Some references about Genetic Algorithm and Evolutionary computation

Identification of multivariable models of fast ferries

J. Aranda, J. M. de la Cruz, J. M. Díaz

EUROPEAN JOURNAL OF CONTROL. Vol 10, n.2, May 2004, pp. 187-198. 

This work presents the formal approach for identifying continuous transfer functions of the vertical dynamics of a high-speed ship as a non-linear optimisation problem with linear restrictions. The proposed solution is described with a hybrid optimisation method (genetic algorithm + classic non-linear optimisation algorithm with restrictions).

Parallel Evolutionary Computation: Application of an EA to Controller Design

M. Parrilla, J. Aranda, and S. Dormido-Canto

J. Mira and J.R. ´Alvarez (Eds.): IWINAC 2005, LNCS 3562, pp. 153–162, 2005.

The evolutionary algorithms can be considered as a powerful and interesting techniques for solving large kids of control problems. However, the great disadvantage of the evolutionary algorithms is the great computational cost. So, the objective of this work is the parallel processing of evolutionary algorithms on a general-purpose architecture (cluster of workstations), programmed with a simple and very well-know technique such as message passing. ( more )

Selection and tuning of controllers, by evolutionary algorithms: application to fast ferries control

M. Parrilla Sánchez, J. Aranda Almansa and J.M. Díaz Martínez

CAMS 2004 IFAC Control Applications in Marine Systems. Ancona July 7-9. Pp. 351-356.
Control Applications in Marine Systems 2004: A Proceedings Volume From the IFAC Conference, Ancoma, Italy, 7-9 July 2004 (Ipv - Ifac Proceedings Volume)
by R. Katebi (Editor), S. Longhi (Editor)

Evolutionary algorithms have been shown very efficient tuning controllers, whose structure has been previously established. In this work, a step forward in the automation of the controllers design process will be tried, an algorithm is implemented which is able to select the appropriate controller structure and to tune it. By means of this procedure, a controller to reduce the motion sickness incidence on a high-speed ship, will be designed. The algorithm will be implemented using parallelization techniques.

A real application example of a control structure selection by means of a multiobjective genetic algorithm

M. Parrilla Sánchez and J. Aranda Almansa

IWANN 2003, Vol. LNCS 3562, pp. 153-162, 2003

Control problems are clear examples of multiobjective optimization. In this kind of problems a series of objectives, some of them opposed to each other, will be optimized in order to fit some design specifications.

Moreover, evolutionary algorithms have been shown to be ideal for the resolution of these kinds of problems because they work simultaneously with a set of possible solutions, thereby favoring convergence towards a global optimum. In this document we propose a way of dealing with the different objectives considered and a genetic-evolutionary algorithm that will enable some phases of the controller design to be automated.

Finally, an application example of the methods outlined will be applied to the design of a controller to reduce the sickness index on a high-speed ship. ( more )

Evolutionary algorithm for the design of a multivariable control for an aircraft flight control

J. Aranda, J.M. de la Cruz, M. parrilla, P. Ruiperez

AIAA Guidance Navigation and Control Conference and Exhibit. 14-17 August 2000. AIAA-2000-4556 

Aircraft flight control design is a multivariable control problem with multiple sensors and multiple actuators where various strict requirements from multiple disciplines have to be satisfied. In this paper a method based in Evolutionary Algorithm is used for the design of these multivariable controllers. This study illustrates how a technique such as the multiobjective optimization by evolutionary algorithm can be applied to multivariable control design. Several objective functions can be considered such as the designer would state them. The problem is formulated from the point of view of the designer, not that of the optimizer. The method is applied to the design of an aircraft flight control and their results can be compared to results of other methods. These techniques can be applied with different controller structures and the objectives can be obtained in a way optimum.

Design of a linear quadratic optimal control for aircraft flight control by genetic algorithm

J. Aranda, J.M. de la Cruz, M. Parrilla, P. Ruipérez

Controlo’2000: 4th Portuguese Conference on Automatic Control. Guimarães-Portugal. October 4 - 6, 2000

Aircraft flight control design is a multivariable control problem with multiple sensors and multiple actuators where various strict requirements from multiple disciplines have to be satisfied. In this paper a method based in Evolutionary Algorithm is used for the design of these multivariable controllers. This study illustrates how a technique such as the multiobjective optimization by evolutionary algorithm can be applied to multivariable control design. Several objective functions can be considered such as the designer would state them. The problem is formulated from the point of view of the designer, not that of the optimizer. The method is applied to the design of an aircraft flight control and their results can be compared to results of other methods. These techniques can be applied with different controller structures and the objectives can be obtained in a way optimum. (more )