System Identification I

1D convolutional neural networks and applications: A survey

During the last decade, Convolutional Neural Networks (CNNs) have become the de facto standard for various Computer Vision and Machine Learning operations. CNNs are feed-forward Artificial Neural Networks (ANNs) with alternating convolutional and subsampling layers. Deep 2D CNNs with many hidden layers and millions of parameters have the ability to learn complex objects and patterns providing that they can be trained on a massive size visual database with ground-truth labels. With a proper training, this unique ability makes them the primary tool for various engineering applications for 2D signals such as images and video frames. Yet, this may not be a viable option in numerous applications over 1D signals especially when the training data is scarce or application-specific. To address this issue, 1D CNNs have recently been proposed and immediately achieved the state-of-the-art performance levels in several applications such as personalized biomedical data classification and early diagnosis, structural health monitoring, anomaly detection and identification in power electronics and motor-fault detection. Another major advantage is that a real-time and low-cost hardware implementation is feasible due to the simple and compact configuration of 1D CNNs that perform only 1D convolutions (scalar multiplications and additions). This paper presents a comprehensive review of the general architecture and principals of 1D CNNs along with their major engineering applications, especially focused on the recent progress in this field. Their state-of-the-art performance is highlighted concluding with their unique properties. The benchmark datasets and the principal 1D CNN software used in those applications are also publically shared in a dedicated website.

1D Convolutional Neural Networks and Applications: A Survey

During the last few decades, magnetorheological (MR) elastomers have attracted a significant amount of attention for their enormous potential in engineering applications. Because they are a solid counterpart to MR fluids, MR elastomers exhibit a unique field-dependent material property when exposed to a magnetic field, and they overcome major issues faced in magnetorheological fluids, e.g. the deposition of iron particles, sealing problems and environmental contamination. Such advantages offer great potential for designing intelligent devices to be used in various engineering fields, especially in fields that involve vibration reduction and isolation. This paper presents a state of the art review on the recent progress of MR elastomer technology, with special emphasis on the research and development of MR elastomer devices and their applications. To keep the integrity of the knowledge, this review includes a brief introduction of MR elastomer materials and follows with a discussion of critical issues involved in designing magnetorheological elastomer devices, i.e. operation modes, coil placements and principle fundamentals. A comprehensive review has been presented on the research and development of MR elastomer devices, including vibration absorbers, vibration isolators, base isolators, sensing devices, and so on. A summary of the research on the modeling mechanical behavior for both the material and the devices is presented. Finally, the challenges and the potential facing magnetorheological elastomer technology are discussed, and suggestions have been made based on the authors’ knowledge and experience.

/pdf/a-state-of-the-art-review-on-magnetorheological-elastomer-4upfpmpbpd.pdf

A state-of-the-art review on magnetorheological elastomer devices

A review of vibration-based damage detection in civil structures : from traditional methods to Machine Learning and Deep Learning applications

In the past few years, intelligent structural damage identification algorithms based on machine learning techniques have been developed and obtained considerable attentions worldwide, due to the ad...

https://journals.sagepub.com/doi/pdf/10.1177/1475921718804132

A novel deep learning-based method for damage identification of smart building structures:

https://opus.lib.uts.edu.au/bitstream/10453/111482/3/inverse%2bmodel_revision%2b%28XG%29.pdf

Semi-active control of magnetorheological elastomer base isolation system utilising learning-based inverse model

Experimental study of semi-active magnetorheological elastomer base isolation system using optimal neuro fuzzy logic control

In recent years, an adaptively tuned magnetorheological elastomer (MRE) isolator for a base isolation system has been designed and tested with the benefits of low power cost, fail safe manner and fast responses. To make full use of this striking device for design of smart structures, a highly precise model should be developed to effectively and accurately forecast the shear force of the device in real-time so as to adopt a proper control strategy to improve the responses of the protected structures. In this work, a novel mechanical model is presented to characterize this nonlinear hysteresis for its implementation in structural vibration control. This model employs the displacement and velocity of the device as well as the applied current as the inputs and just has the limited constant parameters to be identified compared with some classical hysteretic models such as Bouc–Wen, improved Dahl and LuGre models. Performance evaluation of this novel hysteresis model has been conducted based on the testing data from an MRE base isolator. The results show that the proposed model has high modelling accuracy and is able to perfectly portray the unique and complicated behaviours of the device with various excitations.

http://iopscience.iop.org/article/10.1088/0964-1726/25/5/055029/pdf

A hysteresis model for dynamic behaviour of magnetorheological elastomer base isolator

Utilising the unique features of MRE materials for vibration isolators has been intensively studied over the last several years. Real-time control of the MRE isolators holds the key to unlock MRE materials' unique characteristics, i.e. instantly changeable shear modulus in continuous and reverse fashion. However, one of the critical issues for the applications of real-time control is the response time delay of MRE vibration isolators, which has not yet been fully addressed and studied. This paper identified the inherent response time of the MRE isolator and explored two feasible approaches to minimise the response time delay. Experiments were designed and conducted to evaluate the effectiveness of the proposed approaches on minimising time delay on: (i) the transient response of current of a large coil that generates magnetic field and (ii) the transient response of shear force from the MRE isolator. The results show that the proposed approaches are effective and promising. For example, the proposed approach is able to reduce the force response time from 421 ms to 52 ms at rising and from 400 ms to 48 ms falling edges respectively. Such level of short response time of the MRE isolators demonstrates the feasibility of application of real-time control and hence is the essential step on the realisation of real-time control of vibration suppression system based on MRE isolator.

/pdf/investigations-on-response-time-of-magnetorheological-1sfn7poug9.pdf

Xiaoyu Gu

Papers

A novel deep learning-based method for damage identification of smart building structures:

Semi-active control of magnetorheological elastomer base isolation system utilising learning-based inverse model

Experimental study of semi-active magnetorheological elastomer base isolation system using optimal neuro fuzzy logic control

A hysteresis model for dynamic behaviour of magnetorheological elastomer base isolator

Investigations on response time of magnetorheological elastomer isolator for real-time control implementation