Genetic algorithm usage for optimization of saturator operation
Ukrainian Food Journal
Introduction. It is carried out the research of the optimal control adaptive system of the apparatus of II saturation. Qualitative indexes of the adaptive control system efficiency are defined. Materials and methods. Adaptive control system of a sugar refinery apparatus of II saturation was studied. Simulation modeling based on classical and hybrid genetic algorithms is used to determine the optimal performance parameters. Results and discussion. The simulation studies of the quality of
... quality of functioning of the structural model of the adaptive system of optimal control using the classical genetic algorithm are carried out, as well as the research of the modified genetic algorithm with the addition in the classical algorithm the hybrid functions, namely fmincon, fminsearch patternsearch fminunc. Adaptive system of optimal control has significantly lower integral quadratic criterion І = 545 comparing to the existing one, which integral quadratic criterion I = 658. Moreover, control time is also reduced. Adaptive control system requires T = 109 s, while existing control system requires T = 212 s. Usage of hybrid functions allowed to additionally reduce integral quadratic criterion to І = 529-541 and to speed up the system, required time T = 98-105 s. Also, a study of the fmincon method without the use of a genetic algorithm was carried out, this model showed a lower execution time T = 88 s, but the value of the integral quadratic criterion I = 604 was higher. The best results in terms of integral quadratic criterion and time (I = 529, T = 98 s) were obtained for genetic algorithm combined with fmincon hybrid function. Developed adaptive control system for saturator operation significantly outstands the existing one by all the main indexes. That is why it is highly recommended to replace control system on the sugar refinery with the studied adaptive one. Conclusions. The novelty of the research results is the scientific substantiation of the feasibility of using classical genetic and hybrid genetic algorithms in the implementation of adaptive optimal control systems.