Tunable linear feedback control of urban drainage systems using models defined purely from data.

Area of Science:

• Environmental Engineering
• Control Systems Engineering
• Urban Water Management

Background:

• Real-time and model-predictive control offer adaptive urban drainage systems (UDS).
• Existing control algorithms face challenges in computational cost, trust, coordination, and labor.
• Linear feedback control presents advantages in computational efficiency, interpretability, and coordination.

Purpose of the Study:

• To develop an automated method for generating tunable linear feedback controllers for UDS.
• To design controllers requiring only system response data, bypassing the need for calibrated software models.
• To evaluate the effectiveness of the proposed method in enhancing UDS performance.

Main Methods:

• Estimating network connectivity via causal inference tools.
• Identifying a linear time-invariant (LTI) dynamical system to approximate the UDS.
• Designing and tuning a feedback controller based on the LTI UDS approximation.

Main Results:

• The method successfully generated tunable linear feedback controllers using only system response data.
• Evaluations across 190 design storms demonstrated strong performance in flooding safety, erosion prevention, and water treatment.
• The study indicates that effective UDS controller synthesis requires surprisingly basic system knowledge.

Conclusions:

• An automated, data-driven approach enables the synthesis of effective, safe, and tunable controllers for UDS.
• This method significantly reduces the reliance on complex process-based models or extensive system knowledge.
• The findings support near-turnkey controller generation directly from sensor data for improved UDS management.