Decision Making and Learning in Wagon-Repairing

Abstract

We consider production management using such elements of artificial intelligence as human-machine learning and decision-making procedures with fuzzy or qualitative guidelines. These mechanisms include heuristic elements, and use the knowledge of decision-makers and experts, their language, and the rules formulated in this language. The problem of the control mechanism optimal synthesis, with the guidelines of the expert, is set. This mechanism models the functions of people in conditions of uncertainty. The obtained optimal control mechanism includes learning and stimulation procedures. The proved theorem offers a solution to the problem of optimal synthesis of the learning mechanism using the dichotomous classification procedure, supported by an iterative pattern recognition procedure. This iterative procedure uses both a formal recursive algorithm and the results of recognition by an expert on images of emerging situations. As a result, adaptive norms are formed to evaluate the effectiveness and stimulate production. The use of the optimal control mechanism illustrates by the example of the wagon-repair production of a large-scale corporation Russian Railways.