Multi-factor predication of diesel engine by using artificial neural networks

The paper presents an algorithm which combining a neural network observer, it give more flexible and accurate control on the engine operation. In recent year, several researchers already proposed the algorithm which is based on a model-based adaptive control and uses artificial neural networks, and they assumed engine system is also controled by severe non-linear dynamics and their modelling. Especially, the neural network model observer is related with effective torque and brake torque. And they also proposed the model predictive control strategy which uses information multivariables and considers engine dynamics to do multi-step ahead prediction, and is adapted in on-line mode to cope with system uncertainty and time varying effects. Even though they already proposed this kind of nonlinear model method, they never think about combining all non-linear factors together. In this paper, we proposed a well-trained artificial neural networks which used as a predictive model for a combination of non-linear factors. Finnally, experiments shows the predicative results and experimental results are in a good agreement.