For electric power research, we used natural language processing to enable our client to quickly mine thousands of industry reports, and find patterns in narrative data to create an “early warning system” for plant operators.
To classify the data, we developed an electric power industry dictionary, and then applied NLP algorithms to analyze text in maintenance reports, regulatory filings, public reports from ADAMS database, and more.
Improve operations with machine learning
To help operators improve plant conditions, we created an ML-driven decision analysis tool that provided insights on actions to take to correct unusual patterns and reduce plant downtime.
This machine learning tool used aggregated power generation data from multiple plants in its optimization rules, and improved over time as new datasets were added.
Increase effectiveness with ML coaching
We helped client analysts define and implement the best predictive analytics approaches by assessing their machine learning models and the quality of datasets, and providing strategic guidance.
Our ML coaching helped identify and correct biases in ML models before implementation, and increased confidence in utility staff for predictive modeling.