부산대학교 도서관

The article proposes an adaptive algorithm that generates all object signals, including those for which measurements are not performed due to the difficulties associated with online measurements. The algorithm is modeled on the idea of the Kalman filter using its equation; however, the selection of gains is optimized differently, i.e. the constant values depend on the adopted ranges of adaptation errors. Moreover, the knowledge of the statistics of all noise signals is not imposed and there is no linearity constraint. This approach allowed us to reduce the complexity of calculations. This algorithm can be used in real-time systems to generate signals of objects described by non-linear differential equations and it is universal, which allows it to be used for various objects. In the conducted research, on the example of a biochemically contaminated river, only easily measurable signals were used to generate the object signals, and in addition, in the case of the absence of some measurements, the functioning of the algorithm did not destabilize. [ABSTRACT FROM AUTHOR]