Characterization of novel variable friction device and development of data driven controller in intelligent semi-active control system for multi-hazard mitigation capabilities

Thumbnail Image
Cao, Liang
Major Professor
Simon Laflamme
Committee Member
Journal Title
Journal ISSN
Volume Title
Research Projects
Organizational Units
Journal Issue
Is Version Of
Civil, Construction, and Environmental Engineering

Civil infrastructures, including buildings and energy, lifeline, communication, and transportation systems, provide significant services and benefits to our communities. These systems need to be designed, constructed, and maintained to sufficiently resist the effects of service and extreme loads to ensure continuous daily operability and public safety. A solution to increase structural performance vis-a-vis service and extreme loads is a performance-based design (PBD) approach. The strategy of PBD is to appropriately size structural stiffness and supplemental damping for a given system to restrict motion to a prescribed performance. In the case where there are multiple excitations either individually considered or combined, termed multi-hazard, a PBD approach becomes difficult to implement with passive strategies as they require a high level of redundancy. Semi-active control systems can perform over a wide excitation bandwidth, ideal for multi-hazard. However, this technology is not yet widely accepted by the field nor implemented. Several factors are impeding its application: 1) lack of mechanically reliable damping device with large resisting forces capability; 2) large uncertainty in the dynamic parameters of large-scale structures; and 3) lack of robust controller for unknown excitations.

In order to enhance acceptability of semi-active control systems in civil field, an intelligent semi-active control system for multi-hazard mitigation is proposed. This control systems comprises a novel semi-active friction device and a new data driven controller.

The novel semi-active friction device, termed Modified Friction Device (MFD), has enhanced applicability compared to other proposed damping systems due to its cost-effectiveness, high damping performance, mechanical robustness, and technological simplicity. The promise of the MFD has been shown theoretically before. Its mechanical principle is based on a duo-servo drum brake, which results in a high amplification of the input force while enabling a variable control force. Here, we fabricate the first prototype of the MFD and experimentally demonstrate its principle. A three-stage dynamic model is proposed to characterize its dynamics. The proposed model can be used for performance-based design and to develop effective control algorithms.

The new data driven controller, termed input space dependent controller (ISDC), is based on real-time identification of an embedding that represents the essential dynamics found in the input space, or in the measurements. The ISDC is an excellent candidate for multi-hazard applications, because it 1) utilizes local and limited measurements only; 2) does not require prior evaluation or training; 3) is capable of extracting key features from unknown excitation; and 4) adapts to systems nonstationarities.

The intelligent semi-active control system is simulated on two representative structures, one short building located in Japan, and one tall building located in Boston, MA, with both structures being subjected to non-simultaneous multihazard loads. Various control cases are considered, including passive, sliding mode and the proposed ISDC. Results show that the proposed semi-active control system can effectively mitigate all different hazards based on limited and local measurements.

This dissertation is based upon work supported by the National Science Foundation under Grant No. 1300960. Their support is gratefully acknowledged. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author and do not necessarily reflect the views of the National Science Foundation.

Subject Categories
Fri Jan 01 00:00:00 UTC 2016