On the Robustness Mechanism of H-infinity Filter Based on Gain Maladjustment Gene

H-infinity filtering achieves satisfying robustness at the cost of some estimation precision, while its filtering precision is not assured at the same time, and even always be divergent. The illegibility of the algorithm structure makes the study only a qualitative analysis. Based on a Krein space state-space model, a suit of generalized H-infinity robust filtering algorithm and some important concepts, such as "generalized filtering gain" and "gain maladjustment gene", etc, were founded. Problems about H-infinity filtering robustness were discussed and some correlated conclusions, such as "suboptimal H-infinity robust filtering precision upper limit", "H-infinity filter robustness mechanism" and "H-infinity filter divergence reasons" and so on were proposed. Then, an analysis of filtering strategy effect led to the norm condition of optimal H-infinity filtering disturbances attenuation gene. The last but not the least, H-infinity robust filtering algorithm was decomposed in two taches: Kalman filtering and making robustness, and a suit of "structuring decomposed algorithm" was established, which is based on gain maladjustment gene and clearly opens out the essence of H-infinity robust filtering.