A Technique of Technical State Control Based on Prediction

The article introduces a concept of technical object state management with prediction which allows increasing efficiency of auto-mated maintenance and repair systems. According to observable object state data, a decision about reasonability of actions aimed to prevent failures or minimize their consequences is made. In order to estimate prediction efficiency a special criterion is formu-lated. It is based on a ratio of availability of a system with prediction to availability of a system without it and shows how an object down state time changes after introduction of prediction. A gain value dependence on parameters of a technical object and maintenance and repair subsystem is found.

Founded on the proposed criterion, an optimal decision rule for upcoming failures detection is inferred.

An example of an object with two states with Gaussian probability density functions is studied. For this example the expressions of optimal decision threshold value are determined, ranges of system parameter amounts where prediction is applicable are defined.