Research on data-driven energy efficiency optimisation algorithm for air compressors
Zhongping Ning, H. Zeng, Zhen Tian
Abstract
In this paper, a data-driven multi-objective optimisation algorithm is proposed for the energy efficiency optimisation problem of air compressor system. The energy consumption prediction model is constructed by XGBoost, and the improved NSGA-II algorithm is used to simultaneously optimise the three objectives of energy consumption, pressure stability and equipment life, and the LSTM load prediction model is combined to achieve the feed-forward control of the system. The single-unit parameter optimisation and group synergy strategies are designed for fixed-frequency units and inverter units respectively, and the energy efficiency is improved by adjusting the exhaust pressure, loading and unloading differential pressure and dynamic unit selection. Experimental validation shows that the algorithm has good adaptability under both stable and fluctuating operating conditions, especially under low load and rapid load change conditions. Comparison tests with traditional control methods verify the superiority of the algorithm, which not only significantly reduces energy consumption, but also improves pressure stability and reduces the frequency of equipment startup and shutdown.