Forecasting of Photovoltaic Power Yield Using Dynamic Neural Networks Naji Al-Messabi*, Yun Li*, Ibrahim El-Amin**, Cindy Goh* *School of Engineering, University of Glasgow, Rankine building, Glasgow, U.K., n.al-messabi.1@research.gla.ac.uk **Electrical Engineering Department, King Fahd University of Petroleum and Minerals, Dhahran, Saudi Arabia, imelamin@kfupm.edu.sa Abstract— The importance of predicting the output power of Photovoltaic (PV) plants is crucial in modern power system applications. Predicting the power yield of a PV generation system helps the process of dispatching the power into a grid with improved efficiency in generation planning and operation. This work proposes the use of intelligent tools to forecast the real power output of PV units. These tools primarily comprise dynamic neural networks which are capable of time-series predictions with good reliability. This paper begins with a brief review of various methods of forecasting solar power reported in literature. Results of preliminary work on a 5kW PV panel at King Fahd University of Petroleum and Minerals, Dhahran, Saudi Arabia, is presented. Focused Time Delay and Distributed Time Delay Neural Networks were used as a forecasting tool for this study and their performance was compared with each other. Keywords-Irradiance; Time-series forecasting; Dynamic Neural Networks I. INTRODUCTION There has been a concern on how to dispatch renewable sources to electric grids. This includes controlling the PV system in the best way that guarantees safety, security, and economical operation of energy dispatch. Thus it is inevitable to explore ways of forecasting the power yield of these sources to support optimal generation planning studies. PV plants usually consist of PV arrays connected to inverters that change the DC output to AC and perform Maximum Power Point Tracking (MPPT) function. The AC power is then sent to power transformers that will step up the voltage and connect the PV system to either the distribution or transmission power network. The overall PV generation plant configuration is shown in Fig. 1. Figure 1. Overview of PV generation plant The output power produced by PV arrays usually varies with sunshine or irradiance. The irradiance is zero at night and starts to increase gradually during the day and reaches its maximum level in the afternoon and then decreases back to zero again. The out power exhibits similar non-linear behavior. Clouds and heavy dust might cause a sudden fall in the yield of PV plants. Fig. 2 shows typical output of PV plants taken on hourly basis. This data was for a 10MW PV plant, Fig. 3, in MASDAR city, Abu Dhabi, U.A.E, connected to power distribution network at 11kV level. Figure 2. Power generated from 10MW PV plant on two consecutive days (1- 2/1/2010, Source: Abu Dhabi Distribution Company (ADDC)) Figure 3. Masdar 10MW PV plant, Abu Dhabi, U.A.E. The output of PV plant shows non-linear behavior which varies from day to day. Data from smaller scale solar systems with smaller recording time intervals of 10 minutes exhibits output of more vivid nonlinearity. This is shown in Fig. 4. The output shown is that of a 5kW PV array, Fig. 5, installed at King Fahd University of Petroleum and Minerals (KFUPM), Dhahran, Saudi Arabia, connected to small local loads. This fact encouraged researchers to propose nonlinear predictors 10MW PV plant output 0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 1 3 5 7 9 11 13 15 17 19 21 23 Hour Power (MW) Day 1 Day2 U.S. Government work not protected by U.S. copyright WCCI 2012 IEEE World Congress on Computational Intelligence June, 10-15, 2012 - Brisbane, Australia IJCNN