Diesel engine model based on neural networks topology

Abstract EN

In this paper, neural networks engine model is presented, which represents a Diesel engine under a wide range of operating conditions.

Experimental data obtained from 57 tests on Diesel engine under different loading conditions and at three ambient temperatures of (20, 30 and 50 oC) were used to develop a neural network model.

The input data cover applied torque, fuel mass flow rate, and ambient temperature, and the output data cover engine speed, indicated power, air box differential pressure, and indicated mean effective pressure.

The feed-forward neural network is used to build engine model and back-propagation training algorithm with two hidden layers is adopted due to its accuracy of prediction.

The results show that the simulation data from the model and the measured data of the engine tests are in good agreement.

The possibility of using the presented engine model with controller is tested using PID controller.

The value of fuel mass flow rate is used as controller input and speed engine as output of the system when the applied torque is changing