One-Hour-Ahead Load Forecasting Using Neural Networks

Senjyu, Tomonobu; Takara, H.; Uezato, K.; Funabashi, Toshihisa

doi:10.1109/mper.2002.4311688

Cited by 96 publications

(4 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The hybrid load forecast [13][14][15][16] had shown an improvisation of load forecast accuracy. AI based techniques such as artificial neural network (ANN) [17][18][19] uses weather ensemble data as one or more of the input to the developed models. The ANN technique is proven reliable in prediction error, however very large historical data is needed.…”

Section: Introductionmentioning

confidence: 99%

Autoregressive method in short term load forecast

Baharudin

Kamel

2008

2008 IEEE 2nd International Power and Energy Conference

View full text Add to dashboard Cite

Short-term load forecasting plays an important role in planning and operation of power system. The accuracy of this forecasted value is necessary for economically efficient operation and also for effective control. This paper describes the methods of autoregressive (AR) Burg's and modified covariance (MCOV) in solving a short term load forecast. The methods are tested based on historical load data of New South Wales, Australia. The accuracy of discussed methods are obtained and reported.

show abstract

Section: Introductionmentioning

confidence: 99%

Autoregressive method in short term load forecast

Baharudin

Kamel

2008

2008 IEEE 2nd International Power and Energy Conference

View full text Add to dashboard Cite

show abstract

“…4 clusters × 4 contract types = 16 partial regression coefficients are created for each hour and each power distribution line. In order to create the final multiple regression formulas, the validity of dividing the clusters must be established based on the dispersion of these partial regression coefficients.…”

Section: A Methods For Creating Prediction Modelsmentioning

confidence: 99%

The load forecasting technology in the electric power distribution system

2005

View full text Add to dashboard Cite

SUMMARYThe authors proposed a method for creating a prediction model for dynamically predicting the load curve in a power distribution system whose composition varied day by day, and then validated its effectiveness for comparisons with measured current values.In the results, the authors divided power distribution lines into interval increments using the contract power category composition ratio and then created a prediction model for various interval patterns based on the substantial dependence of the consumer load curve on contract type. Furthermore, when confirming the prediction precision using their model, the authors found that the absolute average error was 7.7 A, with 95% within 30 A. The relative average error rate was confirmed at 9.6%. Upon systematizing the current value prediction method and performing opinion surveys and evaluations of users running trials in the power distribution system operation business, the authors were able to confirm that the prediction precision of their method was sufficient for the system's needs.

show abstract

“…Large amounts of input variables not only complicate the neural network structure, but also increase the corresponding learning time. However, in scenarios where fast or even real-time load forecasting is required, the learning time of the neural network model is critical [13]. To overcome this time-consuming characteristic, most previous works on neural network optimization focused on the algorithmic level.…”

Section: Introductionmentioning

confidence: 99%

Use of High-performance Graphics Processing Units for Power System Demand Forecasting

He¹,

Meng²,

Dong³

et al. 2010

Journal of Electrical Engineering and Technology

View full text Add to dashboard Cite

-Load forecasting has always been essential to the operation and planning of power systems in deregulated electricity markets. Various methods have been proposed for load forecasting, and the neural network is one of the most widely accepted and used techniques. However, to obtain more accurate results, more information is needed as input variables, resulting in huge computational costs in the learning process. In this paper, to reduce training time in multi-layer perceptron-based short-term load forecasting, a graphics processing unit (GPU)-based computing method is introduced. The proposed approach is tested using the Korea electricity market historical demand data set. Results show that GPU-based computing greatly reduces computational costs.

show abstract

One-Hour-Ahead Load Forecasting Using Neural Networks

Cited by 96 publications

References 9 publications

Autoregressive method in short term load forecast

Autoregressive method in short term load forecast

The load forecasting technology in the electric power distribution system

Use of High-performance Graphics Processing Units for Power System Demand Forecasting

Contact Info

Product

Resources

About