Abstract
Traditional methods for electronic engine control systems rely on a large number of lookup tables. Map-based open-loop control systems inherently have no ability to compensate for the variations from mass production, aging, fuel properties, and environmental factors. However, an advantage of a closed-loop control system is that it can measure the results from the actual combustion and adapt to environmental factors. In particular, cylinder pressure-based control can maintain optimal engine performance by directly using the combustion information. Therefore, engine control based on the cylinder pressure has the potential to provide numerous functions such as pilot injection optimization, IMEP balancing, start of combustion (SOC) control, and maximum pressure control.
In this paper, SOC control achieved by utilizing in-cylinder pressure signals for a common rail direct injection (CRDI) diesel engine is described. The difference pressure (DP), which is the difference between the in-cylinder firing pressure and motoring pressure, directly reflects the combustion condition. Consequently, the crankshaft angle, where the DP has a value of 10 bar, namely CADP10, is selected as the variable for SOC detection. The feedback signal, CADP10, is used to design an adequate closed-loop controller for SOC control. Observation of the cylinder pressure signal from successive cycles showed substantial variation in the combustion process on a cycle-by-cycle basis. The controller with high gain may be excited by the high frequency components of the feedback signal. The gains of the feedback controller should be restricted to a certain level to avoid large increase in the cyclic variation. For these reasons, an adaptive feedforward controller is applied and the performance of the feedback controller is improved. The feedforward controller consists of the radial basis function network (RBFN) and the feedback error learning method was used for the training of the network. In this paper, the RBFN has two inputs which are engine speed and target SOC, and has one output, SOEFF. The feasibility and performance of the proposed controller were examined with the experiments with the transient engine operation.
Keywords:CRDI, diesel engine, SOC control, neural network, RBFN