Forecasting the load flow of Engine Driven Pump based on Light Gradient Boosting Machine Model

The flow rate of aircraft hydraulic system is one of the important parameters to measure the performance of aircraft. The stability of flow rate of hydraulic pipeline is related to the safety of hydraulic system and even aircraft flight. Aiming at the problem that it is difficult to measure the load flow of aircraft hydraulic system, a prediction method based on ensemble learning for the load flow of Engine Driven Pump (EDP) is proposed. According to the different working conditions of EDP, the Light Gradient Boosting Machine (LightGBM) model is established. Through Particle Swarm Optimization algorithm optimizes the super parameters of LightGBM algorithm, and realizes the regression prediction mode. Finally, compared with the multiple linear regression, random forest, experimental results indicate that the proposed model obviously can improve the prediction accuracy.