Change Point Estimation in the Mean of Multivariate Multiple Linear Profiles by Probabilistic Neural Network

The change point is a useful concept in the statistical process control (SPC) that assists quality engineers to reduce the time and cost of detecting assignable causes. In this paper, probabilistic neural network (PNN) is implemented to estimate the step and drift change points in Phase II monitoring of the mean of multivariate multiple linear profiles. The performance of the probabilistic neural network to estimate the change point is evaluated utilizing Monte Carlo Simulation in terms of accuracy and precision of the estimates. The results of simulations show that the proposed network has entirely better performance than maximum likelihood estimator (MLE) in small shifts, considering mean square error (MSE) criteria, but MLE method owns better performance in medium to large shifts. In general, in all shift types, MLE has a better performance regarding precision, and the proposed PNN performs better in the sense of accuracy. Finally, performance of the proposed probabilistic artificial neural network in an illustrative case is investigated.