Application of hidden Markov model to transient identification in nuclear power plants

Abstract

In this thesis, a transient identification based on a novel stochastic approach with the hidden Markov model (HMM) has been evaluated experimentally for the classification of nine types of transients in nuclear power plants (NPPs). A transient is defined as when a plant proceeds to an abnormal state from a normal state. Identification of the types of transients during an early accident stage in NPPs is crucial for proper action selection. The transient can be identified by its unique time dependent patterns related to the principal pvariables. The HMM, a double stochastic process can be applied to transient identification which is a spatial and temporal classification problem under a statistical pattern recognition framework. The trained HMM is created for each transient from a set of training data by the maximum-likelihood estimation method which uses a forward-backward algorithm and the Baum-Welch re-estimation algorithm. The transient identification is determined by calculating which model has the highest probability for given test data using the Viterbi algorithm. The training and test are collected from the real-time test simulator. Several experimental tests for the base model are performed, then the selected model is suggested and tested to verify its performance and robustness. For choosing a selected model, experimental tests have been performed with normalization methods, clustering algorithms, and a number of states in HMM. There are also several experimental tests including superimposing random noise, adding systematic error, and untrained transients using the selected model. The selected model is robust within 10% of random noise, the identification rates are quite dependent on the selection of systematic errors and the partial transients are classified as unknown transients. The improved model which are using heuristic training methods is proposed to improve identification accuracy. The selected model and improved model have high classification ...