Hydraulic dynamics in split fuel injection on a common rail system and their artificial neural network prediction

Abstract Injection dynamics in engine common rail systems influence cycle fuel injection rate and mass, and further the combustion and emissions features. We therefore experimentally studied the hydraulic dynamic behaviors on a common rail injection system under changed two-stage injection strategies and then constructed an artificial neural network to predict these hydraulic dynamic behaviors. The injector inlet pressure dynamics were firstly measured with changed injection pressure and then associated with the test conditions in both time and frequency domain. Further, an artificial neural network model was constructed and trained to predict the hydraulic dynamic features at different test conditions, with the trial-and-error method used to identify an appropriate neural network configuration. The sensitivity of the injection hydraulic dynamics to the injection parameters was also evaluated. It is found that the injection pressure places a more significant influence on the pressure fluctuation amplitudes than injection dwell time and pilot injection energizing time, particularly on the minimum injector inlet pressures. With elevated injection pressure, an increase in pilot injection energizing time causes an increased fluctuation amplitude under most injection dwell times. The time intervals between the maximum/minimum injector inlet pressures during the pilot and main injections increase with injection dwell time and decrease with extended pilot injection energizing time. The pressure fluctuation amplitudes in the frequency domain are sensitive to the injection dwell times. The constructed back propagation neural network with an optimized configuration shows good prediction capability, and the sensitivity analysis results on injector inlet pressure are consistent with those obtained experimentally.