Summary: This paper focuses on the methodologies into which the fuzzy expert system is embedded into well-developed neural network based algorithms to achieve improved performance for short-term electric load forecasting. Its performance is then contrasted with an Artificial Neural Network (ANN) and a Fuzzy Inference System (FIS). These models have been developed to produce a short-term forecast of the electrical load demand on an hourly basis. These techniques have been tested on actual load and weather data for the year 2000 provided by the Czech Electric Power Company (CEZ), Czech Republic. The illustrative results obtained through the application of soft computing approach show the adequacy of the adopted approach. FNN shows better convergence speed than a pure ANN and FIS approach, which confirms its applicability in the real life applications.
For the entire collection see [Zbl 0997.00023].