A New Algorithm for Long-Term Estimation Based on AR Model
2014 ◽
Vol 614
◽
pp. 440-443
◽
Keyword(s):
Ar Model
◽
Traditional estimation methods have poor performance for long-term data forecast. Using Wiener model to estimate, power spectral density of the input signal, and cross-spectral density of the input and output signals are needed, that are difficult to obtain. And the large amount of calculation is needed using Wiener model. Using AR model and Kalman model, estimated results tend to mean of the training set while the estimated distance increases. For these cases, a new algorithm for long-term estimation based on AR model, named sampling AR model, is presented. Grouping the training set and using a different group of the training set to estimate each value. Sampling AR model improves the accuracy of long-term estimation.
1974 ◽
Vol 96
(2)
◽
pp. 676-679
◽
Keyword(s):
2014 ◽
Vol 224
◽
pp. 118-123
◽