Precautionary statistical criteria in the monitoring quality of technological process

Abstract

The quality of products depends on a stability of production process. In practice to identify reasons of the process degradation, the Shewhart control chart is typically used. At the construction of control X-charts, it is supposed that the dispersion of sample means in subgroups of data measured in process is caused by the influence of random factors and the limited sample size. In such cases it is an improbable event to obtain the output sample values, which are outside the interval ±3σ. Its appearance indicates the presence of systematic influence and it is the need to adjust the controlled parameters of technological process. Based on practical experience in the ISO 7870-2: 2013 standard it is recommended to pay attention to “… any unusual structure of data points, which may indicate about a manifestation of special (non-random) reasons”. In the numerical example presented in this work the analysis of a structure of points on the control chart showed the presence of non-random values, although if the sample mean values were within interval ±3σ. The indicator of existence of non-randomness was the probability that the minimum number of consecutive selective averages, which got to a certain area did not exceed 0.003. As a result of executed analysis the criteria are established and the algorithm is developed. It helped to identify the dysfunction of technological process at an early stage.