Adequacy Assessment of a Wind-Integrated System Using Neural Network-based Interval Predictions of Wind Power Generation and Load

Ronay Ak (1) , Yan-Fu Li (1) , Valeria Vitelli (2) , Enrico Zio (3, 1)

1 SSEC - Chaire Sciences des Systèmes et Défis Energétiques EDF/ECP/Supélec
2 Department of Biostatistics
3 LGI - Laboratoire Génie Industriel - EA 2606

In this paper, we present a modeling and simulation framework for conducting the adequacy assessment of a wind-integrated power system accounting for the associated uncertainties. A multi-perceptron artificial neural network (NN) is trained by a non-dominated sorting genetic algorithm-II (NSGA-II) to forecast point-values and prediction intervals (PIs) of the wind power and load. The output of the assessment is given in terms of point-valued and interval-valued Expected Energy Not Supplied (EENS). We consider different scenarios of wind power and load levels, to explore the influence of the uncertainty in wind and load predictions on the estimation of system adequacy.

Keywords: prediction intervals; Adequacy assessment; multi-objective genetic-algorithms; neural networks; prediction intervals.

Domains: Electric power; Artificial Intelligence [cs.AI]; Machine Learning [stat.ML]; Statistics [math.ST]; Statistics Theory [stat.TH]

Ronay Ak, Yan-Fu Li, Valeria Vitelli, Enrico Zio. Adequacy Assessment of a Wind-Integrated System Using Neural Network-based Interval Predictions of Wind Power Generation and Load. 2013. ⟨hal-00864843⟩