Optimization design of a fully variable valve system based on Nelder-Mead algorithm

Abstract

This paper presents an optimal design for the soft-landing of the fully variable valve system in internal combustion engines based on Nelder-Mead algorithm and integrated simulation. The system is used to enhance the engine performance in terms of emissions and fuel economy. The electromagnetic linear actuator (EMLA) and the magnetorheological (MR) buffer are modelled with COMSOL Multiphysics, and a co-simulation platform is constructed with MATLAB/Simulink. The prototype test results are basically consistent with the integrated simulation, indicating the feasibility of the integrated simulation platform. After that, the structure of the magnetorheological buffer is optimized based on the Nelder-Mead algorithm. The optimization objective is the adjustable coefficient and the optimized parameters are the effective length, the annular gap width, the piston diameter, the groove depth and length of the coil. The adjustable coefficient of the optimized MR buffer was increased from 8.8 to 25.46. And the co-simulation results showed that the optimized system has a great improvement in soft-landing performance, with the valve seat velocity reduced from 0.51 m/s to 0.03 m/s and the bounce height reduced from 0.6 mm to nearly 0 when the lift is 8 mm.