Planning of distributed renewable energy systems under uncertainty based on statistical machine learning

Protection and Control of Modern Power Systems

Table 5 Typical scenario generation methods comparison

Method	Characteristic
Wasserstein distance	(1) High accuracy.
	(2) Able to generate extreme scenarios.
	(3) It can only handle a single continuous variable, that is, a weather correlation.
K-means	This method can handle multiple continuous variables.
Semi-supervised learning	(1) Able to generate multivariate typical scenarios.
	(2) Use the data features of a small number of samples as markers to retain the probability features of all state variables for PPF in massive scenarios.
	(3) Ensure the calculation accuracy of uncertainty planning.