evaluation that these approaches are very useful for identification and control.
展开▼
