A New algorithm to find number of hidden neurons in Radial Basis Function Networks for Wind Speed Prediction in Renewable Energy Systems

作者：Sheela K Gnana^*; Deepa S N

来源：Control Engineering and Applied Informatics, 2013, 15(3): 30-37.

• Mean Square Error
• Neural networks
• Radial basis function networks
• Renewable energy systems
• Wind Speeds

This paper proposes a new algorithm to find hidden neurons in Radial Basis Function Networks for wind speed prediction in renewable energy systems. The proper selection of hidden neuron is important in the design of neural network architectures. The random selection of hidden neuron may cause over fitting and under fitting problem in the network. To find the number of hidden neurons, 101 various criteria are examined based on the error values of mean squared error, mean absolute percentage error and mean absolute error. The minimal error values are considered as the best solution to find hidden neurons in Radial Basis Function network. The proposed new algorithm is tested on real time wind data. The significance of number of hidden neurons is analyzed using these criteria. The criteria are satisfied with convergence theorem. The experimental results show that as compared to alternative methods proposed algorithm performs better in terms of testing errors. The new algorithm is effective, accurate with minimal error than other approaches. Simulations infer that with minimum error the proposed algorithm can be used for wind speed prediction in renewable energy systems.

• 出版日期2013-9