基于神经网络的直升机协调转弯控制律设计

针对直升机协调转弯控制提出了一种控制律设计方法。首先,在小扰动线性模型基础上,建立描述直升机全包线范围内的T-S模糊模型。其次,分析直升机协调转弯机理,确定相应的控制律结构,利用参数优化算法确定控制参数。最后用所得的控制参数样本训练径向基神经网络,实现全包线范围内的连续调参控制律。仿真结果表明,直升机转弯性能完全满足指标要求,所提出的控制律设计方法可行、有效。

Controller Design for Helicopter Coordinated Turn Based on Neural Network

This paper proposes a controller design for the coordinated turn of a helicopter.First,it building a T-S model to describe the varying characteristics of a helicopter to improve simulation accuracy.Then,using the parameter optimization algorithm to search for control parameters at 12 typical flight conditions,which are used to train a radial-basis function neural network until it can output proper control parameters for any flight condition within the envelope.In the end,the simulation result shows the turning performance fully satisfies control requirements so that this design is feasible and effective.