improvement of modeling quality through monitoring techniques

modeling is one of the most interesting areas in various fields of science. unfortunately data quality, which has an important role in the modeling, is not considered. in fact, most often processes encounter disturbances which results in the collection of abnormal data and may lead to a model different from the real behavior of the process. on the other hand, most of real industrial processes are time varying and developing on-line models to capture the variations of the process is very appealing. high capability of intelligent models has attracted considerable attention. therefore, on-line intelligent models can effectively characterize both time invariant and time varying processes. current on-line modeling techniques adapts the primarily identified process model with the new changes in time varying processes without consideration of abnormal situations. this will affect the model. to overcome this problem, this paper proposes to combine process monitoring techniques with modeling approaches. although the proposed approach is not restricted to a specific process monitoring or modeling approach, wave-net on-line techniques and recursive principal component analysis (rpca) methods are invoked. a double continuously stirred tank reactor (cstr) is considered as a case study. the results show the effectiveness of the proposed approach.