A neural network ensemble model for on-line monitoring of process mean and variance shifts in correlated processes

In this paper, a neural network-based identification model is proposed for both mean and variance shifts in correlated processes. The proposed model uses a selective network ensemble approach named DPSOEN to obtain the improved generalization performance, which outperforms those of single neural network. The model is capable of on-line monitoring mean and variance shifts, and classifying the types of shifts without considering the occurrence of both mean and variance shifts in one time. This model is unique since all learning-based methods developed so far can only detect mean or variance shift, but are incapable of classifying types of shifts. The result is significant since it provides additional useful information about the process changes, which can greatly aid identification of assignable causes. The simulation results demonstrate that the model outperforms the conventional control charts in terms of average run length (ARL), and can classify the types of shifts in a real-mode.