Sliding interval width optimization based on particle swarm algorithm for variable-speed diesel engines

To address the impact of interval width on the optimization results of variable-speed diesel engine fuel consumption, this study proposes an improved particle swarm optimization (PSO) algorithm for sliding window width optimization. First, a backpropagation (BP) neural network model is established to predict fuel consumption rates. During speed optimization, the BP neural network is invoked for calculations, and its accuracy is validated with training and testing correlation coefficients of 0.998 and 0.997, respectively. The universal characteristic curves derived from the BP model closely align with experimental data, confirming its reliability. After optimizing the sliding window width, the variable-speed diesel engine operates along the minimum fuel consumption rate curve under varying loads, effectively mitigating the adverse effects of interval width on fuel efficiency. The results demonstrate a maximum fuel-saving rate of 4% under low-load conditions and varying degrees of improvement across other load ranges. This approach provides a foundation for real-time fuel consumption optimization in practical applications.