郭凯谱, 李红飞, 范玲玲,等.数据驱动的非线性系统自适应故障诊断方法[J].河南理工大学学报（自然科学版）,2023,42(6):134-141.

GUO K P, LI H F, FAN L L,et al.Data-driven adaptive fault diagnosis method for nonlinear systems[J].Journal of Henan Polytechnic University(Natural Science) ,2023,42(6):134-141.

数据驱动的非线性系统自适应故障诊断方法

郭凯谱1, 李红飞2, 范玲玲1, 吉鸿海3

1.北京信息科技大学 自动化学院，北京   100192；2.国家工业信息安全发展研究中心，北京  100040；3.北方工业大学 电气与控制工程学院，北京  100093

摘要:针对一类离散时间非线性系统，基于数据驱动自适应滤波故障诊断（data-driven adaptive filtering for fault diagnosis，DDAF-FD）方法，实现对执行器和传感器故障的同时在线估计。首先采用动态线性化技术，将非线性系统等价转化为类线性模型，解决非线性系统难以精确建模的问题；其次，仅利用系统I/O数据，在数据驱动滤波和递归最小二乘算法的框架下，设计一种数据驱动自适应故障诊断方法，实现对两种故障失效因子的实时准确估计；最后，利用Lyapunov方法验证所提方法的稳定性，并利用对比仿真实验验证了该方法的有效性。

关键词:数据驱动滤波;动态线性化;故障诊断;最小二乘法

doi:10.16186/j.cnki.1673-9787.2021120122

基金项目:国家自然科学基金资助项目（6170020910）

收稿日期:2021/12/30

修回日期:2022/04/29

出版日期:2023/11/25

Data-driven adaptive fault diagnosis method for nonlinear systems

GUO Kaipu1, LI Hongfei2, FAN Lingling1, JI Honghai3

1.School of Automation，Beijing Information Science and Technology University，Beijing  100192，China；2.National Industrial Information Security Development And Research Center，Beijing  100040，China；3.School of Electrical and Control Engineering，North China University of Technology，Beijing  100093，China

Abstract:For a class of discrete time nonlinear systems，the simultaneous online estimation of actuator and sensor faults was realized based on the data-driven adaptive filtering for fault diagnosis（DDAF-FD） method.Firstly，the dynamic linearization technique was used to transform the nonlinear system into a quasi-linear model，which solved the problem that the nonlinear system was difficult to model accurately.Secondly，only using system I/O data，a data-driven adaptive fault diagnosis method was designed under the framework of data-driven filtering and recursive least squares algorithm，and the real-time accurate estimation of the two fault failure factors was realized.The stability of the proposed method was proved by Lyapunov method.The effectiveness of the proposed method was verified by the comparison of simulation experiment.

Key words:data-driven filtering;dynamic linearization;fault diagnosis;least squares algorithm

017_2021120122_郭凯谱_L.pdf