A novel hybrid model based on neural network and multi-objective optimization for effective load forecast. (1st September 2019)

• Record Type: Journal Article
• Authors: Singh, Priyanka; Dwivedi, Pragya
• Is Part Of: Energy. Volume 182(2019)
• Page Start: 606
• Page End: 622
• Publication Date: 2019-09-01
• Subjects: Load forecasting -- Artificial neural network -- Multi-objective optimization -- Hybrid forecasting model -- Multi-objective follow the leader (MOFTL); Power resources -- Periodicals; Power (Mechanics) -- Periodicals; Energy consumption -- Periodicals; 333.7905
• DOI: 10.1016/j.energy.2019.06.075
• Languages: English
• ISSNs: 0360-5442

Abstract

Abstract: In recent years, increased attention has been paid by the researchers to predict accurate and stable load due to its effect on the economy and need for proper management of power systems. However, most of the previous research focused only on either reducing load forecast error or enhancing the stability, very few studies focused on these two issues simultaneously. Introducing a forecasting model to solve both independent objectives at the same time is a challenging task due to the complex behavior of the load pattern. Therefore, to achieve two objectives simultaneously, we propose a novel multi-objective algorithm (MOFTL) based on Follow The Leader algorithm. The effectiveness of MOFTL has been shown by comparing the results with three newly presented MOWCA, MOPSO and NSGA-II multi-objective algorithms. Moreover, to validate the performance of MOFTL, we have combined MOFTL with neural network termed as MOFTL-ANN to solve the problem of electricity load forecasting. The proposed hybrid model outperforms baseline models over two real-world electricity data sets namely England region and ERCOT region. MOFTL-ANN shows improvement of 17.42%, 6.81%, 10.77% and 59.69% MAPE values for England region and 4.20%, 4.16%, 1.14% and 21.85% MAPE values for ERCOT region over NSGA-II-ANN, FTL-ANN, BPNN, and GRNN. Highlights: A novel multi-objective follow the leader (MOFTL) algorithm is pro-posed. MOFTL is tested on benchmark problems and compared with MOWCA, MOPSO, and NSGA-II. Hybrid model (MOFTL-ANN) obtains higher accuracy and stability for forecasting model.