针对具有不确定干扰输入信息系统,讨论了一种滚动时域估计(moving horizon estimation,MHE)方法.在系统含有未知干扰输入的基础上,利用最小方差无偏特性,得到了不受不确定输入影响的估计器.再融合预测控制的滚动优化原理,把系统的硬约束直接表述在优化问题中,并在每个采样时刻通过极小化优化问题的性能指标,估计出系统的初始状态和作用在系统上的扰动.仿真时与递推滤波方法比较,结果表明,MHE能处理系统约束,具有比递推滤波方法更好的估计性能.
A moving horizon estimation (MHE) strategy is discussed on the system with the uncertain disturbance inputs. An estimator unaffected from the unknown disturbance inputs is obtained with the properties of unbiased minimum variance. Based on the moving horizon strategy, the state can be estimated by minimizing the performance object of the optimal problem in every sampling time. Simulation results and comparison results with the recursive filter are given, and the results indicate that the method of MHE is more effective than the constrained linear system.