Skip to Main Content
Autodesk

Use Machine Learning to estimate subcatchment inflows to network

Allow users to skip parameterized subcatchment hydrology estimates which may become outdated over time and replace it with machine learning on measured data from TSDB. Flow meter could train algorithm which learns to predict flows based on rainfall. This flow is then allocated to upstream contributing subcatchments. As TSDB grows, flow forecasts potentially improve.
  • Attach files
      Drop here to upload
    • Masoud Mahdisoltani
      Reply
      |
      Aug 22, 2023

      What would be the algorithm? You are essentially asking for automated calibration, which has been studied in numerous papers, yet hasn't made its way to any "real-world" models, as far as I know, due to their limited capacity. If there's a viable algorithm out there for this, then asking for implementing it in InfoWorks ICM is a fair request. But expecting Innovyze to develop an algorithm seems off-topic.