可重复使用火箭在大气层内的回收着陆采用气动力和推力完成高精度的定点垂直软着陆。在动力下降段中，存在推力偏差、气动模型偏差和风扰动等各种干扰，将降低终端着陆精度并损失性能指标，制导系统面临着抗干扰的难题。为解决该难题，提出了一种组合干扰补偿制导方法，根据干扰是否可建模描述，将干扰分为可模型化的干扰和不可模型化的干扰，分而治之地处理。可模型化的干扰是指可以用模型描述的干扰，考虑将其用于最优制导中来提升性能指标；不可模型化的干扰是指难以用模型描述的干扰，考虑仅实时补偿其对终端约束的不利影响。在组合干扰补偿制导框架中，首先，设计了干扰估计器对两类干扰进行实时估计；然后，设计了邻近最优干扰补偿制导算法，利用可模型化的干扰的估计值对最优制导指令进行实时修正补偿，在保证终端约束条件下利用可模型化的干扰提升性能指标；最后，设计了终端不变性干扰补偿制导算法，通过计算不可模型化的干扰引起的终端约束摄动，实时补偿不可模型化的干扰对终端约束的不利影响来保证终端不变性。仿真结果表明，所提出的组合干扰补偿制导方法能够在保证终端着陆精度的同时，提升性能指标的最优性，同时对于各种干扰具有较强的鲁棒性。

The reusable rocket recovery landing in the atmosphere uses aerodynamic force and thrust to complete a high-precision pinpoint vertical soft landing. In the powered descent phase, there are various disturbances such as thrust deviation, aerodynamic model deviation and wind disturbance, which will reduce the terminal landing precision and lose the performance index, so the guidance system faces the difficulty of disturbance rejection. To solve this issue, a combined disturbance compensation guidance method is proposed, which divides the disturbances into modelable and unmodelable disturbances according to whether they are modelable to be described or not, and treats them sepa-rately. The modelable disturbances are those that can be described by a model and are considered for optimal guid-ance to improve the performance index; while the unmodelable disturbances are those that are difficult to be de-scribed by a model and are only considered to compensate their adverse effects on the terminal constraints in real time. In the combined disturbance compensation guidance framework, first, a disturbance estimator is designed to estimate the two types of disturbances in real time. Then, a neighboring optimal disturbance compensation guidance algorithm is designed which corrects and compensates the optimal guidance command in real time using the estimat-ed values of modelable disturbances, to improve the performance index using modelable disturbances while ensuring the terminal constraints. Finally, a terminal invariance disturbance compensation guidance algorithm is designed to ensure terminal invariance by calculating the terminal constraint perturbations caused by unmodelable disturbances and compensating the adverse effects of unmodelable disturbances on terminal constraints in real time. Simulation results show that the proposed combined disturbance compensation guidance method can improve the optimality of performance index while ensuring the terminal landing precision, and has strong robustness to various disturbances.