Qualitative measure of the environment risk level for the fuzzy control systems of environmental safety

• Published in: Journal of physics. Conference series Vol. 1703; no. 1; pp. 12046 - 12053
• Main Authors: Novikova, S V, Tunakova, Y A, Shagidullin, A R, Novikova, K N
• Format: Journal Article
• Language: English
• Published: Bristol IOP Publishing 01.12.2020
• Subjects: Algorithms; Artificial neural networks; Control systems; Criteria; Decisions; Environmental management systems; Fuzzy control; Fuzzy logic; Fuzzy systems; Management decisions; Model testing; Neural networks; Physics; Qualitative analysis; Regression models; Risk levels
• ISSN: 1742-6588; 1742-6596
• DOI: 10.1088/1742-6596/1703/1/012046

When creating an environmental management system that should generate a control actions set for regulating the quality of natural objects, it is necessary to be able to "calculate" the qualitative value of the environment state. A qualitative assessment of the environment state is a generalized measure of natural risk. For these purposes, it is very convenient to use the so-called "soft" models that take into account the fuzzy nature of information - a fuzzy inference system and neuro-fuzzy systems. The main advantage of the neuro-fuzzy model over the regression one is the consideration of the ambiguity and fuzziness of the concepts "Acceptable", "Medium", "High", "Unacceptable" and "Catastrophic" risks, as well as the possibility of adaptive correction due to the training of a fuzzy neural network on experimental data. The calculation of the qualitative values for degree of danger to the environment is important not only in itself, but also as a criterion for making management decisions to normalize or prevent dangerous situations. To develop environmental recommendations on management impacts, the concept of "Risk Measure" is considered in the work, which corresponds to the management decisions made. The criterion "Risk Measure" takes values in the range from 1 to the number of tuples of the experimental sample, and the higher the Measure value, the more global the measures are recommended. Each value can be associated with either a single action or a set of actions. Thus, greater flexibility in the use of the system is achieved, and the range of recommended measures that can compensate for the emerging risks is significantly expanded. The article describes an algorithm for calculating a quality risk measure. The theory is confirmed by computational experiments using in fuzzy control systems.