Genetic algorithms, their applications and models in nonlinear systems identification

The Genetic Algorithm was used to estimate the hydraulic compliance of the hydraulic system on the UBC teleoperated heavy duty excavator. Using real recorded and simulation data from the excavator, the Genetic Algorithm has successfully identified the compliance of single link and multi-link hydraulic system of the excavator. A Parallel GA ( PGA ) was implemented with 16 T800 Transputers. It achieved a speedup factor of 12 over a traditional GA. With such a high speedup factor, real-time monitoring of hydraulic compliance and other hydraulic parameters is becoming possible. New mechanisms such as the distributed fitness function, the active error analysis were used to enhance the performance of a PGA. A PGA which incorporated these mechanisms actually outperformed a traditional GA in key areas such as variance of the estimated parameter and parameter tracking ability. Finally, a physical model that explains the fundamental properties of GAs was introduced. The physical model ( a hypercube ) not only provides an excellent explanation of GAs searching power, but also gives insight to GAs users ways to improve and to predict the performance of GAs in most applications.Applied Science, Faculty ofElectrical and Computer Engineering, Department ofGraduat