Research On Vehicle Emission Simulation Based On Neural Networks

Recently, some countries have introduced policies on traffic control and carbon emission due to the deterioration of the environment caused by automobile emission and the standard of emission limitation is rising gradually.In order to control the automobile emission, the rule self-adjusting process is adopted the main measuring method while CVS plays a subsidiary role. But the improvement on automobile emission is primarily focused on the internal combustion engine while the testing and enhancement are normally achieved through the engine test-bed experiment. This may unavoidably cause the trouble of dismounting repeatedly. In the process of engine test-bed experiment, there would be a multitude of experimental operating point. Thus lots of time and energy might be wasted on the test-bed experiment. It may seem workable to apply the mathematical modeling to calculate the emissions of internal combustion engine precisely but actually the factors which may have an influence on the emission process are various and complicated. So this method would not be practical and easy to carry out. Given all the potential issues and advantages of the normal solutions,this text will introduce and define a new approach which would only need the basic physical calculation and it’s easier to calculate the mode-test emission more accurately. The main work of this text would be:1. Analyzing the processes and result of the internal combustion engine test-bed experiment precisely and using the test data to establish related GT-power models via GT-power software.2. Introducing a model which uses BP neural net to stimulate the experiment of internal combustion engine test-bed after an impeccable research on neural net theories and internal combustion engine emission theories. Then we could use heredity and variation to enhance the fault tolerance and accuracy about BP neural net.3. Conducting researches about force distribution on the bodywork and engine during 10-15 mode-test standard testing circle and establish the mathematical model on internal combustion engine working parameters.4. Calculating the emission under mode-test through neural net.5. This article is based on the generalization ability of artificial neural net and calculates the number of internal combustion engine emission by establishing the input end based on Genetic Algorithm’s torque and outgoing end based on the emission and fuel consumption data.And also,this article apply biogenetics arithmetic to update the traditional BP neural net and enhance the generalization ability of it.Then apply the net to the complete vehicle mode-test and stimulate the emission process under 10-15 mode-test.In conclusion, this method provides a new approach of researching the automobile emission. To some extent, we could use this way to simplify the work on vehicle research, and save the energy and cost at the same time. Specifically, it has certain theoretical significance and application value of engineering.