Analysis and design of fuzzy filter algorithms

The concept of constructing a fuzzy filter algorithm based on the largeness of the residuals of the processed data is proposed, which results in smoothing the processed data through an equivalent smoothing window. Based on this approach, three fuzzy filter algorithms are constructed: the fuzzy least square smoother, the fuzzy recursive least square filter, and the fuzzy kalman filter. This approach can also include some extra information such as the richness of the processed data and the possibility of parameter variation in decision making. By doing so, this approach can also be used for abnormal data rejection, forgetting factor adjustment, and parameter tracking. Simulations of applying this method for observation and comparison are conducted, and some comments are given