Snytyuk O. Synthetic models and methods of agricultural production structure optimisation

The thesis is devoted to a problem of agricultural production optimisation in conditions of a priori information boundedness and indeterminacy. In this paper the analysis of modern methods of structural and parametrical identification of production function and methods of its optimisation is carried out. The formalisation of the task a determination of a production structure that will answer it a maximum efficiency is suggested. Is established that the composition of models artificial neural nets and genetic algorithm enables solutions of the task of identification and consequent optimisation of production function. The procedure of magnification of selfdescriptiveness of initial datas with using of a box-counting technologies and principal components method is offered. Forecasting quality of an optimised neural nets models complex: feedforward nets with back propagation, counterpropagation networks and radial-basis nets is investigated. The process technology of consulting services to the agricultural producers is developed: from gathering an information and forecasting up to the practical recommendations for definition of sowing squares and expenditures optimum volumes.