船舶柴油机是目前普遍采用的船舶原动机设备，被视为船舶的心脏，为全船提供推进动力。柴油机系统层次多、结构复杂，包含大量的摩擦学系统，导致许多零部件在运行过程中产生摩擦磨损故障，降低船舶的安全性。智能故障诊断能够及时排除故障，预防事故产生，但是面对磨损故障模式复杂多样、诊断中存在强不确定性及磨损故障状态动态变化等问题，当前智能诊断方法显得适应性不足且效果欠佳。因此，本项目以动态置信规则推理为基础，联合概念漂移检测与更新、多目标优化等方法，以船舶柴油机主要摩擦学系统为对象，开展船舶柴油机磨损故障智能诊断研究：(1) 在幂集辨识框架下，建立适用于单一、耦合故障的置信规则推理诊断模型；(2) 综合采用动态置信规则推理建模、多模型集成和动态权重更新机制实现诊断模型更新，从而对故障状态漂移进行响应； (3) 提出静、动态磨损故障诊断模型结构与参数协同的多目标优化方法，提高诊断准确性，并控制模型复杂性。

Marine diesel engine is the most common used prime mover, which is considered as the heart of ships, generating the power for propulsion. Marine diesel engine consists a lot of tribological systems with complex structure, therefore friction and wear faults will occur in many parts and systems during engine operation, decreasing ship safety. Intelligent fault diagnosis can identify faults timely to avoid severe accidents. However, the current intelligent fault diagnostic methods cannot well solve the problems in wear fault diagnosis, including the complexity of wear fault modes, the strong uncertainty in fault diagnosis, and the dynamic variation of fault condition. To solve these problems, this project will choose the main tribological systems of marine diesel engines as research objects, and use dynamic belief rule based inference methodology, concept drift detection and update, and multi-objective optimization to study the intelligent wear fault diagnosis of marine diesel engines: (1) under the power set framework, develop the fault diagnostic model on the basis of belief rule based inference methodology for single and coupling wear faults; (2) update the fault diagnostic model by synthetically using the dynamic belief rule based inference methodology, multi-models integration, and dynamic weight update to response to fault condition drift; (3) propose a multi-objective optimization model synergistically optimizing the structure and parameters of fault diagnostic model to keep a balance between model accuracy and complexity.