Modelo matemático para predecir la producción de energía eólica a corto plazo, utilizando redes neuronales.

This degree work is focused on the development of a short-term wind energy prediction model (one-day horizon) for the Villonaco Wind Power Station (CEV), based on artificial intelligence, with artificial neural networks (RNAs), especially NARX (nonlinear autorregresive network with exogenous inputs). Taking as a starting point the records of wind speed and active power from 2014 to 2018. Before making use of the data provided, it was necessary to pre-process the information, fill in wind and power data gaps using the Excel software. It was also necessary to carry out a treatment of wind and active power data, thus suppressing outliers. Before the RNAs training process, a mathematical model was designed using the yEd Graph Editor® software, which allowed generating a flow chart of the wind energy prediction process. In the training process, use was made of Neural Network ToolboxTMMatLab® in the Neural Time Series tool with NARX networks, which specializes in data series that evolve over time. After a training and evaluation process, various topologies in the structure of the input and output data were tested. It was finally defined by training an ANN for each month of the year, and data were entered for its learning from 2014 to 2017, thus leaving 2018 to validate the ANNs. To evaluate the performance and precision of the RNAs, the mean squared error estimator (MSE) provided by MatLab® at the end of training, in the open loop and closed loop of the neural network, was used, as well as the Pearson correlation coefficient (R). Another type of evaluation was performed with the 2018 data, which is data that was not used in training, so the MSE and mean relative error (ERM) were used to evaluate the level of precision. Keywords: bin method, artificial intelligence, Narx networks, wind forecasting, wind turbines, wind power.

• Huvudupphovsman: Vicente Guayanay, Jayden Byron (author)
• Materialtyp: bachelorThesis
• Språk: spa
• Publicerad: 2020
• Ämnen: <INTELIGENCIA ARTIFICIAL>< REDES NARX>< PREDICCIÓN DE VIENTO>< AEROGENERADORES>< POTENCIA EÓLICA>
• Länkar: https://dspace.unl.edu.ec/jspui/handle/123456789/23545