Fuzzy Inference Methods for Forming Expert Systems for Forecasting Innovative Processes

Abstract

The paper highlights the results of the development of methodological support for the inference mechanism as part of expert systems, which are used for forecasting innovative processes in industrial production enterprises. The relevance of expert systems in the studied subject area is due to the high organizational complexity and significant uncertainty of innovative processes. The inference of logic as part of an expert system performs the function of reproducing decision logic for a given subject area. The solution to this problem is based on the following conditions: the logical conclusion is realized by the criterion of effectiveness; the effectiveness is assessed as a measure of compliance between the required and expected values of three key parameters - the target effect, the cost of resources, the limited time; the performance indicator is presented in the vector form (as a three-dimensional vector of performance parameters); the function of measure of compliance is implemented through fuzzy inference; the content of each of the effectiveness parameters (composition of factors) is disclosed in accordance with the ontology of the innovation process; the composition of effectiveness factors is represented by a structural model; factors are evaluated by deterministic and expert methods; relations between factors are realized in the form of fuzzy rules, as well as through deterministic relationships; the parameters of fuzzy inference  are formed in accordance with the general methodology of fuzzy-set modeling, the inference function is implemented through a weighted evaluation of the rules, and weighting coefficients are determined by expert methods; expert assessment of weighting factors is carried out in the context of the studied strategies of the innovation process. The developed methodical support of the mechanism of fuzzy inference is aimed at taking into account the technical and economic characteristics of the innovation process, the strategic context of its implementation and significant limitations of the organizational system. The developed methods of fuzzy inference provide a correct and intuitive translation of mental reasoning of the decision-maker (in linguistic form) into the language of mathematics; this is facilitated, inter alia, by using the method of weighted evaluation of rules when implementing the inference function that meets the mental criteria of heuristic analysis.