The biomedical signals such as ECG, EMG, EEG, and etc are very Important information to diagnose patients The signal is hard to filter the noise because that is mixed with a lot of noise and biomedical signal has the properties of nonlinear and time-variance. So, we will filter under the measure environment for system or patient. But the general adaptive fillet has brought on the distortion of signal because the adaptive filter adjust the filter coefficient with the fixed order of filter, that filter has the unsuitable order in each other environment. So we propose the dynamic structure adaptive filter that is used for improving that disadvantage. In experiment, we obtain the optimal order of adaptive filter and have food results.