This paper addresses the tracking control problem of the tilting quadcopter with unknown nonlinearities. A novel tilting quadcopter conception is proposed with a fully actuated version, which suggests that the translational and rotational movements can be controlled independently. Based on the Euler-Lagrange equations, the dynamics of tilting quadcopter is developed with uncertainties, where Neural Networks (NNs) are utilized to approximate the unknown nonlinearities in systems. We construct a novel auxiliary filter to obtain the estimation errors explicitly to achieve better approximation ability of NNs. By introducing new leakage terms in the adaptive scheme, the weights of identifier of NNs can converge to their optimal values. And a simple online verification is provided to test the parameter estimation convergence, which relaxes the requirement of persistent excitation condition. Moreover, we propose an Adaptive Finite-time Neural Control for the tilting quadcopter, where all the tracking errors can converge to a small neighborhood around zero in finite time as well as the estimation errors. Finally, comparative simulation results are presented to illustrate the effectiveness and superiority of our proposed control.

本文解决了具有未知非线性的倾斜四轴飞行器的跟踪控制问题。提出了一种具有完全驱动版本的新型倾斜四轴飞行器概念，这表明可以独立控制平移和旋转运动。基于欧拉-拉格朗日方程，倾斜四轴飞行器的动力学是在不确定的情况下开发的，其中神经网络 (NN) 用于逼近系统中的未知非线性。我们构建了一个新的辅助滤波器来明确地获得估计误差，以实现更好的 NN 逼近能力。通过在自适应方案中引入新的泄漏项，NN 的标识符的权重可以收敛到它们的最佳值。并提供了一个简单的在线验证来测试参数估计的收敛性，放宽了对持续激发条件的要求。此外，我们为倾斜四轴飞行器提出了一种自适应有限时间神经控制，其中所有跟踪误差都可以在有限时间内收敛到零附近的小邻域以及估计误差。最后，通过比较仿真结果来说明我们提出的控制的有效性和优越性。