Management of short term load forecasting in South African power networks

Accurate short term load forecasting (STLF) is a prerequisite for proper generation scheduling and reliable operation of power utilities. Conventional methods of STLF, suffer from the disadvantages such as lack of ability to accurately model the weather parameters affecting the load, lack of robustness for representing weekends and public holidays and of being computation intensive. Application of intelligent techniques like the Adaptive Neuro Fuzzy Inference System (ANFIS) which combines the low-level computation power of neural networks with the high-level reasoning capability of a fuzzy inference systems, helps to alleviate these problems by defining the STLF problem with linguistic variables from historical load data. This paper reports on the development and application of ANFIS-based model for short term load forecasting for South African power networks considering temperature and humidity as the main weather parameters affecting the load. The model is tested and validated with real time load data obtained from South African networks.