PREDICTIONS OF PERFORMANCE AND EMISSION PARAMETERS OF A BIOETHANOL BLENDED VCR SI ENGINE USING ARTIFICIAL NEURAL NETWORKS

Abstract

The influence of alternate fuels in the place of fossil fuels play a prominent role in the automobile sector. This is due to combustion of fossil fuels; the emissions were coming out of the engine which affects the human health and the environment. In order to decrease the emissions (like CO and HC emissions) coming out of the SI engine, an oxygenated fuel (Bioethanol) is selected as an alternate fuel for the SI Engine. To decrease the complexity and to save the time with the thermodynamic calculations, Neural Networks are very useful for predicting the required targets. The performance and emission parameters data is selected from the base paper and are arranged accordingly inputs and targets for the training purpose of neural networks. The selected inputs are % of Gasoline, % of Bioethanol, Specific Gravity, Calorific value, Octane Number, Compression ratio, Brake Power, Mass of Fuel, EGT and the selected targets are BTE, % of CO Emissions, ppm of HC Emissions. The results obtained from the L-M Algorithm are correlated with the experimental results from the base paper using statistical tool such as MAPE and R2. The MSE Value of best validation performance is 0.010456 at the epoch 3 which is in acceptable limits. After observing the experimental and predicted results, it is clarified that the ANN is successful in simulating the engine model.