航空液压系统流量智能预测方法研究

液压系统是飞机重要机载系统之一，它为飞机输出能源驱动，其性能、稳定性和可靠性直接影响飞机的安全性。流量是衡量液压系统稳定性的重要判据，实时监测液压系统管路流量可对系统特性分析、故障诊断提供有力的支持。但由于流量传感器造成的流阻对系统特性有显著影响，因此在航空液压系统中未广泛使用。针对传感器带来的流阻问题，深入分析了与流量相关的参数，提出基于梯度提升回归树的航空液压系统流量预测模型，通过关键参数预测液压系统的流量。最后试验结果表明， GBRT模型相比最小二乘线性回归模型、决策树回归模型、极端梯度提升树XGBoost模型，在预测准确度、训练时间、测试时间等指标中取得了较好的表现，验证提出方法的有效性。

Research on intelligent flow forecasting method of aviation hydraulic System

Hydraulic system is one of the important airborne systems of aircraft. It outputs energy to drive the aircraft, and its performance, stability and reliability directly affect the safety of the aircraft. Flow rate is an important criterion to measure the stability of hydraulic system. Real-time monitoring of pipeline flow rate of hydraulic system can provide powerful support for system characteristic analysis and fault diagnosis. However, the flow resistance caused by the flow sensor has a significant influence on the system characteristics, so it is not widely used in the aviation hydraulic system. Aiming at the problem of flow resistance caused by sensors, the flow-related parameters were analyzed in depth, and a flow prediction model of aviation hydraulic system based on gradient lifting regression tree was proposed to predict the flow of hydraulic system through key parameters. Finally, the experimental results show that the GBRT model has better performance in prediction accuracy, training time, test time and other indicators than the least square linear regression model, decision tree regression model and extreme gradient lifting tree XGBoost model, which verifies the effectiveness of the proposed method.