Title :
A New Unconstrained Optimization Method based on Feedback Control System
Author :
Lu, Baiquan ; Zheng, Zhou Jian ; Jie, Yuan Min
Author_Institution :
Shanghai Key Lab., Shanghai
fDate :
May 30 2007-June 1 2007
Abstract :
In this paper, a control system with optimization calculation ability is presented, in which neural networks is used as the controller and the cost function is used as the control plant. Thus some characters of the control system, such as perturbation or chaos in this system, can help us to find the global optimization solution since usual BP control algorithm is very easy to fall into a local minimum point. When a perturbation is added at input side of the controller or there exists chaos in this system, it is possible that its trajectory escapes from a local minimum point to other places for finding global minimum point; in the way until the global minimum point is got. Carrying on some simulations, as a result, it indicates that the optimization method based on the control system is useful for finding global optimization solution of unconstrained optimization problems.
Keywords :
backpropagation; chaos; control systems; feedback; mathematics computing; neurocontrollers; optimisation; perturbation techniques; BP control algorithm; feedback control system; neural network; new unconstrained optimization method; Ant colony optimization; Automatic control; Automation; Chaos; Control systems; Cost function; Feedback control; Neural networks; Optimization methods; Simulated annealing; Chaos; Control system; Global Optimization; Neural Networks; Perturbation; component;
Conference_Titel :
Control and Automation, 2007. ICCA 2007. IEEE International Conference on
Conference_Location :
Guangzhou
Print_ISBN :
978-1-4244-0818-4
Electronic_ISBN :
978-1-4244-0818-4
DOI :
10.1109/ICCA.2007.4376807
