Sequential Procedure for Improving the Efficiency of CI Engine by Using Artificial Neural Networks

This work provides a sequential approach to improve efficiency of Combustion Ignition engines which involves both performance and emissions by using the artificial neural network (ANN). In recent years continuous work is going on for improving the output and reducing the emissions especially for Combustion Ignition engines which are mostly used for transportation purposes. In view of the above, the experimental data of a four stroke Combustion Ignition engine is taken as reference. However, the experimental data is split into three categories as input data, target data and output data in neural networks. All these data are trained using neural network toolbar in MATLAB with ten hidden layers by which error deviation are calculated, in order to reduce error deviation between neural network and experimental values, design of inlet manifolds is varied and performance parameters along with emissions is calculated and compared with neural network values. The results showed minimum error over the emission and performance parameters of CI engines from the manifold designs and ANN model. These results provide a sequential approach to improve efficiency of Combustion Ignition engines with the help of neural networks.