The GA-ANN expert system for mass-model classification of TSTO surrogates

A hybrid-heuristic machine learning methodology, based on hybrid genetic algorithm (GA) and artificial neural network (ANN) data classification methods is proposed as an expert system for assessing viability of surrogates of a two-stage-to-orbit (TSTO) vehicle. The methodology is integral to the inverse design method for spaceplane systems. Since spaceplanes do not exist therefore archival mass-model data is also non-existent and inverse design method is used to generate optimal vehicle configuration data. The GA-ANN offers an expert system whereby when a new vehicle configuration is evolved its mass-model is first optimized using GA and then the optimal solution is processed through the ANN classifier to assess the viability of solution. If classification result fails the process is repeated until a qualified result is obtained. Results are validated using mass-model parameters of HTSM (hypersonic transport system Munich) vehicles.