P. Srinivas Reddy, M. Chandra Sekhar Reddy, Narsimhulu Sanke and D. Teja Santosh
Abstract
The Low Heat Rejection engine (LHR) is useful to reduce the engine coolant heat losses thereby improving the combustion engine performance. The key performance measurements namely Brake Thermal Efficiency (BTE), Brake Mean Effective Pressure (BMEP) and Brake Power (BP) are calculated for analyzing the performance and efficiency of the engine when the blend of algae oil, diethyl ether and copper nanoparticles is injected into medium grade LHR engine. However, factors such as injection pressure, injection timing, exhaust gas temperature, coolant load and volumetric efficiency have their effect on the estimations of these performance parameters. To estimate the LHR engine performance for better engine efficiency, machine learning based Multiple Regression modeling is implemented. The physical medium grade LHR engine experimental setup is done and the experiments are carried out. The operational parameters (factors) and key performance parameters data are gathered from these experiments, exploratory data analysis is then carried out, relevant operational parameters that influence the performance parameters are identified which are then used for learning the regression- based models that estimate the performance of the medium grade LHR engine. These models are validated to ensure have better generalization capabilities to unseen data. It is observed that a high R-squared value of 1 for BTE and BP performance parameters and very less Mean Square Error value of 5.259072701473412e-31 for BMEP performance parameter are achieved. These values prove that machine learning help optimize LHR engine operation and reduce downtime.