Optimization of injection parameters, and ethanol shares for cottonseed biodiesel fuel in diesel engine utilizing artificial neural network (ANN) and taguchi grey relation analysis (GRA)

Abstract The increase of fossil fuel powered industrial processes and vehicles has resulted in the exhaustion of petroleum reserves and pollution of the environment. Because of its clean-burning, renewable, and biodegradable qualities, biodiesel is becoming more and more recognized as a potential diesel fuel alternative. The present study investigates engine performance and emission characteristics of cottonseed oil (CSBD20) and diesel blends tested on single-cylinder compression ignition engine by several injection timings, injection pressures, and ethanol shares. Performance parameters such as brake thermal efficiency (BTE), brake-specific fuel consumption (BSFC), exhaust emissions such as hydrocarbons (HC), carbon monoxide (CO), nitrogen oxides (NO x ), carbon dioxide (CO 2 ), and smoke were considered as output factors, considering injection timing (IT), ethanol share (ES), injection pressure (IP) as input factors utilizing artificial neural network (ANN) and taguchi grey relation analysis (GRA). The ANN model accurately predicts the input-output relationships of ethanol and cottonseed biodiesel blends, as validated by experimental comparisons. The predicted values for BTE, BSFC, HC, CO, NO x , and smoke show close alignment with experimental results, with marginal errors of 6.2 %, 2.8 %, 7.1 %, 4.7 %, 6.8 %, and 5.6 %, respectively, confirming its reliability. In addition, this study utilized Taguchi grey relational analysis (GRA) to find optimum engine operating conditions. The analysis revealed that the optimal engine operating conditions were IT at 27° CA bTDC, ES at 15 %, and IP at 200 bar. Furthermore, confirmation tests are also conducted at optimum operating conditions, and the revealed values are closer to taguchi GRA experiments and ANN predicted values.