3 ms·
Invenia has been interested in electricity grids since 2006 and has provided many kinds of services to interested parties, for example load forecasts. We are co
by oxinabox 7y ago
Invenia has been interested in electricity grids since 2006 and has provided many kinds of services to interested parties, for example load forecasts. We are continuing to work on many things in the area. Recently we published some of our blue sky research on grid optimization using modern machine learning ideas (https://www.climatechange.ai/CameraReady/43/CameraReadySubmission/icml_invenia_cameraready.pdf https://www.climatechange.ai/CameraReady/43/CameraReadySubmi...).
- 7thaccount 7y agoLoad forecasting is a reasonable thing. There are already enough vendors for that to where I wouldn't focus in that area personally, but if it is working for you, then that is good. I read your linked paper last night and I'm not 100% sure what to make of it. If I understood it (big assumption :)) it seems like the intent is to use heuristics (Ex: particle swarm optimization) to seed a warm start for a nonlinear solver for the ACOPF problem? Is that right? Although the nonlinear ACOPF problem is academically interesting, the industry is using every last ounce of performance from the best MIP solvers (CPLEX & GUROBI) for the DCOPF unit commitment problem. The 5-minute economic dispatch problem only takes a few seconds as it is formulated as a LP problem, however, the unit commitment process is a MIP problem though and I don't think the nonlinear solvers will have anywhere near the performance we need even with a warm start. There is some interesting work being done on ACOPF (ARPA-E competition project that ASU is heading up).