Improving Market Clearing Price Prediction by Using a Committee Machine of Neural Networks英文教材.pdfVIP

IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 19, NO. 4, NOVEMBER 2004 1867 Improving Market Clearing Price Prediction by Using a Committee Machine of Neural Networks Jau-Jia Guo, Student Member, IEEE, and Peter B. Luh, Fellow, IEEE Abstract—Predicting market clearing prices is an important but difficult task, and neural networks have been widely used. A single neural network, however, may misrepresent part of the input-output data mapping that could have been correctly repre- sented by different networks. The use of a “committee machine” composed of multiple networks can in principle alleviate such a difficulty. A major challenge for using a committee machine is to properly combine predictions from multiple networks, since the performance of individual networks is input dependent due to mapping misrepresentation. This paper presents a new method in which weighting coefficients for combining network predictions Fig. 1. Schematic of an ensemble-averaging committee machine with network are the probabilities that individual networks capture the true predictions y^ and weighting coefficients a . input-output relationship at that prediction instant. Testing of the New England market cleaning prices demonstrates that the new method performs better than individual networks, and better than space of mixtures of experts is divided into several regions committee machines using current ensemble-averaging methods. (subspaces) to which different neural networks are assigned. A gating network decides which network prediction will be se- Index Terms—Committee machines, energy price forecasting, multiple model approach, neural networks.