论文部分内容阅读
基于径向基函数(Radial basis function,RBF)神经网络构建了一种带后缘襟翼主动控制(Active controlled flap,ACF)的旋翼振动载荷计算模型。采用正交试验方法确立RBF网络训练样本的输入,在CAMRAD II中计算前飞状态下与训练样本对应的旋翼桨毂六力素,并将主通过频率下的分量作为样本输出,对RBF网络进行离线训练。在此基础上采用多周控制器对被控模型进行振动载荷主动控制。随后以2桨叶4m直径ACF旋翼为例,构建了其桨毂减振分析方法,并对桨毂动载荷各分量的减振效果进行了分析。研究表明,采用正交样本训练的RBF网络能够精确映射襟翼偏角与桨毂振动载荷的非线性关系,施加多周控制后,桨毂垂向振动载荷降低接近50%,其他方向的振动载荷也有不同程度的降低。
A model of rotor vibration load calculation based on active radial flap (ACF) is proposed based on Radial Basis Function (RBF) neural network. The input of training samples of RBF network was established by orthogonal test method. The six factors of rotor hub corresponding to the training samples in pre-flight state were calculated in CAMRAD II, and the components of main passing frequency were taken as samples to output to RBF network Offline training. On the basis of this, a multi-week controller is used to control the charged model actively. Then taking the 2-blade 4m diameter ACF rotor as an example, the method of damping analysis of its hub was built, and the damping effect of each component of the dynamic load on the hub was analyzed. The results show that the RBF neural network trained by orthogonal sample can accurately map the nonlinear relationship between the deflection angle and the hub vibration load. After multi-cycle control, the vertical vibration load of the hub decreases about 50% and the vibration load in other directions There are also different levels of reduction.