The performance of the high-pressure (HP) compressor is very important for the two-stage turbocharging system. However, the performance of HP compressor on the engine most of the time is poor at low speed and low mass flow conditions. These will lead to poor engine performance at low engine speed. The purpose of this paper is to improve the performance of the HP compressor at low speed and low mass flow conditions. The Latin hypercube design of the experiment method is used to establish the Kriging model and global optimization by multi-objective genetic algorithm NSGA-II to optimal HP compressor. The simulation results indicated that the flow field within the compressor was improved and the high entropy generation area was reduced. The new design delayed the mixing between the tip clearance leakage vortex flow and main flow. The low-speed performance of the HP compressor was improved. The turbocharger gas stand tests and engine bench tests were carried out. The results showed that the efficiency and pressure ratio of the optimized design is increased by 2.1% and 3%, respectively. The engine achieved better performance in low-speed conditions. The pumping means effective pressure (PMEP) and intake airflow increased by 7% and 4.98%, respectively, while brake specific fuel consumption (BSFC) and soot emissions decreased by 0.56% and 32.8%, respectively.