This repository focuses on obtaining a method for power curve modelling that is more accurate than existing techniques and can be implemented for real time turbine monitoring of wind farms. data is also published in this work for ease of testing wherever required.
A novel filteration technique is introduced in this work based on quantiles set on a normal distribution of the data.
we utilize a novel quantile filtering algorithm, a Radial Basis Function and Multi-layer Perceptron Neural network.
Quantile filtering is based on quantiles set on a normal distribution of data that specifies the location of dataset regarded as clean and the region of dataset regarded as outliers.
we also developed a statistical method of fault detection based on the conversion the distance between points in the 2 dimension space into confidence levels for a clean data.
for the conditions implemented in this work
@ONLINE{
Remove outlier
for x in range(1, 62):
if x <= 3:
F = 0.95
elif ((x > 3) and (x <= 10)):
F = 0.9
elif ((x > 10) and (x <= 20)):
F = 0.92
elif ((x > 20) and (x < 30)):
F = 0.96
else:
F = 0.985
d1[x] = outlier_remover(d1[x], 'wind speed', 0.00001, F)
}
Here is the main python notebook that explains all of the stuff step by step!
below is a picture of the model architecture
if you use this code for your publicatons, please cite it as:
@ONLINE{hse,
author = "Henry Emerald"
title = "WTPC modelling using quantile filtering and neural netowrks (MLP and RBF)"
year = "2022"
url = "https://github.com/henrii1/wind-turbine-power-curve-modelling"
}
Henry Emerald