الفهرس 
Partial Shading Effects on PV Systems
ACCURATE FAULT DIAGNOSIS TECHNIQUE FOR SERIES-PARALLEL PV ARRAYOnly 14 pages are availabe for public view
 |  | from | 98 |  |  |
| | | from | 98 | | |
Abstract
Partial shading on PV arrays leads to a significant power loss and complications in the maximum power point tracking algorithm limiting the full utilization of the PV array power.
Previous solutions to mitigate partial shading effects either requires complex computations or expensive equipment or both.
These limitations motivate the research point of this thesis, aiming to 1) increase the energy yield at partial shading conditions to the maximum possible limit, 2) spot the existence of partial shading conditions without the need for extra measurement devices or complicated technique, 3) introduce a technique to discriminate between normal conditions, faulty conditions, and partial shading conditions to prevent misleading tripping, 4) reach the global maximum PowerPoint without confusing it with local maxima points with a simple algorithm with the need for complicated controllers.
Firstly, a novel PV modeling approach was introduced based on a Hybrid Flower Pollination Algorithm with Clonal selection Algorithm (HFPA-CSA) to emulate the performance of the used PV array accurately and serve as a simulated reference model for further calculations.
Secondly, to deal with the power loss at partial shading conditions an innovative integer-search based reconfiguration technique was introduced. The technique alters the PV configuration at shading conditions to harvest the maximum possible energy yield and decrease the number of local peaks to the minimum. The integer-based search was compared with the previously used binary-based search, showing that the proposed technique always harvests the highest power.
Thirdly, an accurate fault diagnosis technique was introduced with the ability to diagnosis the faulty sections in the PV array successfully and discriminate the partial shading conditions from the healthy or faulty conditions based on current and voltage measurement along with an Artificial Neural Network (ANN) used for the discriminating the partial shading conditions only thus reducing the training process.
Fourthly, Artificial Vision (AV) was used effectively to spot the presence of partial shading on the
PV array by capturing the PV array image then comparing the average pixel intensity for all the
PV modules with each other to detect any shading phenomena, then the shaded area in each module is detected using AV manipulation processes, finally, the Camera Response Function (CRF) for the used camera was deduced and the obtained relation was used to estimate the incident irradiance on each PV module, then a simulated model utilized the obtained irradiance levels and the PV model was used to plot the P-V curve for the PV understudy. The obtained Global Maximum Power Point (GMPP) was compared with the real GMPP using a practical installation where results show the accuracy of the detected GMPP.
The proposed techniques rely on simple computations procedures including the PV modelling and the CRF estimation which are a one-time procedure done at the beginning of system installation and used afterward.
The devices needed for detecting shading conditions and GMPP are limited to a surveillance camera and current and voltage sensors which don’t rise the system cost or complexity.
The proposed techniques are simple and cheap yet results show that it accurately detects partial shading conditions, increases the energy yield and locates the GMPP precisely.