Waveriders are supersonic or hypersonic lifting configurations. They areextensively utilized as the forebody part of hypersonic vehicles. As thecore component to generate the lift and compress the incoming flow, awaverider should be designed for assuring the high performance of a vehicle.Various optimization works had been carried out to improve the aerodynamicperformance. However, most of the optimization procedures are often timeconsuming and unstable when the computational fluid dynamic (CFD) analysisis employed for directly evaluating aerodynamic performance. To aim at thisproblem, an artificial neural networks (ANN) based response surface methodwas proposed. First of all, a number of waverider shapes are chose as thenet-training samples, and the aerodynamic performance of each sample isevaluated by CFD analysis. Next, with respect to the training couple, thecontrol parameters of each waverider and its aerodynamic coefficients areprovided to a pre-constructed ANN. The weight of each connection in the ANNis adjusted until the error between the ANN output and the CFD result areacceptable for every training couple. Finally, the ANN is embedded in theoptimization loop as the response surface of the time consuming CFDprocedure. Two numerical cases in the design point of Mach 6 and Reynoldsnumber 7$\times$10$^6$ are carried out to validate the presented method, asingle-objective optimization for maximize the lift-to-drag ratio ($L/D$), anda multi-objective problem to improve the integrated performance of awaverider with the maximal $L/D$, the maximal cubage, and the minimal wetarea. The numerical results show that the ANN based response surface methodis stable with lower time consuming.