Introduction: The analysis process optimization of the complex object technical state based on the dynamic programming method requires the considerable computational expenditure, especially in the large dimension of such objects state table.
Purpose: The development of an algorithm for constructing a flexible analysis program of an object technical state allowing to receive near optimal result with smaller computational expenditure as compared to dynamic programming method.
Methods: Branch and bound method, modified by the authors in relation to analysis process of an object technical state using as an optimization index the measure of information semantic usefulness obtained in case of execution of checks of the discrete diagnostic signs proposed by the academician A. A. Harkevich. Results: In constructing the quasi-optimal algorithm at each step of program functioning it is necessary to select such diagnostic sign to which there corresponds the maximum value of upper bound of semantic usefulness of the obtained information for check. For computation of upper bound of an optimized index the known property of Harkevich's measure was used which means that it will reach the greatest value in case of the maximum distinction of probabilities of result of checks of diagnostic signs. The developed algorithm is presented in the form of the sequential steps allowing to define the minimum set of diagnostic signs the check of which provides recognition of each of the given object technical states, as a rule with maximum semantic usefulness of the obtained diagnostic information. The example of implementation of the developed algorithm illustrating an entity of the offered approach is given. Practical relevance: The proposed algorithm can be used in the development of special software of automated analysis systems of the complex object technical state.