Fault Detection and Diagnosis of Grid Connected PV Systems Using Bayesian Neural Network | Chapter 6 | Current Approaches in Engineering Research and Technology Vol. 6

 

At present, artificial neural networks are massively and successfully used for fault diagnosis of grid-connected PV (GCPV) systems thanks to their capabilities of learning and generalization. Solutions can be found with almost all types of Artificial Neural Networks in fault detection and diagnosis of PV systems. In fact, the use of ANN permits the development of intelligent algorithms that can learn all the PV systems working status and then be able to detect and identify possible malfunctions status and causes independently of the ANNs type. In this study, the architecture of both ANNs is multilayer perceptron (MLP), the transfer function used in this step is the sigmoid function. The approach required the Bayesian Regulation algorithm for ANN training. However, each type of ANN owns its proper characteristics from point of view of its accuracy, efficiency, reliability as well as its response of time in term of faults detection and diagnosis compared to others. In this article, fault detection and diagnosis of a small GCPV array based on an experimental setup using Bayesian Neural Network is presented. This type of Artificial Neural Network can secure the identification and localization of the greatest recurring cases in PV generators in particular short-circuit, open-circuit and healthy cases. This study requires four input data such as irradiation, cell temperature, current and voltage of maximum power point. Results show that this study proved an excellent level of accuracy displaying (99.88%) in terms of detection and localization of treated faults with a global response time equal to 16 min22. Fault detection and diagnosis of PV systems is necessary not only to increase system power system reliability but also to reduce the operating costs of PV plans.

 

Author(s) Details

C. Kara Mostefa Khelil
Electrical Engineering Department, Khemis Miliana University, Ain Defla, Algeria and SET Laboratory, Electronics Department, Blida 1 University, BP 270 Blida, Algeria.

 

B. Amrouche
SET Laboratory, Electronics Department, Blida 1 University, BP 270 Blida, Algeria and Renewable Energies Department, Blida 1 University, BP 270 Blida, Algeria.

 

K. Kara
SET Laboratory, Electronics Department, Blida 1 University, BP 270 Blida, Algeria.

 

Please see the book here:- https://doi.org/10.9734/bpi/caert/v6/1207