基于GA-BP神经网络的大型客机气流角估计方法

AirFlow Angle Estimation Method for Large Passenger Aircraft Based on GA-BP Neural Network

为了解决硬件冗余难以克服的气流角传感器共因故障问题,进一步提高飞机气流角信号的可靠性,研究了基于GABP神经网络的气流角估计方法。通过BP神经网络融合姿态角、加速度、风速等参数来实现不依赖气流角传感器的气流角估计;引入遗传算法对神经网络权值和阈值进行全局优化,提高估计精度;对某大型客机的试飞数据预处理后用于模型的训练和测试。仿真结果表明,训练完成的GA-BP神经网络模型对气流角的估计值贴近实际值,稳定性和精度明显高于BP神经网络。上述方法给飞机增加一个余度的气流角信号,可用于传感器故障时为飞机提供可靠的气流角信号。

In order to solve the common cause fault of airflow Angle sensor which is difficult to overcome by hardware redundancy and further improve the reliability of aircraft air flow Angle signal,an air flow Angle estimation method based on GA-BP neural network was studied.BP neural network was used to integrate attitude Angle,acceleration,wind speed and other parameters to estimate the flow Angle independently of the flow Angle sensor.Genetic algorithm was introduced to optimize the weights and thresholds of neural network globally to improve the estimation accuracy.The model was trained and tested with the pre-processed flight test data of a large aircraft.The simulation results show that the trained GA-BP neural network model's estimation of the airflow angle is close to the actual value,and the stability and estimation accuracy are significantly higher than those of the BP neural network.This method adds a residual air flow Angle signal to the aircraft,which can be used to provide reliable air flow Angle signal for the aircraft when the sensor is faulty.