一种发动机热态衰减阶段的故障诊断集成学习方法  

An Integrated Learning Method for Fault Diagnosis of Thermal Decay Phase of Engine

作　　者：郭崇 陈宇峰[1] 陈根浪[2] GUO Chong;CHEN Yufeng;CHEN Genlang(Hubei Institute of Automotive Technology,Shiyan Hubei 442002,China;Ningbo Institute of Technology,Zhejiang University,Ningbo Zhejiang 315000,China)

机构地区：[1]湖北汽车工业学院,湖北十堰442002 [2]浙大宁波理工学院,浙江宁波315000

出　　处：《佳木斯大学学报(自然科学版)》2024年第12期37-40,共4页Journal of Jiamusi University:Natural Science Edition

摘　　要：针对汽车发动机在热态衰减阶段的故障诊断问题,提出了基于隔离森林算法和随机森林算法的故障诊断集成学习方法,隔离森林算法通过构建孤立树来识别异常点,随机森林算法构建多个决策树并综合它们的预测结果来进行异常分类。实验结果表明,融合隔离森林和随机森林算法的集成学习模型比传统的基线模型在发动机数据异常检测和分类的故障检测任务中具备更高的准确性、鲁棒性和应用潜力。In view of the automobile engine in the stage of thermal decay fault diagnosis,this paper puts forward the fault diagnosis of integrated learning method based on isolation forest algorithm and random forest algorithm,isolated forest algorithm by constructing isolation tree to identify anomaly,random forest algorithm to build multiple decision tree and synthesize their prediction results to anomaly classification.The experimental results show that the ensemble learning model with the random forest algorithm has higher accuracy,robustness and application potential than the traditional baseline model in the fault detection task of anomaly detection and classification of engine data.

关 键 词：故障诊断 SMOTE算法 隔离森林 随机森林 

分 类 号：TP399[自动化与计算机技术—计算机应用技术]