scholarly journals Fault Detection and Diagnosis Based on Unsupervised Machine Learning Methods: A Kaplan Turbine Case Study

Energies ◽  
2021 ◽  
Vol 15 (1) ◽  
pp. 80
Author(s):  
Miguel A. C. Michalski ◽  
Arthur H. A. Melani ◽  
Renan F. da Silva ◽  
Gilberto F. M. de Souza ◽  
Fernando H. Hamaji
Keyword(s):  
Machine Learning ◽  
Fault Detection ◽  
Failure Modes ◽  
Hydroelectric Power Plant ◽  
Real Data ◽  
Fault Detection And Diagnosis ◽  
Machine Learning Techniques ◽  
Hydroelectric Power ◽  
Unsupervised Machine Learning ◽  
Detection And Diagnosis

From the breakdown of the Kaplan rotor of a hydrogenerator unit and the monitored data collected during its operation before such a failure, this work presents a post-occurrence data analysis in which a previously developed hybrid method based on unsupervised machine learning techniques is applied to detect and diagnose failure before a unit shutdown. In addition to demonstrating the efficiency and capacity of the developed method in an application with real data, the conducted analysis seeks to shed light on the events that occurred at the considered hydroelectric power plant, helping to understand the failure mode evolution and outcome. The results of the fault detection and diagnosis process clearly demonstrated how the evolution of failure modes took place in the analyzed equipment. The detection of potential failures far in advance would support adequate maintenance planning and mitigating actions that could prevent unit breakdown and the consequent damage and financial losses.

scholarly journals Development and Field Evaluation of Data-driven Whole Building Fault Detection and Diagnosis Strategy

2018 ◽  
Vol 10 (1) ◽  
Author(s):  
Yimin Chen ◽  
Jin Wen
Keyword(s):  
Fault Detection ◽  
Principal Component ◽  
Field Evaluation ◽  
Fault Detection And Diagnosis ◽  
Data Driven ◽  
Machine Learning Techniques ◽  
Component Level ◽  
Automation Systems ◽  
Tightly Coupled ◽  
Detection And Diagnosis

Faults, i.e., malfunctioned sensors, components, control, and systems, in a building have significantly adverse impacts on the building’s energy consumption and indoor environment. To date, extensive research has been conducted on the development of component level fault detection and diagnosis (FDD) for building systems, especially the Heating, Ventilating, and Air Conditioning (HVAC) system. However, for faults that have multi-system impacts, component level FDD tools may encounter high false alarm rate due to the fact that HVAC subsystems are often tightly coupled together. Hence, the detection and diagnosis of whole building faults is the focus of this study. Here, a whole building fault refers to a fault that occurs in one subsystem but triggers abnormalities in other subsystems and have significant adverse whole building energy impact. The wide adoption of building automation systems (BAS) and the development of machine learning techniques make it possible and cost-efficient to detect and diagnose whole building faults using data-driven methods. In this study, a whole building FDD strategy which adopts weather and schedule information based pattern matching (WPM) method and feature based Principal Component Analysis (FPCA) for fault detection, as well as Bayesian Networks (BNs) based method for fault diagnosis is developed. Fault tests are implemented in a real campus building. The collected data are used to evaluate the performance of the proposed whole building FDD strategies.

Sensorless Early Detection of Mechanical Faults: Developments in Smart Rotating Machines

2001 ◽  
Author(s):  
Alexander G. Parlos ◽  
Kyusung Kim ◽  
Raj M. Bharadwaj
Keyword(s):  
Fault Detection ◽  
Failure Modes ◽  
Fault Detection And Diagnosis ◽  
False Alarms ◽  
Detection Accuracy ◽  
Driving Dynamic ◽  
Diagnosis System ◽  
Electric Power Supply ◽  
Detection And Diagnosis ◽  
High Level

Abstract Practical early fault detection and diagnosis systems must exhibit high level of detection accuracy and while exhibiting acceptably low false alarm rates. Such designs must have applicability to a large class of machines, require installation of no additional sensors, and require minimal detailed information regarding the specific machine design. Electromechanical systems, such as electric motors driving dynamic loads like pumps and compressors, often develop incipient failures that result in downtime. There is a large number of such failure modes, with a large majority being of mechanical nature. The precise signatures of these failure modes depend on numerous machine-specific factors, including variations in the electric power supply and driven load. In this paper the development and experimental demonstration of a sensorless, detection and diagnosis system is presented for incipient machine faults. The developed fault detection and diagnosis system uses recent developments in dynamic recurrent neural networks in implementing an empirical model-based approach, and multi-resolution signal processing for extracting fault information from transient signals. The signals used by the system are only the multi-phase motor current and voltage sensors, whereas the transient mechanical speed is estimated from these measurements using a recently developed speed filter. The effectiveness of the fault diagnosis system is demonstrated by detecting stator, rotor and bearing failures at early stages of development and during different levels of deterioration. Experimental test results from small machines, 2.2 kW, and large machines, 373 kW and 597 kW, are presented demonstrating the effectiveness of the proposed approach. Furthermore, the ability of the diagnosis system to discriminate between false alarms and actual incipient failure conditions is demonstrated.

Class Incremental Fault Detection and Diagnosis With Unlabeled Condition Monitoring Data

2020 ◽  
Author(s):  
Xiaomeng Peng ◽  
Xiaoning Jin ◽  
Shiming Duan ◽  
Chaitanya Sankavaram
Keyword(s):  
Fault Detection ◽  
Condition Monitoring ◽  
Failure Modes ◽  
Fault Detection And Diagnosis ◽  
Testing Data ◽  
Supervised Classifiers ◽  
On Line ◽  
Complete Failure ◽  
Complete Set ◽  
Detection And Diagnosis

Abstract Data-driven methods for fault detection and diagnostics (FDD) require a large amount of labeled data and knowledge about complete failure modes set to train a reliable classifier as well as require the same label space in an online testing phase. Typical supervised classifiers in FDD can only predict precedented faults, limiting their performance in identifying unprecedented failure modes in on-line testing data. In addition, in most applications, it may be expensive and time-consuming to obtain sufficient labeled samples. This study focuses on fault detection and diagnosis without sufficient labels or prior knowledge of the complete set of failure modes. This paper proposes a novel FDD framework using active learning and semi-supervised learning to detect both precedented and unprecedented failures with minimum labeling effort. The effectiveness of proposed approach is demonstrated and validated using a synthetic condition monitoring dataset.

Sign in / Sign up

Export Citation Format

Share Document