Warm Start of Mixed-Integer Programs for Model Predictive Control of Hybrid Systems

In hybrid model predictive control (MPC), a mixed-integer quadratic program (MIQP) is solved at each sampling time to compute the optimal action. Although these optimizations are generally very demanding, in MPC, we expect consecutive problem instances be nearly identical. This article addresses question of how computations performed one step can reused accelerate (warm start) solution subsequent MIQPs. Reoptimization not rare practice integer programming: for small variations certain data, branch-and-bound algorithm allows an efficient reuse its search tree and dual bounds leaf nodes. this article, extend ideas receding-horizon settings MPC. The warm-start propose copes naturally with arbitrary errors, has negligible computational cost, frequently enables priori pruning most space. Theoretical considerations experimental evidence show that proposed method tends reduce combinatorial complexity MPC one-step look-ahead optimization, greatly easing online computation burden.