Vous êtes sur la page 1sur 18

DESIGN MODEL SIMULINK

TO DENOISE ECG SIGNAL


USING VARIOUS IIR & FIR
FILTERS
Analisis Sinyal Biomedis
S-1 Teknobiomedik
Universitas Airlangga
2017
Presented by :

Afni Unaizah
Fadhilah Qalbi Annisa
Maydiana Nurul Kurniawati
abstract
Electrocardiogram (ECG) signal is usually corrupted by several artifacts and these must
be removed before diagnosis. T his paper therefore presents the design of various FIR
filters like Kaiser, Rectangular, Hamming, Hanning, Gaussian and Bartllet window
techniques and FIR Equiripple Filter. IIR filters like Butterworth, Chebychev I & II
and Elliptic Filters are also explored to remove the artifacts in ECG signal. To verify
the usability of designed filters, noisy ECG signal is generated by adding random noise,
white noise and 50 Hz interference (hum) to the 100m.mat ECG data sample taken
from MIT-BIH database. Signal to noise ratio (SNR) is calculated to compare the
performance of different filtering methods using Simulink Model. Results obtained
show that FIR Equiripple filter gives a SNR of 7.71 for a 3rd order filter and IIR
Elliptic filter has SNR of 2.62 for first order filter and are therefore recommended.
Material and Method
1. 100m.mat ecg data sample from MIT-BIH database
2. Sampling frequency : 1000 Hz
3. Cut off frequency : 100 Hz
4. Lowpass FIR filter (Kaiser, Rectangular, Gaussian, Hanning, Barlett, Equiripple)
5. Lowpass IIR Filter (Butterworth, Chebyshev I, Chebyshev II, Elliptic
6. Noisy Signal
Random Noise
White Gaussian Noise
50 Hz Interference
Listing Program
Simulink Model
SNR Result
FIR
Orde Kaiser Hamming Rectangular Gaussian Hanning Bartlett Equiripple
1 2.192 2.192 2.192 2.192 - - -
2 2.308 2.192 2.311 1.174 - - -
3 2.304 2.192 2.31 1.964 1.898 1.898 7.503
10 - - - - 1.048 1.1 -3.69

IIR
Orde Butterworth Chebyshev-I Chebyshev-II Eliptic 1.08
1 2.464 2.306 2.722 2.306
2 1.956 3.519 2.167 3.519
10 -0.1832 0.8931 0.2603 0.914
SNR Result
FIR
Orde Kaiser Hamming Rectangular Gaussian Hanning Bartlett Equiripple
1 2.192 2.192 2.192 2.192 - - -
2 2.308 2.192 2.311 1.174 - - -
3 2.304 2.192 2.31 1.964 1.898 1.898 7.503
10 - - - - 1.048 1.1 -3.69

IIR
Orde Butterworth Chebyshev-I Chebyshev-II Eliptic 1.08
1 2.464 2.306 2.722 2.306
2 1.956 3.519 2.167 3.519
10 -0.1832 0.8931 0.2603 0.914
Transfer Function of IIR Filter
Jenis Filter FIR Orde 1 Orde 2 Orde 10

Butterworth

Chebyshev-I

Chebyshev-II

Eliptic 1.08
Difference Equation of FIR Filter
Kaiser orde 1 : y(n)=0.5*x(0)+0.5*x(n-1)
Kaiser orde 2 : y(n)=0.3188*x(0)+0.3624*x(n-1)+0.3188*x(n-2)
Kaiser orde 3 : y(n)=0.2262*x(0)+0.2738*x(n-1)+0.2738*(n-2)+0.2262*(n-3)
Hamming orde 1 : y(n) =0.5*x(0)+0.5*x(n-1)
Hamming orde 2 : y(n)=0.5*x(0)+0.5*x(n-1)
Hamming orrde 3 : y(n)=0.5*x(0)+0.5*x(n-1)
Rectangular orde 1 : y(n)= 0.5*x(0)+0.5*x(n-1)
Rectangular orde 2 : y(n)= 0.3258*x(0)+ 0.3483*x(n-1)+ 0.3258*x(n-2)
Rectangular orde 3 : y(n)= 0.233*x(0)+ 0.267*x(n-1)+ 0.267*x(n-2)+0.233*x(n-3)
Difference Equation of FIR Filter
Gaussian orde 1 : y(n)=0.5*x(0)+0.5*x(n-1)
Gaussian orde 2 : y(n)=0.038*x(0)+0.9240*x(n-1)+0.038*x(n-2)
Gaussian orde 3 : y(n)=0.0257*x(0)+0.4743*x(n-1)+0.4743*x(n-2)+0.02573*x(n-3)
Hanning orde 3 : y(n)=0.5*x(n-1)+0.5*x(n-2)
Hanning orde 10 : y(n)=0.0055*x(n-1)+0.0428*x(n-2)+0.12151*x(n-3)+0.2076*x(n-4)+0.2453*x(n-
5)+0.2076*x(n-6)+0.1215*x(n-7)+0.0428*x(n-8)+0.0055*x(n-9)
Bartlett orde 3 : y(n)=0.5*x(0)+0.5*x(n-1)
Bartlett orde 10 : y(n)=0.012*x(n-1)+0.0517*x(n-2)+0.1164*x(n-3)+0.1918*x(n-4)+0.2563*x(n-
5)+0.1918*x(n-6)+0.1164*x(n-7)+0.0517*x(n-8)+0.012*x(n-9)
Equiripple orde 3 : y(n)=0.2907*x(0)+0.0844*x(n-1)+0.0844*x(n-2)+0.2907*x(n-3)
Equiripple orde 10 : y(n)=-0.1355*x(0)+ 0.1975*x(n-1)+0.1781*x(n-2)+0.1908*x(n-3)+0.2075*x(n-
4)+0.2141*x(n-5)+0.2075*x(n-6)+0.1908*x(n-7)+0.1781*x(n-8)+0.1975*x(n-9) -0.1355*x(n-10)
Transfer Function IIR
Butterworth Orde 1 Butterworth Orde 10

Butterworth Orde 2
Chebyshev 1 Orde 1

Chebyshev 1 Orde 10

Chebyshev 1 Orde 2
Chebyshev 2 Orde 1

Chebyshev 2 Orde 10

Chebyshev 2 Orde 2
Elliptic 1.08 Order 1

Elliptic 1.08 Order 10

Elliptic 1.08 Order 2


Discussion
Conclusion
1. Efektifitas filter dilihat dari besarnya nilai SNR
2. Semakin besar nilai SNR, maka filter semakin baik
3. Berdasarkan jurnal, filter terbaik adalah FIR Equiripple 3rd order dan IIR Elliptic
1st order
4. Berdasarkan percobaan, filter terbaik adalah FIR Equiripple 3rd order dan IIR
Elliptic 2nd order
THANK YOU

Vous aimerez peut-être aussi