Real-time Neural Sliding Mode Linearization Control for a Doubly Fed Induction Generator under Disturbances
More details
Hide details
1
TecNM Chihuahua, Av. Tecnologico 2909, Tecnológico, Chihuahua, México
2
Université de Lorraine, LCOMS, 57000 Metz, France
3
Department of Electrical Engineering, Cinvestav Guadalajara, Zapopan, Mexico
Power Electronics and Drives 2024;9(Special Section - Modern Control Methods of Electrical Drives ):238-256
KEYWORDS
TOPICS
ABSTRACT
This paper presents an experimental implementation of a Neural Sliding Mode Linearization approach for the control of a double-fed
induction generator connected to an infinite bus via transmission lines. The rotor windings are connected to the grid via a back-to-back
converter, while the stator windings are directly coupled to the network. The chosen control scheme is applied to obtain the required
stator power trajectories by controlling the rotor currents and to track the desired values of the DC-link output voltage and the grid power
factor. This controller is based on a neural identifier trained online using an Extended Kalman Filter. Based on such identifier, an adequate
model is obtained, which is used for synthesizing the required controllers. The proposed control scheme is experimentally verified on
1/4 HP DFIG prototype considering normal and abnormal grid conditions. In addition, maximum power extraction from a random wind
profile is tested in the presence of different grid scenarios. Moreover, a comparison with conventional control schemes is performed.
The obtained results illustrate the capability of the proposed control scheme to achieve active power, reactive power, and DC voltage
desired trajectories tracking and to operate the wind power system even in the presence of parameter variation and grid disturbances which helps to ensure the stability of the system and improve generated power quality.