Gertsiuk M. Information technology for river pollution forecasting in emergencies based on regression methods and heuristic algorithms

An information technology development that predicts rivers pollution level with hazardous substances in emergencies scientific task is solved in the dissertation. The work highlights conducted research that allows riverbed pollution forecasting with a limited input characteristics number. The object of study is rivers pollution forecast with dangerous and toxic substances in emergencies. The subject of study is river pollution consequences forecasting methods. The purpose of work is to increase effectiveness of water bodies consequences forecasting process in emergencies, such as toxic and harmful substances discharges into the river bed using the developed information technology. The main tasks of the work are: - analyze existing automated methods, their advantages and disadvantages; - analyze existing approaches to methods creation for river pollution consequences forecasting in emergencies; - develop adjusting pollution concentration forecasting results method; - develop forecasting pollution levels between monitoring points method; - develop determining pollution source method; - develop information technology for forecasting the river pollution consequences in emergencies. Research methods that will allow obtaining new scientific results in mathematical and analytical calculations part are methods of interpolation, recursion, regression, filtering and sorting. Approaches to information systems development and the system different parts interaction with each other became the basis of client-server system development. An obtained results scientific novelty is as follows: - determining pollution source method based on filtering and sorting methods has been developed, which allows to obtain possible potentially dangerous objects list that could pollute the river bed; - pollution concentration forecasting results adjusting method based on neural networks was developed, which, when using regression problems, allows to increase forecasting pollution effects accuracy at existing statistical data expense without necessity to enter detailed input characteristics; - forecasting pollution levels between monitoring points method was developed based on interpolation method using recursion, which allows obtaining a wider range of results and pollution consequences analysis accuracy; - developed information technology for river pollution consequences forecasting in emergency situations based on an information system consisting of different parts and having a defined area of responsibility.