
Model Predictive Control (MPC)
Predicts future behavior with a process model and optimizes on a rolling basis under constraints.
Why Water Treatment Needs Predictive Control
Water treatment processes typically involve long time delays, multivariable coupling and strong disturbances. PID control acts only on the current error, cannot anticipate changes in influent load, and cannot handle multiple objectives and constraints simultaneously.
In each control cycle, model predictive control uses the process model to predict changes over a future horizon, solves for the optimal control sequence that satisfies the constraints, applies only the first step, and recalculates in the next cycle based on new measurements. This receding-horizon approach combines anticipation with robustness.
BidiTech applies predictive control in engineering practice to water treatment with the Yukong® APC Software Suite and the DeepControl™ Edge Controller.

Implementation Toolchain
Process Identification
Process models are obtained from operating data or step tests.
Control Configuration
Define variables, model matrix, constraints and weights.
Offline Simulation
Closed-loop simulation to validate control performance and robustness.
Online Control
Closed-loop operation and performance monitoring on the edge controller.