Active learning for Structural Health Monitoring
Lead Research Organisation:
University of Sheffield
Department Name: Mechanical Engineering
Abstract
The main aim of Structural Health Monitoring (SHM) is the development of diagnostic systems capable of detecting structural degradation in real time, including: damage detection, localisation and quantification, prognosis (remaining life) and detection of performance-related anomalies.
Machine learning techniques have become increasingly important to the field of SHM. Advanced pattern recognition algorithms can now learn complicated relationships from system data, whilst providing a variety of approaches to novelty detection online and in real-time. In the data-based approach to SHM, the absence of data from damaged structures in many cases forces a dependence on novelty detection as a means of diagnosis. Unfortunately, this means that benign variations in the operating or environmental conditions of the structure must be handled very carefully, lest they lead to false alarms.
Unfortunately, complex systems prove costly to test in the laboratory, whilst also having a vast range of potential damage modes. Therefore, retrieving training data to aid wholly supervised methods becomes infeasible; consequently, a high demand for adaptive, semi-supervised learning methods exists within the context of SHM.
Active learning is a form of semi-supervised, or 'query' learning; its main premise is that an algorithm can achieve greater accuracy from fewer training labels if it can choose the data from which it learns. This eliminates the time-consuming process of labelling large volumes of trivial data, as well as significantly reducing the need for extensive sets of training data. Following unsupervised clustering techniques, an active learner then poses queries relating only to unlabelled instances for which it is uncertain - these data points are then identified by an oracle.
With the intention of utilising active learning as an adaptive intelligence in an online environment, the planned work throughout the duration of this PhD aims to provide significant advances in machine learning techniques currently used by the SHM community.
Through collaborations between the University of Sheffield and Vatenfall, the dynamics research group has recently gained access to additional diagnostic data - gathered over a one year period - from the Lillgrund offshore wind farm in Denmark. This large dataset provides an abundance of potential research routes, as well as the further opportunities to test and validate any methods with real data. Analysis will be carried out on a variety of data types and sources, and as a result, the techniques developed throughout this PhD can be applied to a range of real-world problems in the Structural Health Monitoring domain.
Machine learning techniques have become increasingly important to the field of SHM. Advanced pattern recognition algorithms can now learn complicated relationships from system data, whilst providing a variety of approaches to novelty detection online and in real-time. In the data-based approach to SHM, the absence of data from damaged structures in many cases forces a dependence on novelty detection as a means of diagnosis. Unfortunately, this means that benign variations in the operating or environmental conditions of the structure must be handled very carefully, lest they lead to false alarms.
Unfortunately, complex systems prove costly to test in the laboratory, whilst also having a vast range of potential damage modes. Therefore, retrieving training data to aid wholly supervised methods becomes infeasible; consequently, a high demand for adaptive, semi-supervised learning methods exists within the context of SHM.
Active learning is a form of semi-supervised, or 'query' learning; its main premise is that an algorithm can achieve greater accuracy from fewer training labels if it can choose the data from which it learns. This eliminates the time-consuming process of labelling large volumes of trivial data, as well as significantly reducing the need for extensive sets of training data. Following unsupervised clustering techniques, an active learner then poses queries relating only to unlabelled instances for which it is uncertain - these data points are then identified by an oracle.
With the intention of utilising active learning as an adaptive intelligence in an online environment, the planned work throughout the duration of this PhD aims to provide significant advances in machine learning techniques currently used by the SHM community.
Through collaborations between the University of Sheffield and Vatenfall, the dynamics research group has recently gained access to additional diagnostic data - gathered over a one year period - from the Lillgrund offshore wind farm in Denmark. This large dataset provides an abundance of potential research routes, as well as the further opportunities to test and validate any methods with real data. Analysis will be carried out on a variety of data types and sources, and as a result, the techniques developed throughout this PhD can be applied to a range of real-world problems in the Structural Health Monitoring domain.
