Abstract—In this paper, autoregressive moving average (ARMA) has modeling for solar irradiation forecast by combining two types of parameter estimation methods, Forgetting Factor (FF) and Kalman Filter (KF). For this purpose, the geographical location of length, width and average height than 76.75 e, 31.75 N and 1130.3 meters were used. Parameter with regard to the mean absolute error (MAE), root mean square error (RMSE), mean square error (MSE) and R 2 estimator is compared. The result shows that the KF consists of high convergence rate to solve complex problems. Index Terms—Solar irradiation, autoregressive moving average, forgetting factor, Kalman filter. I. INTRODUCTION Today, researchers are focusing on the use of renewable energy sources, this is due to the fact that it will reduce air pollution and meet the current and future energy needs. However, the exact number of energy from these sources is unknown. This is due to the fact that the power generation is dependent on the number of parameters (for example, wind speed, insolation, waves and water plants, among others), which must be dissolved to leave this form of energy. In the production of renewable energy, is one of the most important resources of the incident solar radiation. It is depending on the geography of the earth (height, length, width and surface angle), Sun-Earth position hour, azimuth and inclination angle [1]. In India incident solar radiation varies from 44 to 77 per cent per kWh m2 per day, the other varies from 68 to 82 percent under a clear sky days per year [2]. It is also affected by atmosphere and weather phenomena such as air mass, suspended particles, water vapor, aerosols, clouds, humidity, temperature, SO2, soot, CO2, aerosol Optical depth [3]. Therefore sunshine forecast for energy producers is necessary. Furthermore, it is very important issue for larger solar power stations Grid intergraded. In addition, helps forecast the consumer, energy planning and management company, which is advantageous from both an economic and technical point of view [4]. In this regard, several methods/models have been proposed to be radiation at the ground, the physics-based models, Manuscript received September 28, 2014; revised July 19, 2015. Yashwant Kashyap and Satyanarayan Patel are with the School of Engineering, Indian Institute of Technology Mandi, 175005 India (tel.: 91-9805-911710; e-mail: [email protected], [email protected]). Ankit Bansal is with the Department of Mechanical and Industrial Engineering, 233 MIED, IIT Roorkee, 247667 India (e-mail: [email protected]). Anil K. Sao is with the School of Computing and Electrical Engineering, Indian Institute of Technology Mandi, India (e-mail: [email protected]). moving average, Classic Car regression, auto regression and moving average, Markov Chain and Fuzzy logic includes predict, etc. Furthermore, an adaptive process such as Time Delay Neural Network (TDNN) has also shown that a reliable method for predicting the future evolution of the time series [5]. Other are his key questions radiation forecast weather and clouds. Therefore, it is very difficult to do a straight build forecast forecast models in one 24-hour cycle. One possible solution is the time series based model the exact radiation can evaluate the ground. Therefore, the solar radiation sequence is treated as a time series and mathematical models fit the random process underlying to predict the next values [6]. In the research community, Autoregressive Moving Average (ARMA) methods are widely used and popular time series models compared to other models (as mentioned above) [7]. The ARMA model is able to extract many regions, useful statistical properties and can easily take on the well-known box-Jenkins method [8]. In addition, these models are very flexible; therefore they can be used in various types of time series with different orders. Finally, it offers regular pervasive in individual phases (identification, estimation and diagnostic check) for a suitable model. In the ARMA model of one of the greatest difficulties need is the enormous amount of data. Moreover, this method requires an excessive agreement of knowledge and although it creates often acceptable results the individual results are determined on the level of research knowledge [9]. Recently, ARMA models based several studies have done in many areas, the current flow, supply chain management, business, earthquakes, land use, sales, products, includes transportation and weather forecasts. However, there is a scarcity of the use of this model in energy applications. In this paper ARMA method is used to use the solar radiation that this connection be recursive least squares-based parameter estimation techniques two types of processes (FF KF) .In both methods predict cast together. In the first step aroused the method to estimate the model parameters and the best results with the measured data. The next step is used to define bias and mean error of the model that leads to accurate predictions. II. METHODOLOGY AND DATA A. Data Collection The average value of extraterrestrial radiance is considered nearly about to 1360 Wm -2 . It consists of two parameters: Diffuse Horizontal irradiance (DHI) and direct Normal Irradiance (DNI). These are used to calculate Global Horizontal Irradiance (GHI) as: Comparative Study of Parameter Estimation Methods for Solar Irradiation Forecasting Yashwant Kashyap, Ankit Bansal, Anil K. Sao, and Satyanarayan Patel Journal of Clean Energy Technologies, Vol. 4, No. 3, May 2016 192 DOI: 10.7763/JOCET.2016.V4.278
5
Embed
Comparative Study of Parameter Estimation Methods for …Yashwant Kashyap and Satyanarayan Patel are with the School of Engineering, Indian Institute of Technology Mandi, 175005 India
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Abstract—In this paper, autoregressive moving average
(ARMA) has modeling for solar irradiation forecast by
combining two types of parameter estimation methods,
Forgetting Factor (FF) and Kalman Filter (KF). For this
purpose, the geographical location of length, width and average
height than 76.75 e, 31.75 N and 1130.3 meters were used.
Parameter with regard to the mean absolute error (MAE), root
mean square error (RMSE), mean square error (MSE) and R2
estimator is compared. The result shows that the KF consists of
high convergence rate to solve complex problems.
Index Terms—Solar irradiation, autoregressive moving
average, forgetting factor, Kalman filter.
I. INTRODUCTION
Today, researchers are focusing on the use of renewable
energy sources, this is due to the fact that it will reduce air
pollution and meet the current and future energy needs.
However, the exact number of energy from these sources is
unknown. This is due to the fact that the power generation is
dependent on the number of parameters (for example, wind
speed, insolation, waves and water plants, among others),
which must be dissolved to leave this form of energy. In the
production of renewable energy, is one of the most important
resources of the incident solar radiation. It is depending on the
geography of the earth (height, length, width and surface
angle), Sun-Earth position hour, azimuth and inclination
angle [1]. In India incident solar radiation varies from 44 to 77
per cent per kWh m2 per day, the other varies from 68 to 82
percent under a clear sky days per year [2]. It is also affected
by atmosphere and weather phenomena such as air mass,
suspended particles, water vapor, aerosols, clouds, humidity,