Introduction. It is important to have models that adequately describe the relationship between the integral indicators of the functioning of the system with the particular indicators of the lower levels of management in complex control systems, in particular in RSChS. Traditional approaches based on normative models often turn out to be untenable due to the impossibility of covering all aspects of the functioning of such systems, as well as due to the high variability of the environment and the values of the set of target indicators. Recently, adaptive machine-learning models have proven to be productive, allowing build stable and adequate models, one of the variants of which is artificial neural networks (ANN), based on the solution of inverse problems using expert estimates. The relevance of the study lies in the development of compact models that allow assessing the effectiveness of the functioning of complex multi-level control systems (RSChS) in emergency situations, developing according to complex scenarios, in which emergencies of various types can occur simultaneously. Goals and objectives. The purpose of the article is to build and test the technology for creating compact models that are adequate to the system of indicators of the functioning of hierarchically organized control systems. This goal gives rise to the task of choosing tools for constructing the necessary models and sources of initial data. Methods. The research tools include methods for analyzing hierarchical systems, mathematical statistics, machine learning methods of ANN, simulation modeling, expert assessment methods, software systems for processing statistical data. The research is based on materials from domestic and foreign publications. Results and discussion. The proposed technology for constructing a neural network model of the effectiveness of the functioning of complex hierarchical systems provides a basis for constructing dynamic models of this type, which make it possible to distribute limited financial and other resources during the operation of the system according to a complex scenario of emergency response. Conclusion. The paper presents the results of solving the problem of constructing an ANN and its corresponding nonlinear function, reflecting the relationship between the performance indicators of the lower levels of the hierarchical control system (RSChS) with the upper level. The neural network model constructed in this way can be used in the decision support system for resource management in the context of complex scenarios for the development of emergency situations. The use of expert assessments as an information basis makes it possible to take into account numerous target indicators, which are extremely difficult to take into account in other ways. Keywords: emergency situations, hierarchical control system, efficiency, artificial neural network, expert assessments
Control of heat exchangers in series using neural network predictive controllers
