Multi-Objective Optimization Study of Dual-Fuel Engine Emissions and Economy

Abstract

In the multi-objective performance optimization process for dual-fuel engines, the conflict between emissions and fuel efficiency is a consideration. To find a satisfactory compromise among many optimal solutions, reduce the difficulty of selecting solutions, and respond to different needs, a decision-making preference optimization strategy is introduced. The reliability prediction algorithm of the micro-ignition diesel/natural gas (NG) dual-fuel engine is built using support vector machines (SVM). The performance prediction model is combined with the optimization algorithm, and the preference information of the decision-maker (DM) is introduced into the optimization process, to guide the population evolution process to the direction that the DM is interested in, and achieve multi-objective preference optimization. Selecting nitrogen oxide (NOx) emission and braking specific fuel consumption rate (BSFC) as the optimization targets, the optimal Pareto front surface is obtained. It can be seen from the simulation results that after introducing the decision preference, the evolution of the population can proceed in the direction of interest to the DM. Preference optimization can be achieved by rationally configuring the preference strength parameter δ, the weight vector w, and the reference point g. The combination of control parameters corresponding to the two preferences was downloaded to the ECU for bench test, and compared with the original data, it was found that when low emissions are preferred, the NOx emission meets the IMO Tier-III limit under all working conditions, and the average NOx emission is 1.22g·(kW·h)-1, which is 78.9% lower than the original engine. At the same time, it was found that even with lower emissions, fuel consumption was reduced by 4.94% compared to the original engine. The preference for lower fuel consumption is 3.68% lower than the preference for lower emissions, but the deterioration of NOx emissions is obvious.