A novel neural PWM controller

• Published in: IEE conference publication pp. 375 - 379
• Main Authors: Dinu, A, Cirstea, M.N, McCormick, M
• Format: Conference Proceeding
• Language: English
• Published: London IEE 1998
• Subjects: Applied sciences; Artificial intelligence; Computer science; control theory; systems; Connectionism. Neural networks; Control system design and analysis; Control system synthesis; Control theory. Systems; Current control; Exact sciences and technology; Neurocontrol; Optimal control; Simulation; Simulation techniques; Software; neural PWM controller; inductance; noninductive load voltage; state-space observer; transistors; digital VHDL simulations; digital simulation; switching pattern; neural inductance estimator; VSI-PWM inverter; MATLAB simulations; electric current control; PWM invertors; neurocontrollers; predictive neural current control scheme; power systems; control system analysis computing; predictive control; observers; neural PWM generator; state-space methods; System architecture; System synthesis; Neural network; Computer simulation; Experimental study; Monitoring control system
• ISBN: 0852967098; 9780852967096
• ISSN: 0537-9989
• DOI: 10.1049/cp:19980666

This paper describes a novel approach of a predictive neural current control scheme for a VSI-PWM inverter used in power systems. The new structure is based on a simple state-space observer providing the advantage of a control which is independent of the load type. The observer is able to reconstruct the non-inductive load voltage while a neural PWM generator produces the switching pattern for the transistors in the inverter. The inductance in the load circuit is calculated online by a neural inductance estimator. Analogue MATLAB simulations and digital VHDL simulations have proved the satisfactory operation of the control system.