AI-Based Prognostic Modeling and Performance Optimization of CI Engine Using Biodiesel-Diesel Blends

Abstract

This article describes aspects of the development of an artificial intelligence (AI)-based prognostic modeling and performance optimization of a single-cylinder CI engine powered by biodiesel-diesel blends. It is a tool based on gene expression programming (GEP) followed by response surface methodology (RSM). RSM is employed to establish an explicit mathematical relationship between input and outputs. A database of experimental data on a computerized engine test bench was collected for model development and its testing. The prognostic ability of the GEP model was verified by error analysis, where the coefficient of determination (R 2 ) and mean absolute percentage error (MAPE) varied marginally within the range of 0.979 ± 0.020 and 2.15 ± 0.25, respectively. The model captures adequate trends. Optimum input conditions of engine load, biodiesel-diesel blending ratio, fuel injection pressure, and fuel injection timing are observed to be 60.49 %, 14.32 %, 231.35 bar, and 23.7° bTDC, respectively, while optimized results of brake thermal efficiency (BTE), brake specific fuel consumption (BSFC), and peak in-cylinder pressure (P max ) are found to be 24.28 %, 0.3135 kg/kWh, and 58.95 bar, respectively. GEP approach followed by RSM is observed to be a robust tool. https://dorl.net/dor/20.1001.1.13090127.2021.11.2.19.8