| To meet the increased regulatory requirements on emission and fuel economy, the electronic control units (ECU) for engine is becoming more and more complex, which also makes its exploitation more difficult. The traditional method of debugging ECU through bench tests is almost outdated for its high cost, long exploitative period, unstable environment, etc. The substitute is found to be hardware-in-loop simulation. As an up-to-date applied Tech, ECU hardware-in-loop simulation system (ECU HILSS) can be regarded as a tool with integrated software and hardware, which is adaptive for real-time simulation. During debugging, the HILSS is connected to ECU directly as a virtual engine, receiving the control from the ECU and sending out the homologous running message of the engine, which reduce the tests and cost of all kinds. Besides, operators are provided with a convenient way of conditioning and observing the whole system because of its software controlled mechanism of imparting information. Flexibility, reliability and safety are all enhanced thereby. To develop such an ideal platform for study of the ECU, ECU HILSS is designed and discussed in detail in this paper.The success of ECU HILSS could only be obtained from the high performance of hardware and software. As a sophisticated real-time running system with character of being real-time and tightly engaged, the feasibility of its actualization is studied at first through the aspects of logic and physical enforcement. Based on the demonstration in the project target and the technologic support, the HILSS is completely constructed, which is a tightly coupled multi-processor system composed of a standard personal computer, a high-performance single chip microprocessor system and a fast-running floating point DSP system. The debugging of the outside ECU will become easier by the friendly graphical user interface, and the high-speed signal transfer through all the parts. Besides, the HILSS can be expanded conveniently for its modular components.As the core of the HILSS, the dynamic models affect the workability and the reliability of the whole system. All the objects are modulated by the principle of independence and physical integrality, and then modeled one by one. Compared with other processes, the combustion model is of more importance for the accuracy and the time cost of the system, so more emphases are put on it. After an all-round analysis of the contribution of the combustion models to the HILSS, the Vibe model is used, and the modified Ouler algorithm is chosen.In consideration of the changes of the models to be caused by the future system expansion, it's helpful that analyzing the scheme for modifying the models with some simple tools before the application. MATLAB/SIMULINK is a kind of dynamic simulation software, which can be used to speed up the building of the HILSS. In this paper, it is used as a prejudging tool to test thepossibility of some simplification in the models. Based on the analysis of the use of SIMULINK, a practical method for simplifying the real-time models is proposed.To perfect the use of whole system, neural network and reconstruction of the pressure are introduced into the system as complements. The principle of the neural network models is different from that of the real-time running models. The latter is designed on the engine running, so the effect of the control strategy on engine performance can be observed clearly. While the former works out the solution directly from the testing data, avoiding the complex calculation procedure. So the complementarity between the two models can be used to make a more comprehensive assessment of the ECU. On the bases of some basic analyses of the neural network models, an fuel injection model (emphasized on the injection quantity) made through neural network is designed and studied all-round, and some other application of neural network models in HILSS are also analyzed. Another expansion for the system is the realization of the on-line probation of the models through the recons...
|
PDF Full Text Request