PONGSAK TAMKEAW et al: SOILING LEVEL IDENTIFICATION OF SOLAR PV PANEL FOR CLEANING . . DOI 10.5013/IJSSST.a.20.03.06 6.1 ISSN: 1473-804x online, 1473-8031 print Soiling Level Identification of Solar PV Panels for Cleaning Planning Pongsak Tamkeaw, Prajuab Inrawong, Kaan Kerdchuen Research Unit of Electric Energy Innovation, Department of Electrical Engineering Faculty of Engineering and Architecture, Rajamangala University of Technology Isan 744 Suranarai Rd., Nai-Muang, Muang, Nakhon Ratchasima, 30000, Thailand Email: eee36413@hotmail.com, prajuab@rmuti.ac.th, kaan.ke@rmuti.ac.th Abstract - This paper proposes the combination of image processing and neural network techniques to identify the soiling level of solar PhotoVoltaic (PV) panel in order to plan the cleaning of these panels. The images of cleaned and soiled PVs are taken into the image processing process for identifying HSV (Hue-Saturation-Value is an alternative representations of the RGB color model [8]). Then, the HSV histogram of different PV panel soiling images can determine the relation with soiling level or Soiling Ratio (SR) by using a neural network. The results of solar PV soiling ratio identification are verified by short and open circuits of solar PV panel testing. Keywords – soiling, soiling ratio, image processing, neural networks I. INTRODUCTION In the power generation industry of solar PV generation, the solar PV panel soiling is important for power generation output. The different location of solar PV installation is soiled by different dust. The level of solar PV soiling needs to clean for increasing the power generation output. Solar PV panel cleaning cost should not be higher than the revenue from energy selling. Thus, if we can identify the soiling level of PV panel, the cost balancing between the cleaning cost and energy selling cost should be compromised. In [1], the solar PV panel soiling level is related with short circuit current and open voltage. This short circuit current is also related with the output power. In [2], the regression model of soiled PV panel is developed for prediction the soiling losses in the PV panel. The experimental data of a wide range irradiation are used to train in neural network model for prediction the power output of a PV panel. In Norway [3], the effect of soiling on PV panel is experimented relates with the weight of dust. The soiling loss is estimated relates to the weight measurement of dust. In practical solar PV soiling cleaning, the reviewed techniques are not conveniently cleaning. If we have the conveniently technique to estimate the soiling level, the system operator will can medially decide for worthy cleaning. Thus, in this paper, machine vision techniques for identification of solar PV soiling level is implemented. The images histogram and neural network are used to process the soiling level. The short circuit and power output are used to verify the soiling loss relation with soiling level. II. SOILED PV PANEL AND GENERATED POWER The solar PV panel characteristic is related with solar irradiance. The high solar irradiance falling called solar power on solar PV panel makes high generated power output. If solar PV panel is soiled, the net solar power falling on the cell of PV is also reduced. (a) P-V characteristics (b) I-V characteristics Figure 1. MPP dropping and power reduction caused by the soiling of the KC200GT panel [4]