

Content:
-----------


GearData_Power_model 		Single-input model for the Power signal on the Gear dataset.
Benchmark.ipynb			Benchmark notebook to test a model.
power_validation_X.pkl		Validation dataset data for the benchmark code.
power_validation_y.pkl		Validation dataset labels for the benchmark code.
readme.txt			This file.

Introduction:
---------------

This repository contains one of the models used in "Convolutional neural network framework for wind turbine electromechanical fault detection" by  Emilie Stone, Stefano Giani, Donatella Zappala' and Christopher Crabtree.
The model is saved into the Tensorflow SavedModel format. 

Benchmark:
-----------

The notebook Benchmark.ipynb can be used to test the code on the user machine.
To run the code, copy the flder ./Models/Single-input Models/GearData_Power_model, containing the single-input mode for Power, into the same folder where the files Benchmark.ipynb, power_validation_X.pkl and power_validation_y.pkl are located.

The notebook load the model and the data and computes the accuracy of the model and the confusion matrix on the validation dataset.
The results are:
Test loss: 0.00024988141376525164
Test accuracy: 1.0

Confusion Matrix:
[[450   0   0   0   0]
 [  0 450   0   0   0]
 [  0   0 450   0   0]
 [  0   0   0 450   0]
 [  0   0   0   0 450]]
 
 
 The code has been tested with the following packages:
 Python 3.10.11
 tensorflow 2.12.0
 sklearn 1.2.2
 numpy 1.22.4
 pickle 4.0
