Fatih Onur Hocaoglu

Missing wind data forecasting with adaptive neuro-fuzzy inference system

In any region, to begin generating electricity from wind energy, it is necessary to determine the 1-year distribution characteristics of wind speed. For this aim, a wind observation station must be constructed and 1-year wind speed and direction data must be collected. For determining the distribution characteristics, the collected data must be statistically analyzed. The continuity and reliability of the data are quite important for such studies on the days when possible faults can occur in any part of the observation unit or on days when, the system is on maintenance, it is not possible to record any data. In this study, it is assumed that the station had not worked at some randomly chosen days and that for these days no data could be recorded. The missing data are predicted using the data that were recorded before and after fault or maintenance by an adaptive neuro-fuzzy inference system (ANFIS). It is seen that ANFIS is successful for such a study.

The effect of missing wind speed data on wind power estimation

In this paper, the effect of possible missing data on wind power estimation is examined. One-month wind speed data obtained from wind and solar observation station which is constructed at Iki Eylul Campus of Anadolu University is used. A closed correlation is found between consecutive wind speed data that are collected for a period of 15 second. A very short time wind speed forecasting model is built by using two-input and one-output Adaptive Neuro Fuzzy Inference System (ANFIS). First, some randomly selected data from whole data are discarded. Second, 10%, 20% and 30% of all data which are randomly selected from a predefined interval (3-6 m/sec) are discarded and discarded data are forecasted. Finally, the data are fitted to Weibull distribution, Weibull distribution parameters are obtained and wind powers are estimated for all cases. The results show that the missing data has a significant effect on wind power estimation and must be taken into account in wind studies. Furthermore, it is concluded that ANFIS is a convenient tool for this kind of prediction.

Comparison of six different parameter estimation methods in wind power applications

In this paper, the different parameter estimation methods such as maximum likelihood method, least square method, weighted least square method, moments method, the method based on quantiles and maximum spacing method for the Weibull distribution which are widely accepted in wind power studies are presented and discussed. The effect of the selecting parameter estimation method on the estimation of wind power is also discussed. Then the wind power estimation results that would be obtained by using the parameters estimated with each method are evaluated. The analysis is performed to show the effects of overestimating or underestimating wind power values caused by an erroneous estimation of the probability density function parameters. n n xa0 n n Key words:xa0Weibull distributions, parameter estimation, wind, wind power, wind energy.

Solar Radiation Data Modeling with a Novel Surface Fitting Approach

In this work one year hourly solar radiation data are analyzed and modeled. Using a 2-D surface fitting approach, a novel model is developed for the general behavior of the solar radiation. The mathematical formulation of the 2-D surface model is obtained. The accuracy of the analytical surface model is tested and compared with another surface model obtained from a feed-forward Neural Network(NN). Analytical surface model and NN surface model are compared in the sense of Root Mean Square Error (RMSE). It is obtained that the NN surface model gives more accurate results with smaller RMSE results. However, unlike the specificity of the NN surface model, the analytical surface model provides an intuitive and more generalized form that can be suitable for several other locations on earth.

Mycielski approach for synthetic wind speed data generation

In this paper, a novel Mycielski based approach for wind speed data generation is developed and presented. The efficincy of the proposed approach is tested using hourly wind speed data obtained from İzmir region. To test the efficiecy of the approach, the four year-long measured data are seperated into two parts: data belonging to first three years are used for training whereas the remaining one-year data are used for testing and accuracy comparison purposes. In order to compare the efficiency of the proposed generation method, the same data are used in artificial wind speed generation with the classical method of first order Markov chains. Results indicate that the proposed Mycielski based algorithm produces better quality artificial data as compared to previous methods.

An application of MDLPF models for solar radiation forecasting

Electricity generation from renewable resources is a hot topic due to increasing environmental awareness of people and energy needs. Solar energy is one of the most important clean energy sources. Forecasting of solar radiation is one of the important stages in sizing and managing a PV power plant. Moreover since the smart grid applications are started, accurate forecasting of the energy output of PV system (hence the solar radiation) became a hot topic. In literature there are a huge number of studies tries to find more accurate forecasting models. Among them in this study Multi-Dimensional Linear Prediction Filter Models (MDLPF Models) are studied. To test the performance of MDLPF models, hourly solar radiation data of two different regions (Ankara and Çanakkale) are employed. In forecasting five different MDLPF Models are built. The accuracies of the forecasting results are compared and discussed.

Deformation classification of cutting discs using artificial neural networks

In this study, a classification application is realized to determine the deformation status of the cutting disc. 673 cutting experiments data obtained from a marble cutting machine suited in a laboratuvary of the Afyon Kocatepe University are evaluated for this purpose. During the cutting process, 8 different signals (Axial forces (Fx, Fy, Fz), Noise, Peripheral speed of the disc, Current, Voltage and Power) are measured and collected. The mean values of the each experiments are used as a 8 length feature vector. To determine the deformation class of the disc (undamaged, less damaged, very damaged and broken) these feature vectors are used. On the other hand Artificial Neural Networks (ANNs) are employed as classifiers. It is obtained that proposed method is able to classify the deformation status of the disc with 95,86 % accuracy.

Wind speed prediction with Mycielski algorithm

Wind speed prediction is an important issue in wind related engineering studies. However, the wind data has random behavior like other meteorological events. Therefore, it is difficult to apply conventional statistical approaches. On the other hand, wind speed data has an important feature; large fluctuations from a wind state to a significantly different level is relatively seldom. This feature leads to some patterns that should be exemined in detail. In this study, a novel approach for wind speed modeling using Mycielski algorithm that considers this important future is demonstrated. Developed procedure, predicts future values of wind data by analysing repeatings in the history of data and assumes that the history will be structurely repeated in the future. The prediction capability of the procedure is tested using wind speed data obtained from 2 cities in Turkey: İzmir and Antalya. Reasonable prediction results are obtained and analysis results are reported.

A 2 dimensional solar radiation model

In this study solar radiation data obtained from Eskisehir region is mathematically modeled using a two dimentional (2D) approach. The approach and model is novel in the literature of solar radiation modeling. The analysis is based on mathematical behavior of hourly and daily behavioral cross-sections of the 2D data. It is observed that the deviation of the hourly data within a day exhibits a Gaussian shape, and the deviation of daily data in the year has a sinusoidal behavior. The hourly behaviour of daily data is tested by single and double Gaussian source models, whereas the daily behavior of yearly data is only modeled using sinusoidal function. By this way two different equations corresponding two different surfaces are obtained. It is concluded that single-source Gaussian surface represents the data more accurate than two dimentional Gasussian surface. Consequently, a very simple but an accurate 2D model is obtained for solar radiation data.