Research On The Mean Value Model Of Two-stroke Diesel Engine Based On Improved Seiliger Cycle

With the development of computer capacity and simulation technology,there are many kinds of diesel engine models.At present,the models widely used in the study of diesel engine performance are the crankshaft angle model and CFD model.The simulation speed of these two models is slow,which cannot meet the requirements of large-scale systems such as the marine propulsion system.The mean value model based on the thermodynamic principle,such as mass conservation and energy conservation,not only guarantees the simulation speed of the model but also ensures the simulation accuracy of the model.As a sub-model of the large-scale complex system model,it can meet the needs of the system simulation,and moreover,observe the changes of important performance parameters of the diesel engine,and make a continuous evaluation.In this paper,an improved Seiliger cycle is proposed.The improved Seiliger cycle adds 2’ points,that is,the beginning of combustion.The stage before the combustion process is divided into two small stages,which can make the Seiliger cycle better represent the actual working process of the diesel engine.Based on the improved Seiliger cycle,the two stroke diesel engine process is divided into servel parts,and the Seiliger parameters are defined,and the Seiliger process model is built.Based on the experimental data,the crankshaft angle model of the diesel engine is built;Then the Seiliger fitting model is built,and the equations of the Seiliger process model and the diesel crankshaft angle model are established according to the equivalent standard,and the Seiliger parameter values are solved.According to the Seiliger process model principle and Seiliger parameter value,the mean value model of diesel engine is built in MATLAB/Simulink.When the diesel engine bench test can not carry out the full condition test for the diesel engine,a method of using the diesel engine neural network model to predict the non experimental conditions and verify the simulation accuracy of the mean value model under non experimental conditions is presented.The diesel engine neural network model is trained and built by using the cylinder pressure data collected from the diesel engine platform,and the accuracy is verified.The diesel engine neural network model is used to predict the cylinder pressure data under the non experimental conditions,and the simulation accuracy of the diesel engine mean value model is verified.The dynamic simulation model including diesel engine,load,governor and shafting rotation balance is built,and the changes of parameters,such as burst pressure and fuel injection quantity,in acceleration and deceleration conditions and load sudden change conditions are analyzed.Compared with the control volume model and neural network model of diesel engines,the Seiliger cycle based mean value model has the characteristics of fast simulation speed and high simulation accuracy of parameters corresponding to equivalent standards,which can be used as a sub-model in large-scale complex system simulation to meet the requirements of realtime simulation and its accuracy.