Neural network prediction based on dung beetle optimization algorithm and engine performance emission optimization using multi-objective rime optimization algorithm

In the continuously evolving field of multi-objective optimization algorithm(MOOA), a pioneering approach known as the multi-objective rime optimization algorithm (MORIME) has been developed to address engineering optimization challenges. This paper centres on the collaborative optimization of engine torque and NOx emissions through the decision variables of INJ (injection mass), SOI (start of injection) and Intercooler(intercooler temperature). Initially, a simulation model of the engine is established using GT-Power(GT-Power is a module within the GT-SUITE software designed for simulating engine operating processes. It has been developed by Gamma Technologies, a company based in the United States.) to ensure its reliability. Subsequently, the optimal Latin Hypercube method and the simulation model are employed to generate an initial dataset consisting of 400 workload points. Based on this dataset, artificial neural network (ANN) prediction models for NOx emissions and torque are developed. The experimental results demonstrate that the use of 10 neural nodes yields faster computation times and better outcomes compared to using 20 neural nodes. Additionally, it is noteworthy that all optimized prediction models utilizing the dung beetle optimization (DBO) consistently achieve correlation coefficients exceeding 99 %. The experimental findings demonstrate an improvement in optimizing emissions for the internal combustion section of the China Rejuvenation Plateau internal electric dual-source electric multiple units(EMU), achieved while meeting dual objective constraints. These findings suggest that integrating optimization algorithms with ANN and multi-objective decision-making holds substantial potential for further enhancing overall engine performance.