Identification of nonstationary time series based on SVM-HMM method

Nonstationary time series are occurring when the plant proceeds to an abnormal state or a transient situation from a normal state. So it is necessary to identify the type of fault during its early stages for the selection of appropriate operator actions to prevent a more severe situation. This paper proposes a new architecture for identification of the time series. It converts the output of support vector machine (SVM) into the form of posterior probability which is computed by the combined use of sigmoid function and Gauss model, it acts as a probability evaluator in the hidden states of hidden Markov models (HMM). Experiments show that the architecture is very effective.