Modelling and forecasting electricity consumption (EC) help industry managers make better strategic decisions. In this study, a hybrid approach for predicting EC is proposed which first EC is decomposed into approximate and detail parts based on wavelet transfer (WT). Next, the modeling of wavelet components is accomplished based on the AWT-LSTM and ARMAX-GARCH type models. The key concept of AWT-LSTM is adding an optimal set of adaptive layers through which AWT-LSTM can automatically concentrate on various frequency components based on the dynamic evolution of the input sequence revealing the multi-frequency pattern of EC. Then, the components are predicted separately and eventually, the results are combined nonlinearly through RF algorithm. Besides the lag calendar setting, the affecting exogenous variables are applied on the electricity consumption in the structure of hybrid models which include humidity, water precipitation, solar radiation, effective temperature, wind direction, and wind speed. To measure the predictive power, the proposed models were tested using a 24 hours ahead forecasting horizon using a sample rate of 30 minutes in two different seasons (winter and summer). The empirical results indicate the hybrid model of AWT-LSTM-ARMAX-EGARCH in mean-RF with RMSE = 1.536, MAPE = 1.276, NMAE=0.0116, NRMSE=0.0143 values in summer test and RMSE = 1.249, MAPE = 0.869, NMAE=0.0078, NRMSE=0.0092 values in winter test outperforms the benchmark models (wavelet transform-feed forward neural network-random forest, wavelet transform-long short-term memory- random forest, wavelet transform-exponential smoothing-random forest, and wavelet transform-(seasonal) autoregressive integrated moving average with explanatory variable-generalized autoregressive conditional heteroscedasticity types models). The results suggest that employing the wavelet transform for the decomposition of data and random forest for the nonlinear aggregation of sub-signals (predicted values of detail and approximate and parts) can significantly improve the prediction power of predictive models
