Prediction of IC engine performance and emission parameters using machine learning: A review

The human kind is facing various natural calamities such as Elnino, forest fires, climate change, etc., due to environmental degradation and pollution. The United Nations has come with sustainable development goals in order to protect the environment and life on earth. The major contributors for degradation of environment are power plants and transport sectors. Usage of IC engines for transportation and power generation is inevitable as IC engine technologies are matured and standardised. However, to meet the sustainable development goals of United Nations, the performance has to be improved and the emissions should be near zero to meet the stringent pollution norms. Various performance parameters of IC engines such as brake power, brake-specific fuel consumption, and brake thermal efficiency are to be improved for various engine operating conditions. Similarly, the emissions such as hydrocarbon, NOx, oxides of carbon, etc. are to be reduced. The ideal operating condition with increased performance and to reduced emission is difficult to achieve. Machine learning is a tool which is used to predict the best operating condition for augmented performance and reduced emissions. In this review article, exhaustive reviews of different machine learning techniques used by different authors are discussed in detailed with reference to performance and emission parameters of IC engines. The widely used machine learning algorithms for prediction of IC engine performance and emission are artificial neural networks, relevance vector method, support vector machine, genetic algorithm, response surface method, and gene expression programming.