In order to improve the driving safety and stability of trams passing through the S-shaped curve lines with small radius, the rational optimization of vehicle suspension parameters is proposed. Taking a tram line in Guangzhou as the engineering background, the dynamic model of vehicle-track coupling system was established by using the multi-body dynamics and finite element methods and the influence laws of vehicle suspension parameters on the operational evaluation indicators such as vehicle stability index, wheel-rail force, derailment coefficient, and wheel load reduction rate, were explored. By combining the optimal Latin hypercube sampling technique, BP neural network, and multi-objective genetic algorithm, the optimized suggestions for the tram suspension parameters were obtained. The results show that the operational evaluation indicators of the vehicle are significantly improved after the optimization. Compared with before optimization, the derailment coefficient of the vehicle has decreased by more than 10%. The research results provide theoretical support for the optimization design of suspension parameters for trams.