Process Monitoring Using Maximum Sequence Divergence
Yihuang Kang, Vladimir Zadorozhny · D-Scholarship@Pitt (University of Pittsburgh) · 2015
Process Monitoring involves tracking a system's behaviors, evaluating the current state of the system, and discovering interesting events that require immediate actions. In this paper, we propose a process monitoring approach that helps detect the changes of dynamic systems, monitor the divergence of the system development, and evaluate the significance of the deviation. We begin with the discussion of the data reduction and symbolic data representation. Timeseries representation methods are also discussed and used as examples in the proposed approach to discretize the raw data into sequences of system states. Markov Chains and stationary state distributions are continuously generated for sequences to represent the snapshots of the system dynamics in different time frames. We use the Generalized Jensen-Shannon Divergence as a measure to monitor the changes of the stationary symbol probability distributions and evaluate the significance of the system deviation. We prove that the proposed approach is able to detect the deviation of the systems we monitor and assess the deviation significance in probabilistic manner