PNNL

Abstract

This paper presents a computationally tractable framework for real-time predictive control of multi-chiller plants whose operation involves discrete and continuous control decisions coupled through nonlinear dynamics, resulting in a mixed-integer optimal control problem. To address this challenge, the Differentiable Predictive Control (DPC)—a self-supervised, model-based learning methodology for approximately solving parametric optimal control problems—is extended to accommodate mixed-integer control policies. The proposed framework is benchmarked against a state-of-the-art mixed-integer Model Predictive Control (MPC) solver and a fast heuristic Rule-Based Controller (RBC). Simulation results demonstrate that the proposed approach achieves significant energy savings over the RBC while maintaining orders-of-magnitude faster computation times than MPC, offering a scalable and practical alternative to conventional combinatorial mixed-integer control formulations.