| The traditional CUSUM control charts are based on the assumption that the dis-tribution function is known both before and after the shift.When the distribution of the process is unknown,some authors combined CUSUM statistic with nonparametric methods and proposed some nonparametric CUSUM charts.In this paper,we estimate the in-control density function from historic data and propose a new CUSUM control chart based on logspline density estimation.We consider the additive change and the multiplicative change of the process parameters.We compare the performance of the proposed CUSUM chart with some existing CUSUM charts by numerical simulation.Simulation results show our proposed control chart is more effective than the other existing control charts. |
