An estimation method of the fuel mass injected in large injections in Common-Rail diesel engines based on system identification using artificial neural network

•A fuel-injected mass estimation method with one pressure transducer is realized.•A virtual pressure sensor along a feeding pipe is built based on system identification.•Fluid properties along a feeding pipe are dependent on one set of downstream signal. An innovative method that is capable of evaluating the fuel mass of large injections (greater than 6 mg/shot) in real time is developed for passenger car Common-Rail injection systems, with the aim to reduce emissions and improve power performance. The proposed method requires only one additional pressure transducer installed on the rail-to-injector pipe, and its overall accuracy is within 1.4 mg. The feasibility of the estimation method is analyzed based on the characteristic curve theory, during the hydraulic injection phase, the fluid properties along the rail-to-injector pipe can be approximated based on a single signal, such as the pressure or the flow velocity, measured at the rail-to-injector tube downstream. Based on system identification using artificial neural network and physical equations, the pressure time distribution in the vicinity of the injector inlet is used as input to identify the pipe-passed fuel mass. This derived fuel quantities are verified to be well correlated with the measured injected mass in large injections under a wide range of working conditions. This estimation tool can be installed on a diesel engine with the relations implemented in the electronic control unit, allowing a more accurate evaluation of the injected fuel mass in real time.