Application of Wavelets for Evaluation of Soft Sensors under Irregular Sampling of Output Measurements

摘要

The problem of evaluating soft sensors for plants with irregular output measurement time is considered. In order to find the relationship between inputs and output, the wavelet-transformed inputs are integrated into the regression model framework. A suitable procedure for wavelet type selection is discussed. An example is provided to illustrate successful application of the proposed approach and an industrial case study is given for soft sensor evaluation of a reactive distillation process.