A Distributed Semi-Consensus-Based Data-Driven Fault Detection Approach for Dynamic Systems

Linlin Li

doi:10.1109/TII.2024.3495774

A Distributed Semi-Consensus-Based Data-Driven Fault Detection Approach for Dynamic Systems

2024-01-01

0

PRE

AI

DOI：10.1109/TII.2024.3495774

原文链接

原文求助

摘要

En 中文

In this article, a distributed semi-consensus-based data-driven fault detection scheme is developed based on the process variables collected by sensor networks to ensure the safety of the large-scale dynamic processes. For our purpose, the distributed data-driven process modeling scheme is developed for dynamic systems first by considering the communication topology of the sensor networks. Then, a distributed Kalman filter-based fault detection approach is developed aiming at achieving optimal detection performance at each sensor node. Specifically, the distributed iterative learning algorithm is implemented to calculate the needed parameter matrices for Kalman filter-based residual generator offline with the aid of average consensus algorithm. It is followed by a distributed fusion of local residual signals to perform online optimal fault detection. To avoid the detection delay caused by the traditional average consensus method, the semi-consensus algorithm is developed for the first time to ensure the timely detection of potential faults. A case study on the multiphase flow facility process is given in the end to demonstrate the proposed method.

Keyword:

Kalman filters

Fault detection

Dynamical systems

Delays

Topology

Consensus algorithm

Training

Safety

Process modeling

Network topology

Distributed Kalman filter

distributed fault detection (FD)

residual generator

semi-consensus algorithm

A Distributed Semi-Consensus-Based Data-Driven Fault Detection Approach for Dynamic Systems

摘要

期刊

机构

引用论文