machine learning

1. “MACHINE LEARNING AND DEEP LEARNING
APPLICATION IN POWER OUTPUT
PREDICTION OF DIFFERENT RENEWABLE
ENERGY SYSTEMS AND ITS COMPONENTS”
Presented by:
KONDAPALLI SRINIVASA VARAPRASAD
Doctor of Philosophy (Electrical and Electronics Engineering)
TECHNO INDIA UNIVERSITY, KOLKATA,WB,INDIA

2. Goal of the Project:
Solar Power Output Prediction
Power output from horizontal photovoltaics installed in 15 locations in
the northern hemisphere is predicted. Only location and weather data are
used without information about irradiance.
Wind Power Output Prediction
Estimating the amount of wind energy production per hour in the next
24 hours by applying machine learning (ML) techniques using historical
wind power generation data and weather forecasting reports..

4. Glimpses into the Dataset for Wind Power:
The dataset consists of 11824 rows and 22 columns.

5. Visualisations and missing values
The dataset shows 118224 observations for 22 features but some of the
features have significant periods of missing data. We need to discard
some periods and fill in the missing periods, otherwise our LSTM
model will not converge, or we may end up feeding it with garbage data.

6. LSTM Model
It is special kind of recurrent neural network that is capable of learning
long term dependencies in data. This is achieved because the recurring
module of the model has a combination of four layers interacting with
each other.

7. Predicted vs Actual Power

8. Formulas Used

9. Histogram for Wind Power

10. Correlation for Wind Power

11. Glimpses into the Dataset for Solar Power:
The dataset consists of 21,045 rows and 17 columns.

12. Histogram of Solar power output variable in the entire dataset:

13. Correlation for solar power:
correlation between available features and power output. From the correlation plot,
ambient temperature, cloud ceiling, and humidity are the top three most correlated
features with solar power output..

14. Hyper-parameter tuning for solar power:
Each of the models was tuned using the random search cross validation
approach which enables the selection of the best combination of hyper-
parameters based on the performance of the model on multiple splits of the
training data.

15. Test data scores:
The performance of each model is evaluated using the hold-out
set which is 20% of the entire dataset. The results are
summarized below:

16. Modeling:
Three models (Random Forest — RF, Light Gradient Boosting Machine
— LGBM, and Deep Neural Network — DNN) and a stacked ensemble
were developed and compared with a baseline (K Nearest Neighbors—
KNN) model.

17. Metrics : The R-squared metric is the ultimate metric for selecting the best-performed
model in this analysis. Other metrics useful for assessing the performance of selected
models include root mean squared error (RMSE) and mean absolute error (MAE).
R-squared:

18. RMSE:

19. MAE:
The values of R-squared go from 0 to 1 and the higher the better
while the values of RMSE and MAE have the same unit as the
power output (W) and the smaller the better.

20. Model Stacking
Four disparate models (KNN, DNN, RF, and LGBM) were combined using the stacking
regressor module.

21. Results: Cross-validation scores
The cross-validation (CV) R-squared scores for 1000 random permutations of hyper-
parameters for different algorithms are shown in the boxplot below: As shown on the
boxplot, the LGBM model is the most sensitive to hyper-parameters selection while KNN is
the least sensitive.

22. The best CV scores for each algorithm type is displayed below: