Rolling Horizon Policies in Multistage Stochastic Programming

TL;DR

This paper investigates how to determine the optimal number of stages in rolling-horizon procedures for multistage stochastic programming, providing bounds and heuristics validated through power generation planning experiments.

Contribution

It introduces bounds for infinite horizon MSP and heuristics for finite horizon MSP to improve rolling-horizon decision-making under uncertainty.

Findings

Derived a bound for epsilon-sufficient stages in infinite horizon MSP.

Proposed a heuristic for finite horizon MSP stages.

Validated approaches through hydrothermal power planning experiments.

Abstract

Multistage Stochastic Programming (MSP) is a class of models for sequential decision-making under uncertainty. MSP problems are known for their computational intractability due to the sequential nature of the decision-making structure and the uncertainty in the problem data due to the so-called curse of dimensionality. A common approach to tackle MSP problems with a large number of stages is a rolling-horizon (RH) procedure, where one solves a sequence of MSP problems with a smaller number of stages. This leads to a delicate issue of how many stages to include in the smaller problems used in the RH procedure. This paper addresses this question for, both, finite and infinite horizon MSP problems. For the infinite horizon case with discounted costs, we derive a bound which can be used to prescribe an epsilon-sufficient number of stages. For the finite horizon case, we propose a heuristic approach from the perspective of approximate dynamic programming to provide a sufficient number of stages for each roll in the RH procedure. Our numerical experiments on a hydrothermal power generation planning problem show the effectiveness of the proposed approaches.