ANN meta-model assisted MOPSO application in an EPA-Tier 4 constrained emission-performance trade-off calibration problem of a hydrogen-diesel-EGR dual fuel operation. (15th November 2017)

• Record Type: Journal Article
• Authors: Banerjee, Rahul; Mikulski, Maciej; Chakraborty, Amitav; Roy, Sumit; Bose, Probir K.
• Is Part Of: Fuel. Volume 208(2017)
• Page Start: 746
• Page End: 778
• Publication Date: 2017-11-15
• Subjects: AFR Air Fuel Ratio -- AI Artificial Intelligence -- AMF Adaptive Merit Function -- ANN Artificial Neural Network -- BDO Baseline Diesel Operation -- BP Brake Power -- BSEC Brake Specific Energy Consumption -- BSFC Brake Specific Fuel Consumption -- BSFCeq Brake Specific Fuel Consumption Equivalent -- BTDC Before Top Dead Centre -- BTHE Brake Thermal Efficiency -- C/O Carbon to Oxygen Ratio -- CA Crank Angle -- CAE Computer Aided Engineering -- CFD Computational Fluid Dynamics -- CI Compression Ignition -- CNG Compressed Natural Gas -- CO Carbon Monoxide -- CO2 Carbon Dioxide -- CRDi Common Rail Direct Injection -- DAQ Data Acquisition -- DF Diesel Flow -- DH Diesel-Hydrogen Operation w/o EGR -- DM Decision Maker -- DPF Diesel Particulate Filter -- EA Evolutionary Algorithm -- ECU Engine Calibration Unit -- EGR Exhaust Gas Recirculation -- EHEPR Effective Hydrogen Energy Participation Ratio -- EPA Environmental Protection Agency -- EQ % EGR -- ETC European Transient Cycle -- EVO Exhaust Valve Open -- FCV Fuel Cell Vehicle -- FE Finite Element -- GA Genetic Algorithm -- GDI Gasoline Direct Injection -- GFNN Generalized Feed- Forward Neural Network -- H/C Hydrogen to Carbon Ratio -- H2FCV Hydrogen fuelled Internal Combustion Engine -- H2ICE Hydrogen fuelled Internal Combustion Engine -- HC Hydro carbon -- HCCI Homogeneous Charge Compression Ignition -- HES Hydrogen Energy Share -- HF Hydrogen Flow -- HIL Hardware in the Loop -- ICE Internal Combustion Engine -- IVC Inlet Valve Closing -- IVO Inlet Valve Opening -- kW Kilowatt -- LHC Latin Hypercube -- LNT Lean NOx Trap -- MIMO Multiple Input Multiple Output -- MISO Multiple Input Single Output -- MLP Multi-Layer Perceptron -- MOOP Multi Objective Optimization Problem -- MOPSO Multi Objective Particle Swarm Optimization -- MTDOM Maximum Trade-off Merit -- MCR Maximum Continuous Rating -- N Crankshaft Speed -- NHC NOx + TUHC -- NOx Oxides of Nitrogen -- NSGA-II Non-dominated Sorting Genetic Algorithm-II -- OFAT One Factor At A Time -- PC Personal Computer -- PF Particulate Filter -- PM Particulate Matter -- PPM Parts Per Million -- PSO Particle Swarm Optimization -- RPM Revolutions Per Minute -- SIT System Identification Technique -- SCR Selective Catalytic Reduction -- SFC Specific Fuel Consumption -- SFTP Supplemental Federal Test Procedure -- SI Spark Ignition; ANN based Meta-model -- Multi-objective Particle Swarm Optimization -- EPA-Tier 4 emission constraints -- Emission-performance trade-off calibration -- Maximum Continuous Rating re-calibration -- Diesel-hydrogen-EGR dual fuel operation; Fuel -- Periodicals; Coal -- Periodicals; Coal; Fuel; Periodicals; 662.6
• DOI: 10.1016/j.fuel.2017.07.037
• Languages: English
• ISSNs: 0016-2361

Abstract

Abstract: The present study provides a comprehensive perspective on the unique opportunities that present day artificial intelligence based meta-modelling strategies pose in internal combustion engine system identification paradigms, especially in its multi-objective calibration domains. The pertinence of the advantages of such AI based meta-modelling strategies and the potential of swarm optimization strategies have been rationalized with the chronological growth in the contemporary necessities of the diesel engine. The study clearly establishes the pivotal significance of perusing true multi-objective optimization strategies commensurate with the increasing need to address the emission-performance trade-off challenges of the diesel engines of the day. In order to highlight the credibility and scalability of such meta-model based multi-objective optimization opportunities in internal combustion engine domains, the present study presents a unique case study showing the distinct possibility of harnessing the synergistic potential of a computationally cost effective and commendably accurate meta-model based calibration endeavour in an existing diesel engine. The study was first-of-a-kind foray into the complexities of a hydrogen-diesel dual fuel operation under EGR application of varying thermal signatures along with the introduction of a trade-off merit index in the optimization workflow. The architecture was based on an ANN system identification platform wherein the MOPSO algorithm was employed to improve the emission-performance trade-off characteristics of the dual fuel operation. Further, in order to corroborate the contemporary relevance of the multi-objective optimization endeavour, additional constraints of the EPA Tier 4 diesel emission mandates were imposed to the case study. Validation of the optimization results indicated a 10.2%, 30.6%, 25.4% and 9.4% improvement in the performance-emission trade-off footprint when compared with the corresponding experimental dual fuel-EGR operations at 20%, 40%, 60% and 80% full load steps respectively.