Day Ahead Regional Electrical Load Forecasting Using ANFIS Techniques
- PDF / 3,294,875 Bytes
- 21 Pages / 595.276 x 790.866 pts Page_size
- 89 Downloads / 168 Views
ORIGINAL CONTRIBUTION
Day Ahead Regional Electrical Load Forecasting Using ANFIS Techniques Ram Dayal Rathor1
•
Annapurna Bharagava1
Received: 7 September 2018 / Accepted: 7 August 2020 The Institution of Engineers (India) 2020
Abstract Short-term load forecasting is a powerful tool for improvement of operation, energy efficiency and reliability of power systems. Researchers are continuously working to improve outcomes of short-term load forecasting (STLF). In this paper, three different ANFIS models are developed for STLF. The proposed models are tested for prediction of load demand of Rajasthan region of India, from fifteen minutes to one week ahead for particular time of the day of year 2015. Rajasthan region has a typical load curve as it has a land area of 342,239 km2 and population of 68 million, with acute climatic conditions. The outcomes obtained from proposed models are compared with outcomes of significant strategies available in literature based on ANN. This comparison reveals that the proposed RR (Rajasthan Region) model is a competitive technique among all other strategies. The results are compared on the basis of MAE, APE and MAPE for fifteen forecasting samples. Keywords Load forecasting Artificial neural network Fuzzy logic ANFIS
Introduction Electrical load forecasting plays a powerful role in capacity planning, scheduling, and the operation of power systems [1]. It provides very important information for generation, & Ram Dayal Rathor [email protected] Annapurna Bharagava [email protected] 1
Rajasthan Technical University, Kota, India
control, power dispatch, maintenance, and expansion of power facility with fewer problems to their consumers [2, 3]. Decisions related to unit commitment, economic dispatch, automatic generation control, security assessment, maintenance planning, and energy exchange depend on the trends of upcoming load demand [4]. Accurate STLF results in economic and trouble free operations, improves efficiency with accurate load scheduling and reduces power system reserves and enhances reliability of power grid with reduction in possibility of overloading and blackouts [5, 6]. It decides accurate load demand, with lead times, from a few minutes to several days and schedules spinning reserve for effective control on load flow parameters [7]. Electric load prediction is difficult as it always depends on different unstable factors, like weather variables, social activities, dynamic electricity prices and nonlinear behavior of consumer demand [8]. Many techniques using different methods including artificial neural networks have been used for STLF. ANNbased models are generally used as they perform better with continuously changing environmental parameters, take short time in development and are simple and flexible in design [9]. These are efficient for online implementation in energy control centers but require large training time and pose problem of convergence for complex function approximations [10]. ANNs are unstable, depend on data, and can easily fall int
Data Loading...
