Machine Learning Predictions of Electricity Capacity

This research applies machine learning methods to build predictive models of Net Load Imbalance for the Resource Sufficiency Flexible Ramping Requirement in Western Energy Market. Several are used this research, including Reconstructability Analysis, developed systems community, and more well-known such as Bayesian Networks, Support Vector Regression, Neural Networks. The aims identify variables obtain a new stand-alone model that improves prediction accuracy reduces INC (ability increase generation) DEC decrease Requirements Market participants. accomplishes these aims. built paper wind forecast, sunrise/sunset hour day primary predictors net load imbalance, among other variables, show average size capacity requirements can be reduced by over 25% with margin error currently industry while also significantly improving closeness exceedance metrics. reduction would yield an approximate cost savings $4 million annually one nineteen market Analysis performs best tested.