Publications

Abstract : An extensive set of data is acquired during engine testing, which is then utilized to evaluate the engine performance characteristics. When engine modifications are carried out in order to improve performance, the whole testing process needs to be repeated. Artificial intelligence-based prediction models can be utilized to reduce the repetitions in engine testing. The data gathered during testing aids in the development of a prediction model that can estimate expected test results with a minimum number of trials. The objective of this study is to predict the in-cylinder pressure of a diesel engine based on the crank angle and load using a model built using artificial neural networks (ANN) in machine learning with MATLAB. ANN prediction model is developed from the data gathered from testing a single-cylinder diesel engine. In ANN, the back propagation algorithm is used to develop the prediction model, which is then validated and compared to the real test data. The best ANN prediction performance is obtained at a mean square error of 0.0012, and the correlation factor is obtained at around 0.9999 for training, testing, and validation. On validation, it is revealed that the ANN prediction model had a high level of accuracy for the outputs and target values. This proven prediction model can predict the in-cylinder values for any single-cylinder diesel engine. The proposed predictive model is envisaged to reduce the time and cost involved during engine development and process improvement.