Using Hybrid Wavelet Approach and Neural Network Algorithm to Forecast Distribution Feeders

In this paper, using an algorithm based on the combination of data based on neural network virology and bacterial nutrition algorithm, improves the performance of the neural network prediction method. Also, the selection of two types of downstream and upstream filters in the wavelet transformation increases the predictive efficacy of neurological prediction. Based on the results, the optimized clustered neural network method has a more favorable response than the other methods. By selecting the appropriate filter and multichannel processing method, the maximum error percentage has improved by 15%. However, compared to the neural network prediction method, the proposed method has more computational volume due to the use of wavelet transform and also three times the use of neural prediction. Due to the large number of layers and used neurons, the neural network method has a much higher computational volume than the linear prediction method, where the linear prediction method has a higher error than the proposed method depending on the data used for training.