Parametric Optimization of Control Systems Pumping Station by the Genetic Algorithms

• Main Authors: M. P. Dounaev, N. N. Kucyi, N. D. Lukyanov
• Format: Article
• Language: Russian
• Published: MGTU im. N.È. Baumana 2014-01-01
• Series: Nauka i Obrazovanie
• Subjects: genetic algorithm; PID-regulators; hydraulic power pack; direct indicators of quality
• Online Access: http://technomag.edu.ru/jour/article/view/668

<p>Now the frequency regulation with its well-known advantages is becoming more popular for using in electric drives. One of these systems is the system of liquid flow stabilization for pumping station of the second flood, which is discussed at the article. There are functional elements of the system: the frequency converter, the induction motor, pump set, check valve, part of the pipeline between the flow sensors and PID controller. In this case are used PID controller with universal input and RS-485 OWEN TRM101.</p><p>The article tries to solve the main problem, which is to set optimal controller settings. The article presents a mathematical model of the system, built on the basis of transfer functions. The specific values of the parameters for Bratsk pumping station of the second flood are found.</p><p>To solve the problem of parametric optimization for control of pumping station authors use a genetic algorithm, which allows us to use different quality criteria. A set of parameters for the genetic algorithm is represented, as well as a method of selection and a method of the transition between generations. The problem of parametric optimization was solved by quadratic integral criterion and by the criterion, composed of direct measures of quality (regulation time and the overshoot). The PID controller optimal values of adjustable parameters were obtained both for considered criteria and for their corresponding transient processes. Comparative analysis of results was produced. When proving the optimality of the obtained settings for the quadratic criterion the proximity of the gradient components of the target functional to zero is demonstrated. In the case of a criterion based on a direct measure of quality, is presented a picture of convergence of adjustable parameters to the same value when run with different starting points.</p><p>It is presented that the use of a criterion based on a direct measure of quality, can significantly reduce the overshoot with a comparable regulation time.</p>