Wind Turbine Fault Detection Using XGBoost & Random Forests

Please cite the following reference in your future publications.

Zhang, D., Qian, L., Mao, B., Huang, C., Huang, B., & Si, Y. (2018). A data-driven design for fault detection of wind turbines using random forests and xgboost. IEEE Access, 6, 21020-21031.

######################################################################
###Wind Turbine Fault Detection Using XGBoost, Random Forests and SVM###
######################################################################

Zhejiang Uniersity, Ocean Energy Lab, Insititute of Ocean Engineering and Technology

Yulin. Si
Mail:Yulinsi@zju.edu.cn

Liyang. Qian.
Mail:spectrum@zju.edu.cn

########################################################################

Directories:

.../FAST_V8/CertTest -- FAST input files (Read the FAST user's guide before use)

.../FAST_V8/Simulink/XGB_TreeModels -- XGBoost dump models
.../FAST_V8/Simulink/FaultDetection.mdl -- FD process simulink models (FAST V8 & MATLAB 2015b X86)
.../FAST_V8/Simulink/FDIBenchMarkData.m -- Simulation parameters setting
.../FAST_V8/Simulink/mat2data.m -- Transfer .mat data to .csv data
.../FAST_V8/Simulink/run.m -- Run the simulation (Note to set the path and name of .fst file)

.../Python/RF_XGBoost_Training.py -- Training and predicting with RF, XGBoost and SVM (Installed libraries first)
.../Python/Dump_XGBoost_Model.py -- Select features with RF and predict using XGBoost, classifier dumped as .txt file

########################################################################

How to observe the FD results:

1)Make sure how to run a FAST-Simulink combined model

2)Set parameters correctly and run 'run.m'

3)Results in scopes (FaultDetection/Fault Detection Subsystem/...)

########################################################################

How to save simulation data, train model and test model:

1)Make sure how to run a FAST-Simulink combined model

2)Set parameters correctly

3)Change one of the 'Terminator module' to 'To File' module. i.e. FaultDetection/Fault Detection Subsystem/claasification fault 2/Terminator2

4)Run 'run.m' and get a .mat file. Name it 'sensordata.mat'.

5)Run 'mat2data.m'. Transfer it to a CSV file. Prepare a training set and a testing set. Name them 'testdata.csv' and 'traindata.csv'

4)Run the 'RF_XGBoost_Training.py' in python 3.6. Note that you need install necessary py library in advance. They are sklearn, pylab, numpy, pandas, xgboost, scipy. 'Dump_XGBoost_Model.py' give a dump file of XGB and you can apply it in simulink model.