Genetic Algorithm Based Software for Optimization and Design of Piles on Slopes (original) (raw)
International Conference on Civil Engineering in the New Millennium: Opportunities and Challenges (CENeM-2007), Bengal Engineering and Science University, Shibpur, Howrah, WB, India , 2007
The present paper deals with genetic algorithms (GA) to find out the minimum Factor of Safety (FOS) of a virgin slope vis-à-vis stone column reinforced slope and highlights some potent points that emerged during the search process. GA-a new generation and nontraditional search and optimization procedure is fast becoming popular for its increased versatility in handling complex engineering problems, since it does not require problem specific knowledge for carrying out the search. It is a tool guided by stochastic principles instead of gradients and hence one has the liberty to start with a wide range of values (or a ‘population’ of points) for the decision variables to process the value convergence of the objective function to near optimal solution at a very rapid rate. The output is found to be independent of the value of the initial vector. The slope stability analysis is carried out on the computer using GA as a function optimizer in slope analysis. The present paper is a sequel to author’s previous study (2001, 2003a, 2003b) wherein the raw data of a virgin slope example cited by Spencer (1967) is used as input, and the output was obtained by integrating GA with Bishop's simplified method of slices (1960) for virgin slope and Author’s (2001) method for stone-column reinforced slope. The search process, besides unfolding increase in FOS of stone-column reinforced slope over virgin slope against failure on a cylindrical slip surface, unleashed some other powerful features that are described in the paper.