Zastosowanie estymacji jądrowej do monitorowania procesu o nieznanym rozkładzie

The control chart is a tool of statistical quality control, which is widely used in factories. The fulfillment of its basic assumptions ensures faultless assessing the monitored process. Infringements the assumptions of classical control charts can cause false signals in the case of a regulated process, either lack of signal or the signal delayed in time, when process is out-of-control. Incorrect assessment of the accuracy of the manufacturing process is of course the economic impact. In this paper, based on actual data an attempt to determine control limits for the manufacturing process of the distribution of the controlled characteristics, which is significantly different from a normal distribution, was taken. The result of this work is the method of determining the control limits based on the quantile of a random variable estimated by the kernel estimation. The article pays attention to the economic consequences of infringements the assumptions of classical control charts.