Data-driven fault diagnosis and prognosis for process faults using principal component analysis and extreme learning machine

Qi, Ruosen; Jie, Zhang

doi:10.1109/indin45582.2020.9442177

Cited by 2 publications

(2 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Similarly, a PCA model was used to formulate fault direction matrix for fault detection, and data-driven fault prognosis models were reconstructed by fault magnitudes. 30 Fault reconstruction and fault magnitude prognosis for process faults was executed by both linear autoregressive and extreme learning machine (ELM) models.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Sensor fault-tolerant attitude determination system based on the nonlinear interacting-multiple-model approach

Akbarinia

Ousaloo

2022

Proceedings of the Institution of Mechanical Engineers, Part G:

View full text Add to dashboard Cite

This paper presents a methodology for satellite sensors' fault detection and diagnosis (FDD) in a fault-tolerant attitude determination system using a nonlinear interacting-multiple-model approach (IMM). Moreover, the fault-tolerant attitude determination system utilizes the IMM approach in the presence of sensors' total fault to estimate faulty sensors and attitude states. This methodology relies on the most conventional satellite attitude sensors including magnetometer, sun sensor, and gyroscope measurements. The presented scheme leads to a significant computational time-saving method due to considering a pre-filter Kalman for fault detection, hence, the proposed FDD approach consists of two stages. In the first stage, the pre-filter Kalman detects any sensor fault and attitude states. In contrast, in the next stage, fault diagnosis and optimal attitude estimation are obtained through the IMM approach. Therefore, the IMM algorithm will be activated to extract the optimal estimation and identify faulty sensor(s). A Monte Carlo type approach is used to verify this technique for various initial conditions. Through numerical and Monte Carlo simulations, it is shown that the proposed scheme is robust against sensors' total fault, and the system is capable of detecting and isolating accurately the faulty sensor. Hardware in the loop tests using satellite dynamics simulator are also used to validate the presented system performance.

show abstract

Section: Introductionmentioning

confidence: 99%

“…In Ref. 30, results of an FDI system employed to outlet temperature sensors of a double pipe heat exchanger were presented using a fractional nonlinear strategy based on a bank of two fractional high gain observers. Garcia et al have presented a fault-tolerant attitude determination system for the NanosatC-BR2 satellite.…”

Section: Introductionmentioning

confidence: 99%