PNRPU Bulletin. Electrotechnics, Informational Technologies, Control Systems

In this article application of genetic algorithm for optimization of a control system by gas-turbine units as a part of the software modeling suite KMES is shined. Genetic algorithm - the latest, but not the unique way of the solution of optimization problems. For a long time two main solutions of such tasks - direct-search and local-gradient are known. At these methods the merits and demerits and in each case it is necessary to think what to choose from them. The typical practical task, as a rule, multimodal and multi-dimension, that is contains many parameters. For such tasks there is no universal method which would allow to find absolutely exact solution quickly enough. However, combining direct-searc and gradient methods, it is possible to hope to receive at least the approximate decision which accuracy will increase at increase in time of calculation. Keywords: control system, genetic algorithm, chromosome, gen, population, crossover, mutation, fitness function, selection.

Article deals with two-dimensional orthogonal cutting-packing optimization which is the most often in practice. 1.5-DBPP (1.5-Dimensional Bin Packing Problem) and 2-DBPP (2-Dimensional Bin Packing Problem) were considered. Difference between these two problems is following. In 1.5-DBPP we are given half-infinite strip of material while in 2-DBPP sheet of material has fixed length. Algorithm-decoder was proposed, this algorithm was called group decoder. To find group of objects which should fill current block proposed algorithm uses linear cutting methods based on dynamic programming. Choose of precisely group but not single object let us more effectively spend space on the sheet and reduce the amount of waste. Computational experiments were held in which proposed algorithm was compared to existing ones; use of simulated annealing methods to find better priority lists was also considered. In these experiments group decoder applied together with simulated annealing showed the best results. Moreover results group decoder without simulated annealing weren't worse than existing algorithms with simulated annealing. Experiments were hold on two data sets. First data set was test consisted of randomly generated rectangles. For second data set tests were generated so as to one be able to place rectangles on a sheet in a way to sheet was used for 100 %. On several tests from second data set proposed algorithm was able to find optimal solution.

The induction motor with squirrel-cage rotor was studied. The effect of voltage falling studied with various type of torque resistance. The experiments were performed on a mathematical model of the induction motor in the programming environment Simulink. Changing rotation speed was calculated for different values of time and depth falling voltage. There are presented conclusions.

Application of ethylene propylene rubber in cable and wire products is considered. Prospects for use of modern crosslinkable materials as cable insulation is considered. Evaluation scope for rubber insulation and protective cable sheath is given, shows the properties of rubber. Comparison with other types of insulating materials on the example of cross-linked polyethylene insulation and paper is produced, the advantages and the disadvantages in comparison with these types of insulation is indicated. Recommendations for selection of equipment for the production of products with ethylene propylene rubber: the forming tool and pressing aggregate, curing environment are given. Approximate modes of processing this type of material are specified. Recommendations for compaction and curing conditions are given. Features of vulcanization technological process are considered: temperature conditions, related equipment for ensure and control the quality of the product.

The estimation of thermal modeling at the design stage by iteration. For equipment functional-hub concept design acceptable phased ideology of modeling of thermal modes of constructions RES, adequate block hierarchy design and with the downward direction in which structural system that can be applied this method of modeling is divided into the following isothermal elements: elements of BOC, cooling systems, etc. on the same structural level (body shell), modules underlying level (discrete body). This should be known external heat - temperature environments or external surfaces of adjacent modules of this and the above levels that consider the design is in heat transfer, the value of which is determined on the previous stages of thermal calculation. Underestimation of the method iteration on stage thermal design leads to necessity of the decision of errors at the stage of manufacturing technological methods.

The internal and external interferences causing changes in the characteristics of automatic control systems of aircraft gas-turbine engines (ACS GTE) are analyzed. Solution of the adaptation problem and monitoring ACS GTE parameters’ is based on the application of identification methods. The purpose of this work is the development of multi-dimensional Kalman filter connected to the output of the built-in mathematical dynamic model of the gas-turbine engine to improve the parameters’ identification and to achieve high-quality automatic control. The possibility of using the Kalman filter for this class of dynamic processes was proved by statistical analysis of experimental statistics method, based on Slutsky criterion. The basic mathematics, which are underlying the algorithms of optimal multidimensional filtering, were considered. The approbation of the proposed algorithms was made by MatLab. The simulation results showed that the using of multi-dimensional matrix Kalman filters in ACS GTE model allows to achieve higher rates parameters’ identification accuracy than analogues used in current technology practice.