Application of artificial intelligence (AI) in characterization of the performance–emission profile of a single cylinder CI engine operating with hydrogen in dual fuel mode: An ANN approach with fuzzy-logic based topology optimization. (24th August 2016)

• Record Type: Journal Article
• Authors: Deb, Madhujit; Majumder, Pinki; Majumder, Arindam; Roy, Sumit; Banerjee, Rahul
• Is Part Of: International journal of hydrogen energy. Volume 41:Number 32(2016)
• Page Start: 14330
• Page End: 14350
• Publication Date: 2016-08-24
• Subjects: Performance -- Emission -- ANN -- NOx -- Soot -- BSEC; Hydrogen as fuel -- Periodicals; Hydrogène (Combustible) -- Périodiques; Hydrogen as fuel; Periodicals; 665.81
• DOI: 10.1016/j.ijhydene.2016.07.016
• Languages: English
• ISSNs: 0360-3199

Abstract

Abstract: The ever-increasing demand for fossil fuels and environmental issues have been the major concerns over the past few decades to search for viable alternative fuels where hydrogen find its suitability to be a viable and promising alternative fuel option on existing IC engine platforms in bridging the contemporary gap to the long term fuel cell based power train roadmap. It's clean burning capability helps to meet the stringent emission norms. Complete substitution of diesel with hydrogen may not be expedient for the time being but the potential use of hydrogen in a diesel engine in dual fuel mode is possible. The study also investigates the use of Artificial Neural Network modeling for prediction of performance and emission characteristics such as BSEC, BTE, NOx, Soot (FSN), UHC, CO2 of the existing single cylinder four-stroke diesel engine with hydrogen in dual fuel mode. Levenberg–Marquardt back propagation training algorithm with logarithmic sigmoid and hyperbolic tangent sigmoid transfer function have resulted in the best model for prediction of performance and emissions characteristics which has been well supported by the trade-off analysis between NOx–Soot (FSN)–BSEC. Fuzzy based analysis has been incorporated into existing ANN model for optimal parameter design which suggests the modesty of the employed transfer function of the existing ANN model. Highlights: Diesel-H2 blends produced higher BTE than pure diesel in all part loads. Diesel-H2 blends reduced BSEC and FSN with increase in NOx emissions. ANN used to predict the performance and emission parameters of the engine setup. MSE, RMSE, MAPE, MSRE, THEIL U2, R, R 2, NSE, KGE metrics used for error analysis. Fuzzy-logic approach was able to predict optimum ANN based transfer function.