Methods for parameter sensitivity assessment in aircraft-missile models

This study was conducted within the project Technical Threat Systems Analysis (THSA). The purpose of this study has been to make an investigation about the use of Monte Carlo and optimisation based methods for sensitivity assessment in aircraft-missile models. A genetic optimisation algorithm has been developed and implemented in ACSL/Math and Matlab. Genetic optimisation algorithms use only function evaluations and no derivative information. Monte Carlo analysis has also been applied on an air to air missile model. We have shown that genetic optimisation methods are promising methods in _nding extremum values in sensitivity analysis studies. Our genetic algorithm implementation performs well on the test examples as compared to the implementations of derivative free optimisation software codes ASA and GAOT.We have also shown that Monte Carlo methods are useful tools for evaluation of complex models such as used in THSA. For the parameter set of the example used, we see clear tendencies in model parameter dependency, e.g. linear dominant behaviour with respect to part of the parameter set.