应用于状态监测的多传感器融合估计

在状态监测的工程实际中, 使用多个同类传感器进行在线测量可以得到更为准确的状态估计．但各传感器测量噪声会出现相关的情况, 而且很难得到相关测量噪声的方差矩阵的精确值, 测量系统往往是不确定的．本文根据系统测量将系统分解为确定和不确定扰动两部分, 分别进行估计, 然后将两者的融合估计结果相加得到了最优鲁棒的融合估计．针对确定部分, 利用同类传感器的测量方差为Pei-Radman矩阵的特性, 通过求解测量噪声方差矩阵的最大特征值得到了一种简便的最优融合估计算法, 该算法避免了求解方差矩阵的逆的过程．针对不确定

Multisensor fusion estimation in state monitoring

In practical state estimation, multiple identical sensors are commonly employed to produce more accurate results. However, measurement noises from different sensors are generally correlated with covariance matrix which is difficult to be determined accurately. Moreover, the measured system may contain uncertain parts. This paper decomposes the measured system into two parts, the certain part and the uncertain part. The states of each part are estimated respectively, and the two estimated results are combined to produce the final fusion estimation. For the certain part, a simple optimal fusion estimation algorithm is proposed for computing the maximum eigenvalue of the Pei-Radman measurementnoise covariance matrix. For the uncertain part, a robust fusion algorithm is also developed in terms of the linear matrix inequality(LMI), based on the polytopic models.