Reducing the computational demands of continually online trained artificial neural networks for system identification and control of fast processes

This paper describes many of the generic factors which influence the computational demands of continually online trained backpropagation artificial neural networks (ANNs) used to identify and control fast processes. The limitations of even parallel hardware in meeting these demands is discussed. An adaptive online trained artificial neural network induction machine stator current controller is considered as a typical fast process. Various modifications are made to the ANN structure to lower computational demands and increase ANN parallelism. The effects of these modifications on the overall controller stability and performance are illustrated by means of simulation results.<<ETX>>

[1]  George Cybenko,et al.  Approximation by superpositions of a sigmoidal function , 1992, Math. Control. Signals Syst..

[2]  Ronald G. Harley,et al.  Identification and control of induction machines using artificial neural networks , 1993 .

[3]  Ronald G. Harley,et al.  Implementation of a neural network to adaptively identify and control VSI fed induction motor stator currents , 1994, Proceedings of 1994 IEEE Industry Applications Society Annual Meeting.

[4]  R.G. Harley,et al.  Optimising multirate motion control algorithms on parallel processors by static scheduling , 1994, Proceedings of 1994 IEEE Industry Applications Society Annual Meeting.

[5]  Ronald G. Harley,et al.  Implementing field oriented control of a voltage fed current regulated induction motor on a single transputer , 1994, Proceedings of 1994 IEEE Industry Applications Society Annual Meeting.