基于二次型规划考虑网络丢包的鲁棒状态估计

提出一种可以体现网络丢包的离散时间线性时不变状态空间模型,并将鲁棒状态估计的问题转化为向量优化问题.为了能够快速有效地对该问题进行求解,通过标量化方法将向量优化问题转化为普通的标量二次型规划问题,然后将状态估计问题转化为对标准l1正则化最小平方问题的求解.结合Kalman滤波的更新过程,提出了能够适用于具有数据包丢失情况下的鲁棒状态估计算法,通过仿真实验验证了算法的有效性.

Robust State Estimation for Network Packet Dropout with Quadratic Programming

The paper presents a new discrete time linear time invariant state space model which considers the state estimation with the network packet dropout. Based on this model, the robust state estimation problem is transformed into a vector optimization problem. To solve this problem fast and effectively, the vector optimization problem is transformed into a scalar quadratic programming problem by the scalarization method. And with the further work, the initial problem can finally be transformed to solve a l1 regularized least squares problem, which usually has a standard and fast solution. Associating with the Kalman filter updating procedure, the new algorithm which can be adapted to the condition with the network packet dropout is proposed. The simulation results show that the proposed algorithm is effective.