Markov决策过程在船舶结构维修中的应用

船舶腐蚀和疲劳是导致船体老化,进而使其逐渐丧失结构承载能力的重要原因.本文采用具有无后效性的Markov链来模拟船体结构老化过程,其状态空间由完好(不需要维修)、可见裂纹维修、油漆和阴极防护维修、腐蚀修理、疲劳裂纹修理、腐蚀疲劳组合修理和完全失效共计七种状态组成.每一状态均可采取无修理、修理和更新三者之一的维修方案.通过迭代来获得最佳维修方案以实现单位时间成本最低的维修优化目标.算例表明,Markov链可以合理地模拟船舶结构的老化过程,并可方便快捷地运用于船舶维修经济论证.

The Application of Markov Decision Process in Ship Hull Maintenance

Ship corrosion and fatigue are two important factors which result in ship hull degradation and the loss of load-carrying capacity. In this paper, Markov Decision Process(MDP) is applied to simulate ship hull deterioration, whose state space consists of 7 states in accordance with deterioration extent of ship hull, i.e. perfection, noticeable crack repair, corrosion protection system repair, corrosion repair, fatigue crack repair, both corrosion and fatigue repair, and failure. With regard to each state, one of three maintenance approaches, e.g. no repair, repair and replacement, is utilized. The optimal maintenance policy is determined by means of maintenance policy iteration process in order to achieve the minimum long-run cost rate. An example is presented to illustrate the application of MDP in ship hull maintenance and its advantage in economical assessment. The importance of parameter determination in MDP is stressed and effort will be made to build up maintenance database.