Implementation of artificial neural network based controller for a five-level converter based UPQC

• Main Authors: Sudheer Vinnakoti, Venkata Reddy Kota
• Format: Article
• Language: English
• Published: Elsevier 2018-09-01
• Series: Alexandria Engineering Journal
• Online Access: http://www.sciencedirect.com/science/article/pii/S1110016817301205

To reduce the cost and complexity of the system, ANN controller is proposed for a five-level converter based UPQC which eliminates mathematical operations such as a-b-c to d-q-0 transformation and complex controllers such as DSP’s and FPGA’s,. This paper proposes a five-level DCC based UPQC with ANN based controller and its performance is tested with nonlinear unbalanced loads and harmonic supply voltage. In addition, voltage related issues such as sag and swell are also considered. ANN control scheme trained with Levenberg-Marquardt backpropagation algorithm is used in this paper for effective generation of reference signals and also for maintaining desired dc link capacitor voltage. Simulations are carried out in Matlab/Simulink software with two-level UPQC, and three-level and five-level diode clamped converter based UPQC using SRF based control and ANN control schemes. The results showed better performance with the proposed concept and are discussed in this paper. The response of dc link capacitor voltage with two-level, three-level and five-level converter based UPQC and its effect on supply currents are discussed. Comparison of %THD in load voltage and supply current with two-level, three-level and five-level converter based UPQC using SRF based control and ANN control schemes is presented to show the superiority of the proposed controller. Keywords: Multilevel Inverter UPQC (MLI-UPQC), Diode clamped converter (DCC), % Total Harmonic Distortion (%THD), Artificial Neural Network (ANN), Levenberg-Marquardt backpropagation (LMBP)