Generation Siting for Grid Reliability: Optimization Modeling and an ML Approach to Cascading Failure Risk
And/Or: How I learned to stop worrying and love AI?! 


Today we will discuss three topics.

First - Power grid modeling to find optimal placement of new generation focusing on the renewable resources of wind and solar. We will use Alaska's railbelt grid as our grid of interest and optimize for overall risk.

Second -  Development of a ML tool to look at the system state and forecast failure probability.  We use both the railbelt grid which is rather unique in structure and a 800 bus meshed grid for investigation.  We run the extensively validated OPA model to produce both the training data and the test data.  The ML tool shows promise with blackout size and as a tool for fast what-if analysis.   The ML model development, strengths and weaknesses and results will be discussed.

Third and finally - AI according to David (and all of you), its rapidly evolving present state, uses and weaknesses and some thoughts/discussion about the future.