Skip to main content
. 2025 Aug 21;16(8):963. doi: 10.3390/mi16080963
Step No. Filtering Steps The Equations Corresponding to the Filtering Steps
a Set initial value based on the training data L1,P1,r1,R1,q1,Q1,Γ1
b Calculation of weighting factors dk=(1b)/(1bk+1)
c One-step forecast L^k,k1=Φk,k1L^k1+q^k1
d One-step mean square error estimation Pk,k1=Φk,k1Pk1Φk,k1T+Q^k1
e Filter residual variance estimation r^k=(1dk)r^k1+dk(ZkHkL^k,k1)
νk=ZkHkL^k,k1r^k
f Measurement noise estimation R^k=(11k1)R^k1+12(k1)(ZkZk1)2
g Calculating filter gain Kk=Pk,k1HkT[HkPk,k1HkT+R^k]1
h State estimation L^k=L^k,k1+Kk[ZkHkL^k,k1]
i Mean square error estimation Pk=[IKkHk]Pk,k1
j Model accuracy discrimination R^kη[HkPk,k1HkT+R^k]
k Variance matching νkvkT>HkPk,k1HkT+R^k
l System noise estimate q^k=(1dk)q^k1+dkX^kΦk,k1X^k1
Q^k=(1dk)Q^k1+dk(KkνkνkTKkT+PkΦk,k1Pk1Φk,k1T)