Day Types Identification of Algerian Electricity Load Using an Image Based Two-Stage Approach

Short term electricity load forecasting is one of the main concerns for electricity producers in regular system planning, where electricity demand is influenced by the day type among other factors that must be identified before modeling to ensure good load balance. This paper proposes a two-stage approach for identifying day types based on an image of the daily load curve. In the first stage, a set of day classes of load profiles using K-Means clustering algorithm is created, while in the second stage, the Time-Series Visualization method is used to build a classification model able to assign different days to the existing classes, detecting visual characteristics from daily load data curves. This classification model could be used in the forecasting process either by including the day-type as an input or by modeling each day-type independently.

[1]  Chen Qian,et al.  A Novel Method Based on Data Visual Autoencoding for Time-Series Classification , 2015 .

[2]  Furong Li,et al.  Development of Low Voltage Network Templates—Part I: Substation Clustering and Classification , 2015, IEEE Transactions on Power Systems.

[3]  Yoshua Bengio,et al.  Why Does Unsupervised Pre-training Help Deep Learning? , 2010, AISTATS.

[4]  Carlos E. Pedreira,et al.  Neural networks for short-term load forecasting: a review and evaluation , 2001 .

[5]  S. P. Lloyd,et al.  Least squares quantization in PCM , 1982, IEEE Trans. Inf. Theory.

[6]  Tomonobu Senjyu,et al.  Next-Day Load Curve Forecasting Using Neural Network Based on Similarity , 2001 .

[7]  Marios M. Polycarpou,et al.  Short Term Electric Load Forecasting: A Tutorial , 2007, Trends in Neural Computation.

[8]  Ping Wang,et al.  Application of Pattern Recognition and Artificial Neural Network to Load Forecasting in Electric Power System , 2007, Third International Conference on Natural Computation (ICNC 2007).

[9]  Belén Carro,et al.  Classification and Clustering of Electricity Demand Patterns in Industrial Parks , 2012 .

[10]  Khadir M. Tarek,et al.  Kohonen Maps Combined to K-means in a Two Level Strategy for Time Series ClusteringApplication to Meteorological and Electricity Load data , 2010 .

[11]  Yoshua Bengio,et al.  Extracting and composing robust features with denoising autoencoders , 2008, ICML '08.

[12]  Amy Loutfi,et al.  A review of unsupervised feature learning and deep learning for time-series modeling , 2014, Pattern Recognit. Lett..

[13]  Damien Fay,et al.  Day Type Identification for Algerian Electricity Load using Kohonen Maps , 2008 .

[14]  Damien Fay,et al.  24-h Electrical Load Data - a Sequential or Partitioned Time Series? , 2003, Neurocomputing.

[15]  Yoshua Bengio,et al.  Random Search for Hyper-Parameter Optimization , 2012, J. Mach. Learn. Res..