Virtual　coupling　technology　has　been　widely　concerned　due　to　its　potential　in　improving　the　operational　efficiency　of　rail　transit.　To　mitigate　the　effects　of　measurement　error　and　actuation　fault　on　coordinated　train　operation,　this　paper　proposes　a　distributed　fault-tolerant　control　strategy　for　the　virtual　coupling　train　system　by　the　integration　of　moving　horizon　estimation　(MHE)　and　model　predictive　control　(MPC).　Considering　potential　faults　for　virtually　coupled　trains　during　operation,　a　faulty　model　is　developed　to　capture　measurement　errors　and　loss　of　actuator　effectiveness.　A　fault-tolerant　control　strategy　is　designed　that　consists　of　two　modules,　i.e.,　fault　estimation　and　fault-tolerant　control.　The　MHE　method　is　employed　in　the　fault　estimation　module　to　estimate　faults　by　solving　a　local　quadratic　optimization　problem　based　on　historical　data.　In　the　fault-tolerant　control　module,　a　reconfigured　train　tracking　control　problem　is　constructed　under　the　MPC　framework　on　the　premise　of　real-time　estimation.　An　accelerated　alternating　direction　method　of　multipliers　algorithm　is　further　designed　to　embed　with　the　fault-tolerant　model　predictive　controller　to　calculate　the　control　commands　in　a　distributed　manner.　Numerical　experiments　are　conducted　and　demonstrate　the　effectiveness　of　the　proposed　approach.　　©　2000-2011　IEEE.