近空间飞行器动态受限控制分配技术研究

发布时间：2018-07-30 08:30

【摘要】：近年来,近空间飞行器相关技术受到了各主要大国的密切关注。近空间飞行器作为一种新型飞行器,飞行速度快、飞行高度高,现有的防空武器难以对该类型飞行器构成威胁,且可重复使用,投入成本低于发射的卫星。因而,近空间飞行器在高空侦察、快速打击等领域具有独特的优势。近空间飞行器常采用多操纵面新型气动布局方式以提高飞行器的可靠性、安全性和可控性等性能。因此如何将虚拟控制指令转化为操纵面的控制指令是近空间飞行器飞行控制系统迫切需要解决的问题。本文针对近空间飞行器飞行控制系统设计过程中遇到的控制指令分配问题,研究了考虑操纵面物理约束的控制分配方法,主要包括以下几项研究内容:首先,在分析操纵面控制效果和不同操纵面的控制力矩在物理约束范围内所能产生的控制效果的基础上,建立了引入控制分配的近空间飞行器数学模型。考虑飞行器操纵面动态,分析了两种典型的驱动器模型对控制指令的动态响应。其次,针对操纵面受限问题,采用基于鱼群算法的混合优化算法实现了近空间飞行器的控制分配。通过引入遗传算法,解决了鱼群算法在后期会陷入局部收敛于最优解的问题。仿真结果表明基于遗传算法的混合鱼群优化方法能有效地实现控制分配。随后,针对近空间飞行器飞行过程中会受到外部干扰,以及控制效率矩阵不确定所导致控制分配误差的情况,设计了能在有限时间收敛的基于滑模干扰观测器的终端滑模控制器。并将该方法与鲁棒最小二乘控制分配方法结合,实现了近空间飞行器在操纵面控制效能为区间不确定情况下的有限时间控制分配。仿真结果充分验证了该控制分配方案的有效性。接着,考虑操纵面控制效能为非线性的情况,将非线性控制分配问题转化为求解分段截点组合和优化对应的分段截点系数的两个优化问题。对控制效率矩阵在当前时刻物理约束范围进行分段,先利用遗传算法获取最优分段区间,再利用鸽群启发算法优化截点系数。仿真结果表明所设计的分步混合优化算法能在操纵面控制效能为非线性的情况下取得令人满意的控制分配结果。最后,考虑飞行器操纵面驱动器动态响应,研究了将径向基神经网络与差分进化算法结合的动态控制分配方法。该方法在充分考虑执行器动态的情况下,利用径向基神经网络的离线和在线逼近非线性函数的能力预测差分算法迭代过程中个体能产生的控制效果,使得差分算法可得到各个操纵面的最优偏转指令。仿真结果显示该方法有效提升了控制分配效果。
[Abstract]:In recent years, the near-space vehicle related technology has been paid close attention by the major countries. As a new type of aircraft, the near space vehicle (NV) has high flying speed and high altitude. It is difficult for the existing air defense weapons to pose a threat to this type of aircraft, and can be reused, and the input cost is lower than that of the launched satellite. Therefore, near-space vehicles have unique advantages in high-altitude reconnaissance, fast attack and other fields. In order to improve the reliability, safety and controllability of near space vehicles, a new type of aerodynamic layout with multiple control surfaces is often adopted. Therefore, how to transform the virtual control instruction into the control instruction of the control surface is an urgent problem to be solved in the flight control system of the near space vehicle. In this paper, a control assignment method considering the physical constraints of the control surface is studied in view of the control instruction assignment problem encountered in the design of the flight control system of the near space vehicle. The main research contents are as follows: first of all, Based on the analysis of the control effect of the control surface and the control effect produced by the control torque of different control surfaces within the range of physical constraints, the mathematical model of the near-space vehicle with control assignment is established. In this paper, the dynamic response of two typical actuator models to the control instructions is analyzed considering the control surface dynamics of the aircraft. Secondly, a hybrid optimization algorithm based on fish swarm algorithm is used to realize the control allocation of near-space vehicle. Genetic algorithm is introduced to solve the problem that fish swarm algorithm will fall into local convergence to the optimal solution in the later stage. Simulation results show that the hybrid fish optimization method based on genetic algorithm can effectively achieve control allocation. Then, the control assignment error caused by the uncertainty of the control efficiency matrix and the external disturbance during the flight of the near-space vehicle will be considered in view of the situation that the control allocation error is caused by the uncertainty of the control efficiency matrix. A terminal sliding mode controller based on sliding mode disturbance observer is designed, which can converge in finite time. This method is combined with the robust least square control allocation method to realize the finite time control allocation of the near space vehicle under the condition that the control efficiency of the near space vehicle is interval uncertainty. The simulation results fully verify the effectiveness of the control allocation scheme. Then, when the control efficiency of the control surface is nonlinear, the nonlinear control assignment problem is transformed into two optimization problems for the combination of piecewise sections and the optimization of the corresponding piecewise truncation coefficients. In this paper, the control efficiency matrix is segmented in the physical constraint range at the present time. The optimal segmental interval is obtained by genetic algorithm, and then the truncation coefficient is optimized by pigeon swarm heuristic algorithm. The simulation results show that the proposed hybrid optimization algorithm can obtain satisfactory control allocation results when the control efficiency of the control surface is nonlinear. Finally, a dynamic control allocation method combining radial basis function neural network and differential evolution algorithm is studied. With the full consideration of actuator dynamics, this method uses the ability of off-line and on-line approximation of radial basis function neural networks to predict the control effect that can be produced by the individual in the iterative process of the difference algorithm. The difference algorithm can obtain the optimal deflection instruction of each control surface. Simulation results show that this method can effectively improve the control allocation effect.
【学位授予单位】：南京航空航天大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：V249.12

【相似文献】