Title :
Multi-objective evolutionary algorithm using population diversity
Author :
Weng Li-guo ; An Wang ; Min Xia ; Zhuangzhuang Ji
Author_Institution :
Inf. & Control Coll., Univ. of Inf. Sci. & Technol., Nanjing, China
Abstract :
Crossover and mutation plays an important role in evolutionary algorithms, probability selection determines the performance of the algorithm. Now the crossover and mutation probability is mainly calculated according to the fitness of individuals, while some shortcomings still exist, such as evolution easier to stall and so on. In the paper, we propose an adaptive adjustment strategy which genetic parameters change based on diversity of population, ensuring maintaining sufficient diversity and enhancing search capability of the algorithm during the evolution. Compared with AGA through the classic test functions and the robot path planning. Simulation results suggest the improved strategy has greatly improvement on the ability of fast convergence and stability of the algorithm, and more easy to jump out the local convergence.
Keywords :
convergence; evolutionary computation; mobile robots; path planning; probability; search problems; adaptive adjustment strategy; algorithm stability; crossover probability; local convergence; multiobjective evolutionary algorithm; mutation probability; population diversity; probability selection; robot path planning; search capability; test functions; Automation; Convergence; Evolutionary computation; Instrumentation and measurement; Robots; Sociology; Statistics; MOEA; crossover and mutation; population diversity; robot path planning;
Conference_Titel :
Instrumentation and Measurement, Sensor Network and Automation (IMSNA), 2013 2nd International Symposium on
Conference_Location :
Toronto, ON
DOI :
10.1109/IMSNA.2013.6743448