A graph based method for faster display for distribution networks
WSEAS TRANSACTIONS on SYSTEMS
ICC'06 Proceedings of the 10th WSEAS international conference on Circuits
A novel learning scheme for Chebyshev functional link neural networks
Advances in Artificial Neural Systems
| Hi-index | 0.00 |
This paper presents a functional-link network based short-termelectric load forecasting system for real-time implementation. Theload and weather parameters are modelled as a nonlinearautoregressive moving average (ARMA) process and parameters of thismodel are obtained using the functional approximation capabilitiesof an auto-enhanced Functional Link net. Numerous and significantadvantages accrue from using a flat net, including rapid quadraticoptimisation in the learning of weights, simplification in thehardware as well as in computational procedures. The functionallink net based load forecasting system accounts for seasonal anddaily load characteristics as well as abnormal conditions, holidaysand other conditions. It is capable of forecasting load with a leadtime of one hour to seven days. The adaptive mechanism with anonlinear learning rule is used to train the network on-line. Theresults indicate that the functional link net based loadforecasting system produces robust and more accurate load forecastsin comparison to simple adaptive neural network or statisticalbased approaches. Testing the algorithm with load and weather datafor a period of two years reveals satisfactory performance withmean absolute percentage error (MAPE) mostly less than 2% for a24-hour ahead forecast and less than 2.5% for a 168-hour aheadforecast.