非线性预测滤波在星敏感器姿态确定中的应用

摘要

为研究无陀螺卫星控制优化问题的新方法,针对环境干扰力矩和估计误差等不确定因素导致滤波器稳定性和估计精度降低,提出了使用星敏感器测量值在线估计模型参数和卫星姿态的非线性预测滤波方法.根据预测滤波理论推导了关于模型误差的损失函数,利用线性化的测量方程求得使损失函数最小化的模型误差值,代人状态方程求数值积分,得到卫星的姿态估计参数.仿真结果证明,不仅简化了计算,且适应性更强.而应用四元数描述卫星姿态,避免了欧拉角法的奇异性问题.仿真结果还表明,方法收敛速度较EKF更快,状态估计精度与EKF相当,并对非线性模型误差具有良好的跟踪性能.%The filter estimation accuracy can be lowered using non-linear attitude motion model while determining satellite attitude without gym, because of the influence of some uncertain factors such as environmental disturbance torque, estimate error and so on. A nonlinear predictive filter solely based on star sensor measurements without angu1ar rate measurements is proposed to estimate satellite attitude online as well as the model parameter. A cost function containing model error is derived based on the predictive filter theory, and the model error is calculated which minimizes the cost function with the linearized observation equation. Then the estimated attitude parameters are numerically integrated by substituting the model error into the state equation. Comparing with the extended Kalman filter, the method is calculated more effectively and more adaptively, and the singularity of the Euler angles is avoided by using quaternion parameters. The simulation result indicates that the method has faster convergence speed than the EKF,and the precision is nearly the same as EKF. Addicfively, the method has a good tracking performance to the nonlinear model error.