Variable Convergence in Evolutionary Optimization and its Relationship to Sensitivity Analysis

Citation

Wright JA, Wang M, Brownlee A & Buswell R (2012) Variable Convergence in Evolutionary Optimization and its Relationship to Sensitivity Analysis. In: Wright J & Cook M (eds.) Proceedings of the 2012 Building Simulation and Optimization Conference. First Building Simulation and Optimization Conference, Loughborough, 10.09.2012-11.09.2012. Loughborough: Loughborough University, pp. 102-109. http://www.bso12.org/-proceedings/papers/2A2.pdf

Abstract
Optimization methods provide building designers with optimized design solutions, whereas sensitivity analysis methods indicate the extent to which the design variables have an impact on building performance, this information being important when uncertainty exists in the design decisions. This paper investigates the extent to which the solutions found from a building design optimization can be used for a sensitivity analysis. It is concluded that the solutions found at the start of the optimization process can be used in a global sensitivity analysis. It is also concluded that the convergence behaviour of the search does not relate to the global sensitivity.