In order to reduce automobile exhaust emissions, proposes the control parameters optimization of parallel hybrid electric vehicle (PHEV) by means of multi-level scanning parameters optimization algorithm. The optimized paratmeters are simulated through ADVISOR software, and the simulated results and the resutls before the parameters optimaztion are compared. Under the CYC_UDDS condition, the simulated results show that HC is 0.520 g/mile，CO is 2.361 g/mile and NOx is 0.407 g/mile in automobile exhaust emissions, which are obviously low compared with the the resutls before the parameters optimaztion. The superiority of multi-level scanning parameters optimization algorithm is proved.