Short-term Load Forecasting Based on Gene Expression Programming With Error Cycling Compensation
Yongli Zhu · Proceedings of the CSEE · 2008
An attempt to apply gene expression programming (GEP) to short-term load forecasting is made, where the error recycling compensation model is suggested and higher forecasting precision is gained. In order to eliminate the pseudo-data, the load samples are filtered and processed generally first, then the load series of the same time but different days are chosen as the training samples, and by means of the flexible expressive capacity of GEP, the models of different time points are evolved according to time-sharing. Then the errors between forecasting models and samples are evolved by means of GEP as well, and finally, the error compensation models are compensated to the former corresponding forecasting models. And the error compensation models will not be evolved until the results are satisfied. According to forecasting results, it indicates that GEP is of high efficiency and the error recycling compensation models can compensate the errors of evolutionary process. After comparison with the results forecasted by means of time series and genetic programming (GP), it proves that the algorithm of GEP in short-term load forecasting is better.