Model predictive control for maintenance operations planning of railway infrastructures

Abstract

This paper develops a new decision making method for optimal planning of railway maintenance operations using hybrid Model Predictive Control (MPC). A linear dynamic model is used to describe the evolution of the health condition of a segment of the railway track. The hybrid characteristics arise from the three possible control actions: performing no maintenance, performing corrective maintenance, or doing a replacement. A detailed procedure for transforming the linear system with switched input, and recasting the transformed problem into a standard mixed integer quadratic programming problem is presented. The merits of the proposed MPC approach for designing railway track maintenance plans are demonstrated using a case study with numerical simulations. The results highlight the potential of MPC to improve condition-based maintenance procedures for railway infrastructure.