Orlovsky I. A synthesis of automation electromechanics systems on the basis of the modified recurrent neuron networks

Українська версія

Thesis for the degree of Doctor of Science (DSc)

State registration number

0512U000226

Applicant for

Specialization

  • 05.09.03 - Електротехнічні комплекси та системи

15-03-2012

Specialized Academic Board

Д64.050.04

Essay

Object of research - transients in electromechanical systems (EМS) with intellectual control systems. The purpose of research - creation of methods of synthesis automated EМS on the basis of the modified recurrent neural networks (МRNN) for maintenance of effective parameters of management. Methods of research - methods of theories of automatic control, the electric drive, identification, artificial neural networks, indistinct sets, optimization (in particular a method of genetic algorithm), matrixes, разностных the differential equations, mathematical modeling. Theoretical and practical results - has received the further development the theory of the electric drive regarding synthesis intellectual EМS managements on the basis of offered МRNN; structures МRNN for identification linear and nonlinear EМS in view of their features are offered; are found by the set criteria parameters of a control system asynchronous traction to the electric drive of a diesel engine-train with use of its models of a kind of the MRNN and genetic algorithms; methods of synthesis are developed are checked up by mathematical modeling, at the research stand and at operation asynchronous traction to the electric drive of a diesel engine-train. Novelty - methods are developed: synthesis for nonlinear EМS mathematical models of kind МRNN with use of training of gradient algorithms on parameters of a mode of object; Transformations of linear and nonlinear mathematical models EМS to models of kind МRNN; synthesis of weight factors МRNN by analytical parities{ratio} on parameters of an operating mode of object and a technique: identifications of parameters EМS on their models of kind МRNN; decrease{reductions} in time of search at use of genetic algorithms of parameters of regulators on model EМS of kind МRNN due to use of the information about EМS; the mathematical model of teamwork of all units of the asynchronous traction electric drive a diesel engine of a train in view of the advanced management of a diesel engine, the generator and traction asynchronous engines is developed; features of synthesis of mathematical models of kind МRNN for EМS with to non-linearity type a backlash and restriction on parameters of an operating mode of the equipment are certain. Degree of introduction - in Open Society of scientific research institute " Converter " (Zaporozhye); in the motor-carload of depot Hristinovka of the Odessa railway; educational process of the Zaporozhye national technical university. Area - an iron and steel industry and transport.

Files

Similar theses