<![CDATA[In this study, deep learning (DL) model is used to predict brake power (BP) of GX35-OHC 4-stroke, air-cooled, single-cylinder gasoline engine. The engine uses E15 (85% gasoline + 15% ethanol) as a fuel due to its high performance and low emissions. A convolutional neural networks (CNN) model is used on time-series data due to their ability to capture temporal patterns and relationships in sequential data, such as engine BP. While studying the performance of the network, it is found that the root mean squared error (RMSE) is 0.0007, explained variance score (EVS) is 0.9999, and mean absolute percentage error (MAPE) is 0.22%. Compared to traditional machine leaning methods, these metrics demonstrate the high accuracy and reliability of the model, confirming its effectiveness in predicting BP. Various performance curves are plotted such as comparing target and predicted values, regression plots (to indicate the generalization capability), learning curve (to demonstrate the model's effective training progress and convergence), Bland-Altman plot (to show the convergence between the actual and predicted values), histogram and density plot (to show a close fit between predicted and actual values), density plot of actual and predicted outputs, and residual plot (to show randomly distributed errors). This high accuracy and reliability of this DL model help in effective real-time engine performance monitoring, and reducing emission levels, especially for the adoption and use of renewable fuels like E15.]]>
Copyrights © 2024
