SELECTION OF THE PARAMETERS OF THE GENERALIZED DISPERSION ALGORITHM FOR MULTIVARIATE STATISTICAL CONTROL OF THE PROCESS SCATTERING

Abstract

To control the stability of the functioning of aviation equipment units based on the results of monitoring a group of indicators, methods of statistical processes control can be used. In the presence of significant correlations between performance indicators, multivariate methods are used. In this case, the control of the average level of the process is carried out on the basis of the Hotelling algorithm, the control of multivariate scattering is carried out using the generalized variance algorithm. If, according to the conditions of the process, it is necessary to ensure the fastest detection of a violation, then the optimization problem of finding such values of the sample size, sampling frequency and position of the control boundaries is solved that minimizes the average time of the unstable state of the process. The initial data are the number of process indicators monitored, the target value of the generalized variance (estimated from experimental data), the characteristic of the permissible increase in scattering, the intensity of process disturbances (parameter of the Poisson distribution); time to search for a violation after its detection and time to calculate the sample element.