基于途径概念的数据融合方法

A New Data Fusion Method based on Approach Concept

目前在多传感器条件下提升数据精度的基本方法是状态估计,其中最常用的方法是卡尔曼滤波及其变形算法。由于卡尔曼滤波是依靠时间积累来增加信息量、提高数据精度,因而其收敛速度不可避免地要受到时间因素的影响。另外卡尔曼滤波作为单纯的数据处理方法也无法利用观测系统中一些已知的物理信息。提出了基于途径概念的数据融合方法,它打破了机械的"传感器融合"的概念束缚,有效地帮助人们挖掘和利用潜在的冗余信息,使多传感器信息能够迅速得到充分利用,从而使收敛速度得到提高。基于途径的数据融合方法并未提供一套固定的算法,它提供的只是一套处理规则。

Now the basic method to improve data accuracy under multi-sensor is State Estimation. Among of them Kalmen filter and its transform methods are commonly used. Because Kalmen filter is depending on the time accumulation to increase information and to improve data accuracy, its convergence speed will be inevitably affected by time. Moreover, Kalmen filter is purely a data process method, some physics information which have been known in observation system can＇t be used. In this paper, a new data fusion method based on approach concept is presented. By the concept the restrain of mechanical concept of ＂sensor fusion＂ is broken through. It would help people to find out and make use of the latent redundancy information availably. So multi-sensor information can be used adequately and rapidly, and convergence speed will be accelerated. No a fixed algorithm is provided in data fusion method based on approach instead of a set of process rules. In this paper an actual applied case is offered. The result showed that new method has remarkably increased convergence speed.