Organisations
Studentship Projects
Project Reference | Relationship | Related To | Start | End | Student Name |
---|---|---|---|---|---|
EP/N509735/1 | 01/10/2016 | 30/09/2021 | |||
1788975 | Studentship | EP/N509735/1 | 01/10/2016 | 30/09/2019 | Lawrence Bull |
Description | A data-driven approach is often used to implement Structural Health Monitoring (SHM); that is, monitoring measured data from a structure or system, in order to define an SHM strategy. Unfortunately, the data recorded from structures in practice are rarely comprehensive. For example, consider data recorded from a wind-turbine in operation, in an offshore wind-farm; while an engineer might have an abundance of measured data, it isn't clear what the measurements represent: is the system operating normally, under extreme environmental conditions, or (more importantly) is the system damaged? This project considered the application of novel methods for monitoring such data online, recorded from systems in operation. Through the application of active learning methods (from the field of machine learning), algorithms were used to monitor data and define when is it most appropriate for the engineer to go and investigate the structure, to describe abnormalities in the measurements. This is significant when the costs associated with investigating the structure are high; for example, an unscheduled visit to the remote ocean site of an offshore wind-farm. Furthermore, the use of semi-supervised learning allows for monitoring algorithms to make use of incomplete label information, in order to improve the diagnostic capabilities of predictive models. Another relevant set of tools, considering the costs associated with inspecting and labelling measured data. |
Exploitation Route | Alternative implementations and algorithms could be investigated in various applications to practical engineering datasets. Including applications outside of Structural Health Monitoring. |
Sectors | Energy,Manufacturing, including Industrial Biotechology,Transport |
Description | Dynamics Research Group Showcase and LVV Laboratory Launch Event 2019 |
Form Of Engagement Activity | Participation in an activity, workshop or similar |
Part Of Official Scheme? | No |
Geographic Reach | International |
Primary Audience | Industry/Business |
Results and Impact | Co-organised the Dynamics Research Group showcase - including lab tours and a poster display of research activities. The annual event aims to encourage engagement with industry and external academia |
Year(s) Of Engagement Activity | 2019 |
Description | Farnborough Airshow 2019 |
Form Of Engagement Activity | Participation in an activity, workshop or similar |
Part Of Official Scheme? | No |
Geographic Reach | International |
Primary Audience | Industry/Business |
Results and Impact | Presented engineering applications airshow attendees and members of industry. Additionally, new laboratory facilities were advertised, to encourage testing and collaboration. |
Year(s) Of Engagement Activity | 2019 |
Description | The 12th International Workshop on Structural Health Monitoring (IWSHM), Stanford, USA, 2019 |
Form Of Engagement Activity | Participation in an activity, workshop or similar |
Part Of Official Scheme? | No |
Geographic Reach | International |
Primary Audience | Other audiences |
Results and Impact | Conference |
Year(s) Of Engagement Activity | 2019 |
Description | The 36th International Modal Analysis Conference, Orlando, USA, January 2018 |
Form Of Engagement Activity | Participation in an activity, workshop or similar |
Part Of Official Scheme? | No |
Geographic Reach | International |
Primary Audience | Other audiences |
Results and Impact | Conference |
Year(s) Of Engagement Activity | 2018 |
Description | The 8th International Conference on Recent Advances in Structural Dynamics (RASD), Lyon, France, April 2019. |
Form Of Engagement Activity | Participation in an activity, workshop or similar |
Part Of Official Scheme? | No |
Geographic Reach | International |
Primary Audience | Other audiences |
Results and Impact | Conference |
Year(s) Of Engagement Activity | 2019 |
Description | The 9th European Workshop on Structural Health Monitoring (EWSHM), Manchester, UK, July 2018 |
Form Of Engagement Activity | Participation in an activity, workshop or similar |
Part Of Official Scheme? | No |
Geographic Reach | International |
Primary Audience | Other audiences |
Results and Impact | Conference |
Year(s) Of Engagement Activity | 2018 |