Emission Mitigation Dispatch in Wind Power Generation with Expedited Machine Learning

In this paper: a machine learning (ML) module is presented for emission mitigation dispatch (EMD) in wind power (WP) generation. It is developed for the mobile edge computing (MEC) platforms. Therefore, the module is based on a lightweight ML scheme: radial basis neural network (RBN). It has been well known that RBN is compact compared with a more popular ML scheme: feedforward neural network. However, this is the first study to combine EMD with RBN, strongly motivated by the MEC implementations. MEC is becoming an important platform in the evolving smart grid program. As a trial, a case study with 36 generators and 50 wind turbines is presented. Two standard ML stages, training and testing, are conducted in the simulation. It is shown that, by using the trained RBN, the time of solving EMD is significantly reduced. It is expected to be a desirable model for MEC.