Distributed task allocation of fleet-level maintenance: Dealing with stochastic durations

In scenarios where systems must execute a series of missions with limited breaks, the difficulty in assigning maintenance activities is of paramount importance. Generally, the efficiency of formation maintenance is dependent of maintenance planning as well as groups scheduling. However, few existing research takes into account the influence of maintenance providers scheduling, which results in inappropriate maintenance decisions. This paper proposes a selective maintenance optimization model for a single equipment system and a fleet-level maintenance task allocation model with random maintenance time. A distributed algorithm is introduced for the addressed fleet-level maintenance task allocation with stochastic maintenance durations. In comparison to those existing models where task durations are assumed to be deterministic, this approach utilizes a selective maintenance model in combination with a decentralized task allocation framework, guaranteeing successful completion of subsequent missions. On basis of a customized heuristic algorithm with an enhanced consensus-based bundle method, the proposed approach seeks the optimal solution to the allocation of maintenance tasks across a fleet by minimizing downtime while maximizing success probability. The effectiveness and validity of the established model is finally verified via a case study involving agricultural harvesters, showing its potential application in dynamic and resource-constrained environments.