Date of Award

Spring 2017

Degree Name

Doctor of Philosophy

Department

Electrical and Computer Engineering

First Advisor

Dr. Ikhlas M. Abdel-Qader

Second Advisor

Dr. Bradely Bazuin

Third Advisor

Dr. Azim Houshyar

Keywords

Multiprocessing, real time, solar panels, defect detections, vision based system, condition monitoring

Abstract

Enabling an algorithm to be executed in parallel on a multicore or a multiprocessor system has become a necessity for many real time applications. Multicore systems are widely used to improve performance and satisfy time and power demands. In this dissertation, solar energy, which has proved itself as the future clean source of energy, is also considered in real time. Optimum utilization of this energy propelled research efforts into many directions of the solar system components. However, while real time operations of Photovoltaic (PV) systems occur without any supervisory mechanisms, many internal and/or external obstacles can occur and hinder a system’s efficiency. To address fault detection in solar systems and thus provide a safer and more time efficient inspection in real time, we are proposing using videos for real time inspection and fault detection for the solar panel.

To monitor the conditions of the solar system and issue an alert when a faulty condition is detected, an integrated multicore CPU system using real time recording and analyzing of thermal and photographic videos has been developed. The system using a multiprocessing module in Python and under a multicore CPU system, inputs thermal and photographic videos into different segments and executes detection algorithms in parallel. Each segment should be processed via a separate thread. Several Pattern Recognition algorithms are investigated for real time fault detection suitability.

Two cameras are used to capture the scene of the solar panels simultaneously while mounted on a drone. The FLIR Vue Pro thermal camera was used for thermal video recording with (NTSC) frame rate, and with a resolution of 336x256 pixels. This resolution is high enough to show an accurate thermal resolution from the solar panels. For visual images, a GoPro Hero 4 Black photographic camera was used in the system; the camera has effective photo resolution of 12.0 MP, and a max video resolution of 3840x2160. These two cameras are mounted on the Yuneec Typhoon Q500 quadcopter. The recorded videos are streamed into the ground workstation where they are processed using the Python 2.7- IDE for Eclipse (Luna Service Release 2 (4.4.2)).

To validate our real time proposed system, we used a mobile solar system that was constructed primarily for this project in the Digital Image and Signal Processing Laboratory (DISPLAY) at Western Michigan University (WMU). This system is composed of two panels of SUNIVA OPTIMUS 60 Cell modules (Model OPT285-60-4-1B0); each panel is rated at 285W. The proposed system, as demonstrated by the results, has the following contributions: 1) Using the multiprocessing module in Python and the thermal and photographic video processing on multicore CPU shows execution time improvement and processor performance enhancements; the average improvement for the processing time of the detection algorithms for thermal and photographic videos was 3.1 times using 2 processes, and 6.3 times using 4 processes; and 2) a multicore real time system for the analysis of thermal and photographic videos, drone mounted, provides the capability to accurately detect defects in the solar panels and give location information in terms of panel location by longitude and latitude.

Access Setting

Dissertation-Open Access

Download

Share

COinS