Lijie Yu

Lijie Yu

Auto-generated profile, based on publicly available information

Lijie Yu

General Electric Power & Water Engineering 30339 Atlanta Georgia USA

Research Focus & Metrics

Academic Career Length

14Years (2013-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
18 October 2015 in Annual Conference of the PHM Society
Verified
Journal Article
Open access
PDF file
WYWeizhong Yan
Weizhong Yan
LYLijie Yu
Lijie Yu
Monitoring gas turbine combustors' health, in particular, early detecting abnormal behaviors and incipient faults, is critical in ensuring gas turbines operating efficiently and in preventing costly unplanned maintenance. One popular means of detecting combustors’ abnormalities is through continuously monitoring exhaust gas temperature profiles. Over the years many anomaly detection technologies have been explored for detecting combustor faults, however, the performance (detection rate) of anomaly detection solutions fielded is still inadequate. Advanced technologies that can improve detection performance are in great need. Aiming for improving anomaly detection performance, in this paper we introduce recently-developed deep learning (DL) in machine learning into the combustors’ anomaly detection application. Specifically, we use deep learning to hierarchically learn features from the sensor measurements of exhaust gas temperatures. And we then use the learned features as the input to a neural network classifier for performing combustor anomaly detection. Since such deep learned features potentially better capture complex relations among all sensor measurements and the underlying combustors’ behavior than handcrafted features do, we expect the learned features can lead to a more accurate and robust anomaly detection. Using the data collected from a real-world gas turbine combustion system, we demonstrated that the proposed deep learning based anomaly detection significantly indeed improved combustors’ anomaly detection performance.Deep learning, one of the breakthrough technologies in machine learning, has attracted tremendous research interests in recent years in the domains such as computer vision, speech recognition and natural language processing.Deep learning, to the best of our knowledge, has not been used for any PHM applications, however. It is our hope that our initial work presented in this paper would shed some light on how deep learning as an advanced machine learning technology can benefit PHM applications and, more importantly, can stimulate more research interests in our PHM community.

3 October 2016 in Annual Conference of the PHM Society
Verified
Journal Article
Open access
PDF file
LYLiJie Yu
LiJie Yu
SSSachin Shrivastava
Sachin Shrivastava
Compressor blades of a heavy duty industrial gas turbine need to sustain long period mechanical stress and vibration induced by high speed rotation and high pressure mass flow. High stress coupled with erosion and corrosion damage during operation is the main driver for blade cracking. Material separation of cracked rotating blade is a serious safety and reliability concern, which not only affects compressor health, but may also cause costly secondary damage at downstream. Early detection of blade anomaly and incipient crack is critical to ensure blade and compressor health and minimize service disruption. In this paper we will introduce a blade health monitoring (BHM) system developed by GE Power. BHM adopts distributed system architecture and operates continuously 24x7 to provide real time rotor blade health assessment. BHM sensors and data acquisition (DAQ) system are installed on the gas turbine to capture blade passing signals (BPS) and assess time of arrival (TOA) for each blade. Advanced signal processing algorithms process the signals locally to calculate key features that associated with blade health. Then finally, a central anomaly detection module, which is fully integrated with GE Power monitoring system, is developed to assess blade health condition and generate anomaly alarms to alert diagnostic engineer.

25 August 2019 in arXiv
Verified
Preprint
Open access
PDF file
WYWeizhong Yan
Weizhong Yan
LYLijie Yu
Lijie Yu
Monitoring gas turbine combustors health, in particular, early detecting abnormal behaviors and incipient faults, is critical in ensuring gas turbines operating efficiently and in preventing costly unplanned maintenance. One popular means of detecting combustor abnormalities is through continuously monitoring exhaust gas temperature profiles. Over the years many anomaly detection technologies have been explored for detecting combustor faults, however, the performance (detection rate) of anomaly detection solutions fielded is still inadequate. Advanced technologies that can improve detection performance are in great need. Aiming for improving anomaly detection performance, in this paper we introduce recently-developed deep learning (DL) in machine learning into the combustors anomaly detection application. Specifically, we use deep learning to hierarchically learn features from the sensor measurements of exhaust gas temperatures. And we then use the learned features as the input to a neural network classifier for performing combustor anomaly detection. Since such deep learned features potentially better capture complex relations among all sensor measurements and the underlying combustor behavior than handcrafted features do, we expect the learned features can lead to a more accurate and robust anomaly detection. Using the data collected from a real-world gas turbine combustion system, we demonstrated that the proposed deep learning based anomaly detection significantly indeed improved combustor anomaly detection performance.

2 October 2017 in Annual Conference of the PHM Society
Verified
Journal Article
Open access
PDF file
XJXiaomo Jiang
Xiaomo Jiang
LYLijie Yu
Lijie Yu
KMKaren Miller
Karen Miller
This paper develops an innovative integrated methodology for turbomachinery event detection and prediction by using multivariate noisy operation data. The method seamlessly integrates probabilistic method with multiple advanced analytics techniques, including wavelets and entropy information theory. Wavelets based multi-scale principal component analysis is employed to de-noise the raw data for each tag/variable. Probabilistic principal components analysis is further developed to extract useful information from multiple corrected variables, and entropy information feature is extracted as a precursor of the event, the measure of disorder in a thermodynamic system. The proposed method is so-called Wavelet PCA Entropy. The method considers uncertainty in multivariate data, and provides proof-of-concept of advanced analytics for prediction of challenging events in turbomachinery. The feasibility of the presented methodology is demonstrated with the prediction of combustor lean blow out event and data collected from a real-world gas turbine. This study provides a novel intelligent approach to turbomachinery damage diagnostics and prognostics.

14 October 2013 in Annual Conference of the PHM Society
Verified
Journal Article
Open access
PDF file
WYWeizhong Yan
Weizhong Yan
LYLijie Yu
Lijie Yu
JSJim Sherbahn
Jim Sherbahn
UBUmang Brahmakshatriya
Umang Brahmakshatriya
Gas turbine health monitoring is a critical process in preventing costly unplanned maintenance and secondary damage. To monitor gas turbine health, control signals are typically collected and analyzed using anomaly detection rules and models to assess failure likelihood based on observed data patterns. An analytic designer will often deal with rule optimization tasks in order to maximize failure detection and reduce false alarms. Manual tradeoff analysis is typically time consuming and suboptimal. In this paper, we attempt to address this issue by introducing a strategy for automatic and efficient rule optimization. By focusing on optimizing rule parameters while keeping rule structure intact, we maximize the rule performance by integrating domain knowledge with data driven optimization techniques. Realizing that automated rule tuning can be computationally expensive and infeasible to complete in reasonable time, we will leverage our recently-developed scalable learning framework - iScale that allows for automatically distributing rule tuning tasks to a large number of cloud computers, which not only dramatically speeds up tuning process, but also enables us to handle big size of historical data for tuning. We also explore different search methods to make rule tuning more efficient and effective and finally demonstrate our rule optimization strategy by a real-world application.