Photovoltaic Power Forecasting Based on Artificial Neural Network and Meteorological Data

Index Terms--Photovoltaic Power Forecasting; Artificial Neural Network Method; BP Network; Aerosol Index; Maximum Absolute Prediction Error.
I. INTRODUCTION
1
Photovoltaic Power Forecasting Based on Artificial Neural Network and Meteorological Data
Jiahao Koபைடு நூலகம், Jun Liu*, Qifan Li, Wanliang Fang
Dept. Electric Power Engineering School of Electrical Engineering, Xi'an Jiaotong University Xi'an, China *Corresponding author: eeliujun@
1
Zhenhuan Chen, Linlin Liu, and Tieying Guan
Dispatching and Communication Center Gansu Electric Power Corporation Lanzhou, China yangcx@ Various PV power output prediction methods have been proposed up to present. Multiple linear regression [3][4] is used to study the solar power output characteristics, combining with weather data and solar radiation data. However, this method requires large amount of data samples, and the fitting results are sensitive to pathological data. Another method based on Markov chain [5]-[7] also needs a large number of raw data, but it shows higher accuracy even if some weather data are omitted. Support Vector Machine (SVM) is also applied to predict PV power outputs [8][9]. However, classical SVM algorithms are more suitable for two-classification cases, while PV power forecasting is a typical multi-classification problem. The most promising method for PV forecasting is the artificial neural network method [10][11], especially Back Propagation(BP) neural network. BP neural network has the advantages of complex nonlinear systems simulation, good approximation performance, strong learning ability and fault data tolerance. However, it also has some inherent defects, such as slow convergence rate, easy to fall into local minimum value and cannot get the global optimal solution, some improvements have been made on BP network to enhance better convergence[12] -[15]. While predicting PV power outputs, few researches have taken aerosols into consideration, which can be induced by desert dust, biomass burning, volcano smoke, power plant emission, and so on. In fact, there is a strong correlation between total solar radiation and aerosol index [16]-[18]. Typically, the higher the aerosol optical depth is, the greater the attenuation of the solar power reaching the surface of the earth is. Therefore, aerosols influence the output of the PV system indirectly. In this paper, a novel PV forecasting model is proposed, based on multiple meteorological data including aerosol index values. Firstly BP neural network is trained by historical meteorological data and hourly average output power data, and then weather data of the objective day are used as the input variables of BP neural network to predict hourly power outputs of PV systems. The efficiency of the proposed method is validated by analyzing the mean absolute percentage error between predicted values and measured values. II. FORECASTING MODEL FOR PV OUTPUT POWER Since there are various factors affecting PV power outputs, it is difficult to figure out the power outputs with a fixed
T
O ensure reliable operation and economic dispatch of modern power systems, generation schedules are made for real-time, one-day-ahead, weekly, monthly and yearly power markets. Hence it is essential to predict power system load curves and power outputs of renewable power plants, such as wind farms and Photovoltaic (PV) power stations. As it is well-known that the outputs of solar power stations have high variations subject to different weather conditions, researches on PV power forecasting have been increasing considerably in recent years. There are two commonly used objectives for power forecasting at present: one is to predict the environmental parameters relevant to the PV system, such as solar radiation [1], then calculate the active power outputs with solar radiation, ambient temperature and other parameters using predefined mathematical models. The other is to predict the active power outputs of the PV systems directly [2]. Since solar radiation data are difficult to predict on an hourly basis, the active power outputs of the PV system are predicted directly in this paper, based on the historical power outputs data and weather forecast data of the objective day.
Abstract--Due to the intermittency and randomness of solar Photovoltaic (PV) power outputs, it is necessary to find a precise method for PV power forecasting. However, conventional methods, using only temperature, humidity and wind speed data, failed to obtain high accuracy when used to predict PV power outputs under extreme weather conditions. Aerosol index which indicates particulate matter in the atmosphere has a strong correlation with PV generated energy. This paper proposes a novel photovoltaic power forecasting model considering aerosol index data as an additional input. Based on weather classification and back propagation artificial neural network approaches, the estimated results of the forecasting model show good coincidence with the measurement data. And the proposed model is able to improve the prediction accuracy of conventional methods using artificial neural network.