Diagnostics of the State of Safety-Oriented Enterprise Management System Using Neural Networks

Abstract

Enterprise management is based on the need to make and justify management decisions that contribute to its development. It is almost impossible to determine the risk of a particular managerial decision, and excessive risk in the implementation of individual projects can lead to loss of business. Therefore, management faces the need to find a balance between benefits and risks, at which, on the one hand, it will be possible to develop a company and, on the other hand, adhere to postulates of safetyoriented management. Since management decisions cannot be foreseen for all possible situations and combinations of risk-benefit ratios, a universal model is proposed. It implies a golden ratio, depending on the limited number of current conditions, that would satisfy an enterprise management from the standpoint of sufficient justification on a decision. The article proposes a probabilistic neural network architecture and Matlab parameters of a probabilistic neural network for diagnosing the states of a safety-oriented control system. The proposed model in the form of a probabilistic neural network generates a response to input data on previous month under estimation, and forms an optimal state for a next month.