超小型旋翼机平面转弯运动的神经网络补偿建模

采用神经网络补偿机理模型法进行超小型旋翼机系统的平面转弯运动建模。首先建立系统的非线性参数化模型，通过把非线性模型在平衡点线性化，得到描述系统平面转弯运动的线性机理模型；然后针对转弯运动，采用对角回归网络来补偿机理模型与非线性模型之间的误差，从而获得超小型旋翼机系统的平面转弯运动的混合模型。仿真结果表明，该混合模型比单纯机理模型有更高的精度，且所获得的补偿神经网络具有很强的泛化能力。

Modeling of a Subminiature Rotorcraft in Horizontal Turning by Using NN-Compensation

A hybrid modeling method of a subminiature rotorcraft in horizontal turning had been proposed by using mechanism model and NN-compensation. First, the nonlinear dynamics of a subminiature rotorcraft was established. Through linearization of the nonlinear dynamics on the trim point, the linear time-invariant mechanism model in horizontal turning had been obtained. Then according to horizontal turning, a diagonal re-cursive neural network was used to compensate the mod- el error between the mechanism model and the real nonlinear model, thus the hybrid model of a subminiature rotorcraft in horizontal turning was achieved. Simulation results show that the hybrid model has higher accuracy than that of mechanism model and the obtained compensated-NN has high ability of generalization.