Proportional-integral-differential Neural Network Based Sliding-mode Controller for Modular Multi-level High-voltage DC Converter of Offshore Wind Power

Abstract

This article presents an improved sliding-mode control method for a modular multi-level high-voltage DC converter. It merges the merits of the proportional-integral-differential neural network and can solve the chattering problem that exists in conventional sliding-mode control on-line. The reaching law parameters of sliding-mode control can be adjusted by the proportional-integral-differential neural network without the previously needed of off-line learning. The Lyapunov function is chosen as the energy function for real-time training of the proportional-integral-differential neural network. In addition, the stability of the control system is carefully studied, and the global optimal solution is achieved. The MATLAB (The MathWorks, Natick, Massachusetts, USA) simulation results show that the proposed method can make the system globally stable, can achieve stronger robustness under system disturbance, and be applied easily to digital signal processor based modular multi-level converter high-voltage DC control systems.

Keywords:

• proportional-integral-differential neural network
• sliding-mode variable-structure control
• modular multi-level converter
• high-voltage DC transmission
• offshore wind power
• direct power control