Abstract
A neural-network-based (NN-based) approach for short-term load forecasting of electrical power is proposed. The principal component analysis (PCA) technique is used to reduce the original electric load variables to several characteristic variables. A single parameter dynamic search algorithm (SPDS) is employed to train the NN. Since the training sample sets can be chosen before forecasting, the interference of the non-correlative samples for the forecasting can be avoided. The effectiveness and the feasibility of on line forecasting of the proposed method are examined using simulated experiments.
Index
Terms
Available to subscribers and IEEE members.
References
Available to subscribers and IEEE members.
Citing Documents
Available to subscribers and IEEE members.
Cart
