DataFlowML runs Inference Engine as a software module inside the ControlLogix chassis, scoring ML models against live PLC tags at controller scan rates (5-100 ms) without cloud round-trips, enabling closed-loop quality prediction faster than any operator can react. No-code model import workflow accepts PMML, ONNX, and Rockwell’s own Analytics Builder models, allowing data scientists to deploy retrained models to factory controllers via the standard Rockwell DevOps pipeline. DataFlowML Studio provides a visual pipeline builder for data pre-processing (scaling, windowing, feature extraction) before model scoring, matching cloud ML pipeline conventions. Scalable from single machine to enterprise by deploying additional DataFlowML modules across ControlLogix chassis without any infrastructure change. Compatible with all ControlLogix 5580/5570 controllers running Logix 5000 v34+.
IoT Product
Industrial IoT
Rockwell Automation
FactoryTalk DataFlowML deploys pre-trained and custom ML models at the ControlLogix controller edge, executing inference on PLC tag data in real time for quality prediction, anomaly detection, and closed-loop control tuning.