Optimization of Gasoline Engine to Maximize Brake Power and Minimize Brake Specific Fuel Consumption with Artificial Neural Network and Killer Whale Algorithm

• Main Author: Biyanto, Totok
• Format: Report
• Language: unknown
• Published: INA-Rxiv 2017
• Subjects: Engineering; Law; FOS Law; Mechanical Engineering; FOS Mechanical engineering; Computer-Aided Engineering and Design; Ricardo; Killer Whale; Killer whale
• Online Access: https://dx.doi.org/10.17605/osf.io/dbjwv; https://osf.io/preprints/inarxiv/dbjwv/

In Formula Society of Automotive Engineering (SAE) competition, the design of efficient and powerful combustion engine is required. This paper discussed optimization of gasoline engine using Killer Whale algorithm. The modelling of gasoline engine was built using Multi-Layer Perceptron - Artificial Neural Network (MLP-ANN). A gasoline engine was simulated using Ricardo Wave commercial software to acquire data for training and testing the proposed ANN. The ANN weights were determined by utilizing Levenberg-Marquardt algorithm. The objective function in this paper is to maximize power, minimize the Brake Specific Fuel Consumption (BSFC) and minimize the operational cost. The optimized variables are engine speed (rpm), Air Fuel Ratio (AFR), Mass Fuel Flow (MFF), Intake Pressure (IP), Intake Air Temperature (IAT), Combustion Start (CS) and Throttle Angle (TA). Root Mean Square Error (RMSE) of ANN modelling is 0.021 kW for power and 0.00032 kg/kW.hr for BSFC. The optimization results show that the power increases to 13%, BSFC decreases to 11% and the cost operation decreases to 23% compare with existing design variables.