In the last few years, predicting electricity consumption has become one of the most essential sectors for both electric utility centers and customers. Anticipating power usage is critical for effective management decisions and company strategies. This research presents an effective and highly accurate hybrid electricity consumption prediction model in which Discrete Wavelet Transform (DWT), Autoregressive Integrated Moving Average (ARIMA), Artificial Neural Network (ANN), and Cuckoo Search (CS) optimization technique is utilized. The reduction in errors generated in electricity consumption predictions and an increase in its accuracy is the primary aim of the suggested approach. The proposed work decomposed the data belonging to Punjab State Power Corporation Limited (PSPCL) using the DWT approach in two levels; these levels signify the lowest and the highest level of energy consumption. After this, ARIMA model implementation was performed on the entire decomposed data for attaining data of Time Series. Further, the data is combined using the Inverse Discrete Wavelet Transform (IDWT), which is improved as well with the use of a nature-inspired CSA approach. The developed model is trained with the implementation of the ANN technique that aids in estimating future electricity utilization using input layers to fed them with the enhanced data. MATLAB software was utilized for validating the performance of the developed approach. The outcomes obtained showcased our hybrid approach is outperforming other similar models in terms of MAP, MAPE, and accuracy.

Load forecasting, Electricity Consumption, ARIMA, Cuckoo search Optimization, DWT, ANN