| 摘要:以失效概率为出发点，从系统摩擦学角度，考虑摩擦学系统概率、非线性等特性，建立发动机摩擦学可靠性设计方法。以活塞环缸套摩擦学系统为例，建立活塞环缸套宏微观多尺度分析模型，在此基础上，针对窜气量、滑油消耗等性能评价指标，基于机器学习，建立高效的可靠性分析代理模型，并对样本分布、尺寸公差、缸套温度等不同因素对系统失效概率影响进行分析。利用指示平均有效压力（indicated mean effective pressure，IMEP）方法，获取活塞组缸套摩擦力，并基于大数据机器学习，对活塞环缸套摩擦副监测参数缸套温度进行智能预测。研究成果可为活塞环缸套摩擦学系统可靠性设计提供理论指导。 |
| 关键词: 发动机 摩擦学 可靠性设计 大数据 机器学习 |
|
| AI-Empowered Engine Reliability Design |
| Tang Yihu,Liu Jinxing,Wang Fengyan |
| 1Shanghai Marine Diesel Engine Research Institute， Shanghai 201108， China;2National Key Laboratory of Marine Engine Science and Technology， Shanghai 201108， China |
| Abstract:Based on the perspective of failure probability and adopting a system tribology approach that accounts for probabilistic and nonlinear characteristics，a reliability design method for engine tribological systems was established.Taking the piston ring-cylinder liner tribological system as an example，a macro-micro multi-scale analysis model was developed.On this basis，an efficient surrogate model for reliability analysis was constructed using machine learning，targeting performance indicators such as blow-by and oil consumption.The effects of various factors，including sample distribution，dimensional tolerances，and liner temperature，on system failure probability were analyzed.Using the indicated mean effective pressure（IMEP）method，the friction force of the piston assembly-cylinder liner was measured.Based on big data and machine learning，an intelligent prediction of the liner temperature — a key monitoring parameter of the piston ring-cylinder liner tribo-pair — was achieved.The research results provide theoretical guidance for the reliability design of piston ring-cylinder liner tribological systems. |
| Key words: engine tribology reliability design big data machine learning |
