Alejandro Blanco-M 
Please, if you use this code/dataset, cite me as the author of the code/data compilation. The data is compiled by myself from a SCADA system that I have been working in.
Copyright by Alejandro Blanco-M. Licensed under Eclipse Public License v2.0.
This is a Fuhrländer FL2500 2.5MW wind turbine dataset.
The dataset is stored in JSON format inside the "dataset" folder. It contains five wind turbines (80,81,82,83,84), each one with three years of data with a time interval from 2012 to 2014. The data frequency is 5 minutes reporting four indicators of each 78 sensors (a total of 312 variables). The reported values for each sensor are minimum, maximum, mean, and standard deviation for each 5-minute interval. The dataset also contains the alarms events, indicating the system and subsystem and a small description.
I have included several functions for {R,MATLAB,...} languages to providing an interface that pre-processes and manipulates the RAW data into a table-like format. The table-like format is composed of the variables at the columns and each five-minute data entry in rows.
In the case of matlab code, I have included a ELM (extreme learning machines) classificator model to make some predictions as an example. The ELM model is provided by Pere Marti-Puig
Please increase the matlab java heap memory to more than 4GB following the next instructions: https://es.mathworks.com/help/matlab/matlab_external/java-heap-memory-preferences.html