(3)计算aki?i?1,2,(4)计算e1(n)和b1(n):,k-1?：a11?K1=-0.9524e1(n)?e0(n)?K1b0(n-1)?x(n)?K1x(n-1)b1(n)?b0(n-1)?K1e0(n)?x(n-1)?K1x(n)代入具体的数据：e1(1)?x(1)?K1x(0)=1.0476,e1(2)?x(2)?K1x(1)=1.0952e1(3)?x(3)?K1x(2)=1.1428,e1(4)?x(4)?K1x(3)=1.1908b1(1)?x(0)?K1x(1)=-0.9048,b1(2)?x(1)?K1x(2)=-0.8572b1(3)?x(2)?K1x(3)=-0.8094,b1(4)?x(3)?K1x(4)=-0.76202(5)计算?12：　?12??1-K12??0=(1-0.95242)?11=10.0929（6）k?k?1,2???ep－1(n)bp-1(n-1)??N-1n?pN-1K2?-=-2??e1(n)b1(n-1)?n?222??e(n)?b??11(n-1)?n?244???e(n)?b(n)??2p-12p-1n?p2(1.0952?(-0.9048)+1.1428?(-0.8572)+1.1908?(-0.8094))=-(1.09522+1.14282+1.19082+0.90482+0.85722+0.80942)5.87==0.95756.13(7)计算?：　???1-K??=(1-0.9575)?10.0929=0.9397222222212(7)计算aki?i?1,2, a22?K2=0.9575,k-1?：(8)计算e2(n)和b2(n):e2(n)?e1(n)?K2b1(n-1)b2(n)?b1(n-1)?K2e1(n)41

代入具体的数据：e2(1)?e1(1)?K2b1(0),e2(2)?e1(2)?K2b1(1)=0.2744,e2(3)?e1(3)?K2b1(2)?0.3678,e2(4)?e1(4)?K2b1(3)?0.4612,b2(1)?b1(0)?K2e1(1),b2(2)?b1(1)?K2e1(2)=0.1439,b2(3)?b1(2)?K2e1(3)=0.2365,b2(4)?b1(3)?K2e1(4)=0.3308（9）k?k?1?3,N-12??4?ep－1(n)bp-1(n-1)??2Kn?p??e2(n)b2(n-1)?n?33?-?N-1=-224?n?p?ep-1(n)?bp-1(n)?????3?e22(n)?b22(n-1)?n?=-2(0.3678?0.1439+0.4612?0.2365)(0.36782+0.14392+0.46122+0.23652)=0.32390.4246=0.7628a33?K3?0.7628a32?a22?a33a21??0.4651a31?a21?a33a22??1.13462?1-a23?33??22?0.392942

?

求得的AR(3)模型为：x(n)?-1.1346x(n-1)-0.4651x(n-2)?0.7628x(n-3)?w(n)预测值为：?4?-1.1346?x3-0.4651?x2?0.7628?x1?wnx预测时，不考虑激励噪声的影响，有?4?-1.1346?5-0.4651?4?0.7628?3?-5.245xe0(n)e1(n)e2(n)e3(n)e3(n)x(n)z-1k1k2*k2k3*k3k1*z-1b0(n)b1(n)b2(n)b3(n)b3(n)

43