Abstract: Abstract: In view of the problems of traditional diesel engine modeling, such as low modeling efficiency and versatility, poor openness of software platform, and abstract modeling method, this paper used Modelica language to set up a simulation model of diesel engine, expecting to be able to solve the above problems and ensure the accuracy of the dynamic and steady state simulation of the diesel engine. Firstly, according to the actual physical structure of diesel engine and object-oriented modeling method of Modelica language, a diesel engine model was divided into several sub-models, such as combustion chamber sub-model, air valve sub-model, and turbocharger sub-model. Secondly, for the sake of delivering the information between different sub-models, this paper defined some interface types according to the actual function of each sub-model. There were 3 kinds of transfer variables between these interfaces, i.e. flow variable, potential variable and real number variable. For example, the turbocharger had a rotating interface and two gas interfaces; the flow and potential variables delivered in rotating interface were the torque and the angle of rotation, respectively, and the flow and potential variables delivered in gas interfaces were the mass flow and enthalpy flux, the temperature and pressure of gas, respectively. Thirdly, the equations describing the internal behavior of sub-models and the multi-domain unified physical modeling specification Modelica language were used to set up the sub-models mentioned above, respectively. Compared to other modeling methods of diesel engine, Modelica has some outstanding characteristics, such as object-oriented modeling, non-causal modeling, exhaust temperature before turbo; the complex physical structure of a diesel engine can be restored by the combination way of the composition and topology structure of the model, and the universality and inheritance of the model are better. As Modelica has the non-causal modeling feature, the modelers need not to consider the causal relationship between the various variables and only need to pay attention to how to use the equations to describe the internal behavior of the model, and therefore the modeling efficiency is improved. These are especially important for the modeling of diesel engine. Finally, taking the D4114B type electricity-generating diesel engine as an example, the simulation model was established by the connection of the interfaces between the sub-models, and the modification of the structure parameters and boundary parameters. The steady state performance of D4114B type diesel engine was simulated under the rated speed with the simulation model, and under 25%, 45%, 70%, 90% and 100% load condition, the simulation values of several parameters were compared with the experimental values, such as explosion pressure and hourly oil consumption. The maximum of the relative error was 8.9%, which showed that the established model could describe the steady state process of diesel precisely. On the basis of steady simulation, the dynamic simulation of the model was realized. The change rules of some transient performance parameters of diesel engine were analyzed when the load was suddenly changed at the rated speed, such as rotational speed, fuel quantity per cycle for single cylinder, and output torque. In order to respond to the changing process of dynamic simulation precisely, some discrete points were chosen from these simulation curves, and the corresponding data of discrete points were listed. The simulation value of rotation speed of diesel engine was compared with the experimental value, and the two were in good agreement, so the model could forecast the dynamic performance of diesel engine well. This paper can provide the reference for the following study of diesel engine based on the Modelica language.