The use of the bi-orthogonal decomposition in a nonlinear controller for electrical drives

Learning algorithms are powerful tools to extract, from experimental data, relations that represent dynamics of complex systems. A direct application for these algorithms is to compensate nonlinear terms that affect dynamics for control proposes. This paper presents some characteristics of the bi-orthogonal decomposition used as a learning algorithm in an “intelligent” electrical drive controller. The authors underline the “memory and generalisation” capabilities of the biorthogonal decomposition in the learning process that uses a training set of experimental data. Moreover, an application of this learning algorithm to a DC motor electrical drive supplied by a DC-DC converter, with a coarse sliding mode position tracking controller, is also presented