Carlson应用方差上界技术和信息分配原理论证了当主滤波器和局部滤波器的维数都相同时联邦滤波器的全局最优性,并且信息分配系数选定后是不变的。为了实现当主滤波器维数大于局部滤波器维数时的全局滤波和局部滤波的最优性,该文提出了一种解析补偿和动态最优信息分配系数确定的方法。结果表明:解析补偿方法附加计算量小,同时可作为一种性能指标用于软故障检测;信息分配系数确定的新方法由于实现了联邦滤波器局部滤波的最优性,从而提高了软故障检测灵敏度。 Carlson applied the upper bound of variance and the principle of information distribution to prove the global optimality of the federated filter when the dimensions of the main filter and the local filter are the same, and the information distribution coefficient is unchanged after the selection. In order to achieve the optimality of the global filter and the local filter when the main filter dimension is larger than the local filter dimension, a method of resolving compensation and determining the dynamic optimal information distribution coefficient is proposed. The results show that the analytical compensation method has a small additional computation and can be used as a performance index in soft fault detection. The new method of determining the information distribution coefficient increases the optimality of the local filtering of the federal filter and improves the soft fault detection Sensitivity.