对结果的影响变小即可。H：加权最小二乘算法和递推算法程序
clearclcNp400u1001101011fori10Np1uixorui4ui9e
dforj10Np1ifmodj20yj1
felseyj0e
duixoruiyje
dvra
d
3400L400z1L2fori3400zi15zi107zi21ui105ui2vie
dHzeros4004fori3400Hi1zi1Hi2zi2Hi3ui1Hi4ui2e
dLambdaeye400fori1400forj1400ifijLambdaij100ielseLambdaij0e
de
de
dtheta1i
vHLambdaHHLambdazi1400subplot121plotitheta11ritheta12kitheta13githeta14btitle惠一龙，加权一次最小二乘算法P106eye4theta20001eye41Ieye4hHfori3400KPhiHiPhi1Lambdaii1PIKHiPtheta2theta2KziHitheta2e
dtheta2
fi1400subplot122plotitheta21ritheta22kitheta23githeta24btitle惠一龙，加权递推算法
I：结果示意图：
惠一龙，加权一次最小二乘算法1惠一龙，加权递推算法1
05
05
0
0
05
05
1
1
15
15
2
0
100
200
300
400
2
0
100
200
300
400
J：结果分析：这个也相对容易一些，只是给予新数据较高的权值，即给予新数据较多的信任度。逐渐忽略老数据。K：偏差补偿最小二乘算法程序
clearclcNp480u1001101011fori10Np1uixorui4ui9e
dforj10Np1
fifmodj20yj1elseyj0e
duixoruiyje
dL480vra
d
1Lzzeros1480fork3480zk15zk107zk2uk105uk2vke
dtheta00001000100010001thetac00001000100010001P0106eye4thetatheta0zeros4479thetacthetac0zeros4479J0D1000010000000000fori3480Hzi1zi2ui1ui2JJziHtheta021HP0HKP0Hi
vHP0H1theta1theta0KziHtheta0P1P0KHP0theta0theta1thetaitheta1sigmaJk1thetac0Dtheta1thetac1thetac0isigmaP1Dthetac0thetac0thetac1thetacithetac1P0P1e
di1480plotitheta1ritheta2kitheta3githeta4btitle惠一龙，偏差补偿最小二乘算法thetathetaL
L：结果示意图
f惠一龙，偏差补偿最小二乘算法5
4
3
2
1
0
1
20
50
100
150
200
250
300
350
400
450
500
M：做这个程序的时候，还是遇到了些困难。比如递推的顺序，一些初值的赋予，还有矩阵的维数等问题。相对而言，偏差补偿估计更接近于真值。N：限定记忆算法程序
clearclcNp480u1001101011fori10Np1uixorui4ui9e
dforj10Np1ifmodj20yj1elseyj0e
duixoruiyje
dvra
d
3400z1398
ffori3400zi15zi107zr