An Improved Affine Combination Adaptive Filtering Method

Abstract

The invention provides an improved affine combination adaptive filtering method which is realized based on an affine combination adaptive filtering system. The affine combination adaptive filtering system is composed of a known expectation filter and a combination adaptive filter, wherein the combination adaptive filter comprises two subfilters. According to the method, the known expectation filter generates expectation signals of the subfilters and the combination adaptive filter, output signals of the two subfilters are combined through combination parameters to obtain an output signal of the whole combination adaptive filter, and the operation of the two subfilters is controlled through the combination parameters; the weight vectors of the two subfilters are updated with the minimum mean square error method, and the combination parameters are updated through the output signals of the two subfilters; real-time track enables the subfilters to be high in performance, high in convergence speed and low in steady-state deviation. By means of further improvement on the combination parameters, error interference caused by the correlation between noise signals and the two subfilters is avoided.

Description

technical field [0001] The invention belongs to the technical field of digital adaptive filtering, and in particular relates to an improved affine combination adaptive filtering method. Background technique [0002] In the design of adaptive algorithms, convergence speed and steady-state error are two important indicators. However, in general adaptive systems, these two indicators often cannot reach the optimal value at the same time, that is, fast convergence speed causes high Steady-state error, slow convergence speed reduces steady-state error. In order to solve this contradiction, the researchers tried to give a convex combination scheme of the adaptive filtering algorithm. In this scheme, the sigmoid function is used to control the sub-adaptive filter of the convex combination, although to a certain extent The contradiction between convergence speed and steady-state error is alleviated, but this combination is interfered by gradient noise in a complex and changeable en...

AI Technical Summary

Problems solved by technology

[0002] In the design of adaptive algorithms, convergence speed and steady-state error are two important indicators. However, in general adaptive systems, these two indicators often cannot reach the optimal value at the same time, that is, fast convergence speed causes high Steady-state error, slow convergence speed reduces steady-state error

In order to solve this contradiction, the researchers tried to give a convex combination scheme of the adaptive filtering algorithm. In this scheme, the sigmoid function is used to control the sub-adaptive filter of the convex combination, although to a certain extent The contradiction between convergence speed and steady-state error is alleviated, but this combination is interfered by gradient noise in a complex and changeable environment, resulting in performance degradation, and it is difficult to exert its superiority in a complex and changeable environment. The root cause of this defect The reason is that the combination parameters cannot follow the output signal of the sub-filter, and thus cannot effectively switch the sub-filter in time

Images