Precisious Estimation of Solar Irradiance by Innovative Neural Network and Identify Exact Hidden Layer Nodes through Novel Deciding Standard

Abstract

In this research carried out work on modeling innovative neural network based on a mixture of six self governing artificial neural network such as three feed forward neural networks (Radial basis function neural network (RBFNN), Multi-layer perceptron neural network (MLPNN) and Back propagation neural network (BPNN)), one feedback neural network (Elman neural network (ENN)) and two new feed forward neural networks (Recursive radial basis function neural network (RRBFNN) and Improved back propagation neural network (IBPNN)) whose output are accumulated and averaged to get the final exact estimation of solar irradiance to assist the solar farm and energy system operator. This research attempt to identify apt hidden layer nodes based on a novel deciding standard for innovative neural networks. Hence, omit prospect of over learning and lack of memory capacity problems. Among proposed various 153 deciding standards, (8(n+1)+1/(n-6))is identified esteem apt deciding standard for identification of hidden layer nodes in the innovative neural network because validation on real-time data source it results very much lower error qualifiers such as 3.7752e-08 of MSE, 3.2158e-05 of MAE, 9.3693e-06 of MRE, 1.9430e-04 of RMSE and 9.3698e-04 of MAPE for solar irradiance estimation. The presented all deciding standards are convergent, which is evidenced by convergence theorem. Proposed the best deciding standard possess superiority compared with other previous and suggested deciding standards for identification of apt hidden layer nodes. The limitation regard with self governing artificial neural networks is overcome by proposing an innovative neural network. Therefore, this network achieves improved estimation accuracy and generalization capability.