Data-Driven Based Battery Health Prognosis with Diagnosis Uncertainties and Insufficient Training Data Sets Prof. Zhimin XI (University of Tennessee - Knoxville)
This paper investigates data-driven based battery prognosis with diagnosis uncertainties and insufficient training data sets. Four types of data-driven prognosis methods are investigated including the neural network, similarity-based approach, relevance vector machine, and a recently developed copula-based approach. The remaining useful life (RUL) predictions of lithium-ion battery capacity are compared with capacity estimation error due to the fact that onboard lithium-ion battery capacity estimation is difficult and almost always contains estimation errors. Thus, robustness of each prognosis methods can be studied for real time capacity RUL estimation. Furthermore, collection of sufficient run-to-failure training data sets for lithium-ion batteries is almost impossible even though it is desirable for all data-driven based methods. Therefore, robustness of these methods in terms of the insufficient training data sets is also studied. These insightful results will help designers choose appropriate prognosis algorithms in designing battery management systems (BMS) for lithium-ion batteries.
A Study on the Application of AAKR Based Early Warning System for ICE Mr. Hyungcheol MIN, Heesoo KIM, Seokman SOHN and Yongchae BAE (Korea Electronic Power Research Institute)
Internal Combustion Engine(ICE) is a major type of power generating plant in the islands area and Emergency Diesel Generator(EDG) for nuclear power plants. All the electricity workload in the islands area are supported by diesel engine but due to harsh environmental conditions and lack of manpower, diesel engines in islands are not managed properly. For instance, components of diesel engines are mostly affected by electrical/mechanical problems. Therefore, engines installed in islands area are vulnerable to unexpected failure and replaced earlier than the expected product life.
In this paper, we suggest an early warning algorithm which detects machine breakdown before it happens to prevent unexpected machine failure. By applying this algorithm, we expect a life extension of the diesel engine through proactive and efficient maintenance.
AAKR(Auto Associative Kernel Regression) is a multi-variate state signal estimation model, which is a core algorithm of the early warning system. This algorithm compares saved normal state data with acquired present state data and calculates the weight to create estimation signals.
In this research, we analyzed 8 diesel engine operation data and confirmed that the algorithm works properly. In the future, based on this research result, we plan to build an IoT based diesel engine early warning system.
Perspectives on Using Deep Learning for System Health Management Dr. Samir KHAN and Takeshisa YAIRI (University of Tokyo)
Having a robust health management and diagnostic strategy is an important part of a system’s operational life cycle as it can be used to detect anomalies, analyze faults/failures and predict the remaining useful life of components. By utilizing condition data and on-site feedback, data models can be trained using machine learning and statistical concepts. Once trained, the logic for data processing can be embedded on on-board controllers whilst enabling real-time health assessment and analysis. More recently, deep learning has gained increasing attention due to its potential advantages with data classification and feature extraction problems. It is an evolving research area and hence its use for aerospace applications must been researched if it can be used to increase overall system resilience or potential cost benefits for maintenance, repair, and overhaul activities. This paper focuses on investigating the application of deep learning for system health management, therefore incorporating reliable redundancy at the adequate point in the system. Deep learning is discussed, recent developments are reviewed to clarify potential applications, after which some research issues relating to their realization are highlighted.