Design optimization of electric control unit casing for diesel dual fuel vehicles in palm oil plantations using backpropagation neural network and genetic algorithm

Diesel vehicles are primary used for transporting fresh fruit bunches (FFB) from palm oil plantations to production facilities within 24 hours to minimize the percentage of free fatty acids (FFA) and ensure high-quality palm oil production. The rise in global oil prices and the government ban on subsidized fuel for plantation industry vehicles have increased the operational costs of transport vehicles in palm oil plantations. Diesel dual fuel (DDF) technology, which allows diesel engines to use both liquid and gas fuels, offers a solution to reduce operational costs and exhaust emissions while maintaining vehicle performance. The operation of DDF in diesel engines is controlled by an electronic control unit (ECU), which requires a casing to protect it from impact, vibration, water, dust, and heat during operation. This study aims to design an ECU casing that protects the electronic components from impact and effectively dissipate heat. The design process involves varying three input parameters: length (120 mm, 130 mm), width (100 mm, 110 mm, and 120 mm), and thickness (1 mm, 1.2 mm, 1.4 mm, 1.6 mm, 1.8 mm, and 2 mm). A total of 36 design variations were simulated using ANSYS software to evaluate temperature response (heat management) and strain response (impact protection). A backpropagation neural network (BPNN) was employed to model the relationship between input parameters and responses using MATLAB software. The genetic algorithm (GA) method utilized the BPNN-generated network as an objective function to optimize the design for maximum casing temperature and minimum strain. The BPNN-GA optimization identified the optimal ECU casing design with a length of 130 mm, a width of 120 mm, a thickness of 1.5 mm, a temperature response of 328.526 K, and a strain response of 0.024488. These predictions were validated using ANSYS, yielding temperature and strain differences of 0.4756% and 2.2092%, respectively.