The article is concerned with designing online identification algorithm for identification plunger model for  wicket gate of Kaplan turbine adaptive control system. Algorithm is based on generalised recurrent least square method and invariant embedding method. Computer modeling using designed algorithm and using real experimental data shows that algorithm is effective.  

Keywords: adaptive control system, diagnostic system, Kaplan turbine wicket gate, plunger, linearized models, identification, recurrent least square

The article is concerned with problems of modern electrohydraulic converters control systems used to adjust wicket gate of Kaplan turbine. Linearized model of wicket gate servomotor is considered. Algorithm for online identification of parameters of servomotor is considered. Algorithm is based on recurrent least squares method. Algorithm for adaptive PI-regulator tunning is based on online optimization of functional of generalised work (FGW). Model experiments using real data show reducing maximum control error of servomotor position from 20 % to 1 %. Also methods for servomotor state diagnostic are considered.

Keywords: adaptive control system, diagnostic system, Kaplan turbine wicket gate, linearized models, identification, recurrent least square, functional of generalised work