Analysis and Optimization of the Idle Sound Quality Based on RBF Prediction Model 2021-01-0660

Idle sound quality of four-cylinder DI gasoline engine is predicted and optimized. The objective parameters of psychoacoustics such as loudness, roughness, sharpness, tonality, fluctuation, and interference are selected for objective evaluation. The numerical valuation method is used to evaluate the 21 engine noise samples at the idle condition. The Kriging, RBF, and RSM prediction methods are respectively used to establish the subjective and objective regression model. The prediction results are compared with the actual subjective evaluation results and show that the RBF method has the highest consistency with the correlation coefficient is 0.912. The sensitivity model of objective evaluation parameters of sound quality based on RBF is established. The contribution of objective evaluation parameters that affect the idling sound quality is calculated quantitatively. The contribution of sharpness to the sound quality of the gasoline engine is the largest (39.6%), followed by loudness, which is 18.2%. Through the dynamic characteristics analysis of the front-end accessory system (FEDA), the common rail pressure simulation of the high-pressure fuel injection system, the modal analysis of the structure, and the transfer path analysis (TPA), then the combinatorial optimization measures are put forward. According to the RBF regression model prediction results, the subjective evaluation result for this engine after optimization is significantly higher than the original engine.