Mixed Control Chart Design Using the Integration of Genetic Algorithm and Monte Carlo Simulation

Pramote Charongrattanasakul, Wimonmas Bamrungsetthapong

Abstract

This research proposed a design of mixed control charts to monitoring the process quality based on attribute data together with variable data called a Mixed NE chart. The integration of Genetic Algorithm (GA) and Monte Carlo (MC) simulation are used to simultaneously assess the efficiency of a Mixed NE chart based on three different scenarios of the control limit coefficients (L_np,L_{EWMAX_bar-R}). In this study, MC simulation is used to evaluate the maximum average of run length in case of in-control process ARL₀ while the GA is demanded to optimize the control limit coefficients that obtain the maximum ARL₀. In addition, the efficiency of the Mixed NE chart according to the process shifts are measured using the average of run length in case of out of control process (ARL₁) and the extra quadratic loss (EQL). The results indicate that the Mixed NE chart performed well for small sample size(n), low smoothing constant (Lamda_W,Lamda_Z) and the optimal design of control limit coefficient is L^*_np > L^*_{EWMAX_bar-R} . Moreover, the optimal design of the Mixed NE chart brings new important perspectives to achieve the most efficiency of control chart.