Main content area

Probabilistic modelling and analysis of stand-alone hybrid power systems

Lujano-Rojas, Juan M., Dufo-López, Rodolfo, Bernal-Agustín, José L.
Energy 2013 v.63 pp. 19-27
Monte Carlo method, algorithms, batteries, data collection, energy, fuels, neural networks, prices, probabilistic models, probability analysis, solar radiation, uncertainty, wind speed
As a part of the Hybrid Intelligent Algorithm, a model based on an ANN (artificial neural network) has been proposed in this paper to represent hybrid system behaviour considering the uncertainty related to wind speed and solar radiation, battery bank lifetime, and fuel prices. The Hybrid Intelligent Algorithm suggests a combination of probabilistic analysis based on a Monte Carlo simulation approach and artificial neural network training embedded in a genetic algorithm optimisation model. The installation of a typical hybrid system was analysed. Probabilistic analysis was used to generate an input–output dataset of 519 samples that was later used to train the ANNs to reduce the computational effort required. The generalisation ability of the ANNs was measured in terms of RMSE (Root Mean Square Error), MBE (Mean Bias Error), MAE (Mean Absolute Error), and R-squared estimators using another data group of 200 samples. The results obtained from the estimation of the expected energy not supplied, the probability of a determined reliability level, and the estimation of expected value of net present cost show that the presented model is able to represent the main characteristics of a typical hybrid power system under uncertain operating conditions.