Liachenko S. Automation of technological branches control sugar production based on neural network approach

The object of research is the process of automated process control sugar production. The thesis is devoted to important scientific and technical problem of automation control processes offices sugar production by synthesis of adaptive control systems based on the use of models and methods for adaptive and predictive control, given the nature of uncertainties class objects in question. First proposed: a method of synthesis models of the technological process of the sugar production based on adaptive approach taking into account the non-stationary processes, a method of constructing neural network models of the process based on static neural networks of direct distribution method of synthesis of neural and neural predictive PID-regulators, neural network model predicting the technological process based on dynamic perceptron. Improved recursive method of constructing unsteady regression model and the organization of information and computer software simulator. Further developed: the method of control of dynamic objects through their static models and assess emerging obtained with the loss of traditional neural network methods for constructing nonlinear models Wiener and Hammerstein, which improves the interpretation model, adaptive model of non-stationary processes sugar production using recurrence algorithms with high speed of convergence, which reduces the time of construction of mathematical models of objects. Developed in theses models and methods allow for new design decisions on process automation process control sugar production.