Comparative Analysis of Machine Learning Techniques to Predict Stability in Smart Grid

The population and economic development are both booming, which makes it necessary to consider the demand for power as having high growth. Therefore, it becomes essential to properly dispense electricity to home & industrial users in line to prevent power loss. Such power losses during electricity distribution can be reduced via smart grids. Consumer power consumption accurately will forecast by using clever techniques like machine learning and artificial intelligence on smart grids. Evaluation and analysis of the different machine learning algorithms are essential steps in choosing the best one to use with smart grids. Support vector machines (SVM), Decision trees (DT), Naive Bayes (NB), logistic regression (LR), K-nearest neighbour (KNN) are the machine learning algorithms utilized in this study to predict the smart grid stability. The simulation outputs demonstrated that excellence of the DT classification technique, which outperforms the other algorithms used in this work, yielding
$100{{\% }}$
accuracy,
$100{{\% }}$
recall,
$100{{\% }}$
precision, and best F1 score. Smart grid dataset in this study is taken from the UC Irvine (UCI) machine learning repository, it is openly available dataset.