Analysis and application of minimum variance discrete time system identification

S. Kotob, H. Kaufman
1976 1976 IEEE Conference on Decision and Control including the 15th Symposium on Adaptive Processes  
ptirnal. design of an on-.line state and parameter estimation algorithm results in a complex structure not readily feasible for adaptive control purposes and/or for implementation in a typical process control or flight computer. Suboptimal designs are often used in practice and despite the ease of implementing these procedures, p obl.ems such as di ,errence and/or inaccuracies are not uncommon. An on-line minimum variance parameter identifier has therefore been developed which embodies both
more » ... racy and comp utati-nal efficiency. The nL,.r formu:..tion, results in a linear estimation problem with both additive and ,C multi p licative noise. The resulting filter is shorn to utilize both the ccvariance of the p arameter vector itself and the covariance of the error in identification. A bias reduction scheme can be used if desired to yield asymptotically unbiased estimates. It is proven that the identification filter is mean square convergent and mean s q uare consistent. The 10 p arameter identification scheme is then used to construct a stable state and parameter estimation algorithm.
doi:10.1109/cdc.1976.267859 fatcat:flpazugg55d4dggxbzm2obtwfi