Search academic texts

Information × Registration Number 0209U010741, 0108U001096 , R & D reports Title Development of methods and information technologies on the basis of stochastic Bayes networks for forecasting of a condition of ecological safety of coastal water areas of Black sea popup.stage_title Head Bardachev Jury Nikolaevich, Registration Date 23-12-2009 Organization Kherson national technical university popup.description2 Object of research - hydrobiological systems as multifactorial systems of difficult interdependent processes. The project purpose is working out of theoretical bases Bayes networks, numerical methods of their construction and likelihood mathematical models of pollution of coastal water areas of Black sea, for estimation, forecasting and the prediscretion of the difficult interdependent situations connected by anthropogenous activity of ecologically dangerous objects which it is characterised by high degree of uncertainty owing to discrepancy and noise pollution available biological data and which have finally likelihood character. Research receptions: the theory likelihood Bayes a conclusion, the theory of counts, numerical methods of combinatory optimisation and digitization casual replaceable, the theory of hydrobiological and ecological systems Theoretical results. In work advantage of use to an estimation of hydrobiological systems of a computing paradigm of Bayes Networks is described. On the basis of the formation theory the likelihood conclusion is offered a technique of construction of networks Bayes for the description of a statics and dynamics of processes of pollution of coastal water areas of Black sea with the further purpose of forecasting of development of hydrobiological processes, definition of damages to dynamic systems taking into account possible uncertainties different character. It is executed comparisons of measure Bayes with a measure of the minimum length and it is executed asimptotical the analysis of characteristics of this measure. Numerical methods of realisation Bayes procedures are considered. In particular, approximation Laplace, iterative quadrature approximation also is significant digitization of a basis of method Monte-Carlo. The developed methodology of construction of dynamic networks Байеса for problem solving short-term forecasting. Procedure of an establishment of dependences between key replaceable and definition of parametres of a network in problems of forecasting and situational likelihood modelling is offered. Practical results. On the basis of network Bayes the information system is developed for the analysis and forecasting and situational modelling of influence of technogenic objects on the general condition бентоса the Sevastopol bays. Working out of information system is based on the real data received at monitoring researches of a condition benthos in the Sevastopol bays. The software for performance of computing experiments with existing and created networks is created. The led examples of practical application of the considered methods to the decision of a wide range of problems of situational modelling. In a course experiments are proved that the increase in quantity of the used indicators considerably improves quality of work of models. The system of models has convenient and intuitively-clear the interface and can be applied to the decision of various problems of situational modelling of a sanitary condition of the Sevastopol bays, the Kerchensky channel and coastal water areas of Black sea. Use area. Fundamental and applied results of research work can be used in research and project institutes of National academy of sciences of Ukraine, at branch institutes of modelling, recognition and forecasting of statistical and dynamic processes. In particular, at Institute of biology of the southern seas in scientific research institutes of a fish economy the Considerable part of work is already entered into educational process of the Kherson national technical university, National technical university "Kiev polytechnical institute". Product Description popup.authors Альомов Сергій Вікторович Бабічев Сергій Анатолійович Вороненко Марія Олександрівна Гожий Олександр Петрович Дідик Олексій Олександрович Литвиненко Володимир Іванович Миронов Олег Глєбович Рогальський Франц Борисович Шульга Марія Олександрівна popup.nrat_date 2020-04-02 Close