Алгоритм построения квазиоптимальной гибкой программы анализа технического состояния объекта

Abstract

Introduction: The optimization of analyzing the technical state of complex objects using dynamic programming requires considerable computational expenditure, especially when the state table of such objects is large. Purpose: We develop an algorithm for constructing a flexible program which would analyze the technical state of an object. This algorithm should allow you to obtain a result close to the optimum, with smaller computational expenditure as compared to dynamic programming. Methods: We use the branch-and-bound method, having modified it for the object state analysis. As an optimization criterion, we use the measure of semantic usefulness of information (proposed by A.A.Kharkevich) obtained when you check diagnostic signs presented as intervals on a real numerical axis with a uniform distribution law. Results: When developing a quasi-optimal algorithm, on each step of the program operation you have to choose a diagnostic sign for checking, which would be corresponded by the maximum value of the upper bound of semantic usefulness of the obtained information. To calculate the upper bound of an optimized index, we used the well known property of Kharkevich's measure: it reaches its highest value when the probabilities of the diagnostic sign check results are maximally different. The developed algorithm is presented in the form of sequential steps allowing you to define the minimum set of diagnostic signs whose checks provide the recognition of each of the given technical states of the object with the highest average semantic usefulness of the obtained diagnostic information. We provide an example of implementing the developed algorithm, illustrating the gist of the proposed approach. Practical relevance: The proposed algorithm can be used in the development of special software for automated systems of analyzing the technical state of complex objects.