BISMBISM generates training data
The mechanistic model supplies extensive, high-quality synthetic data where historical sewer observations are scarce.
BISM / AI + BISM
AI for H₂S prediction and control
SPM, the Swift Prediction Model, combines BISM with machine learning, using a verified process model as the basis for fast prediction and real-time control.
AI + BISM
SPM: combining BISM with machine learning
SPM, the Swift Prediction Model, combines BISM with machine learning, using a verified process model as the basis for fast prediction and real-time control.
Machine learning builds a fast prediction model
Algorithms learn the complex relationship between wastewater quantity, quality and H₂S concentrations to approximate BISM responses.
SPM supports real-time control
Fast H₂S predictions support control adjustments, helping reduce overdosing and control failure.
< 1 sPrediction time in the reported case
100%Reported target compliance in the case
Performance figures refer to the reported research case and its test conditions, rather than a guarantee across all networks and operating conditions. The illustrated control target is 20 ppm.
Water Research X, 2024 ↗