Response attenuation of cable-stayed bridge subjected to central US earthquakes using neuro-fuzzy and simple adaptive control

摘要

Cable-stayed bridges generally present high flexibility, little redundancy, and complex dynamic behavior, which may lead to excessive dynamic responses. Structural control is an option to limit dynamic responses of these structures in order to avoid excessive damage and guarantee an acceptable level of comfort. Most studies in the field of control of cable-stayed bridges are performed considering their nominal parameters. However, bridges are subjected to extreme changes in temperature, cracking, corrosion, snow accumulation, extreme loading and fatigue. Additionally, engineering modeling simplifications, estimates, and assumptions also result in estimation of parameters that are different from the actual ones. Earthquake excitation prediction also involves a great amount of uncertainties. The development of a control scheme that presents satisfactory performance and presents enough robustness is fundamental for its successful operation. The main goal of this study is to fmd control solutions that are able to attenuate seismic induced responses, while providing enough predictability and robustness in face of parametric changes and the variability of earthquake loads. The implementation of a control approach that is cost effective, dependable, predictable and effective may lead to the possibility of accounting for this control solution in design, allowing for more flexible but safe structures. A control approach that is dependable and robust has the potential to guarantee performance limits and impact how structures are designed in the future. Adaptive control is presented in this research as a suitable and robust control alternative to deal with the many uncertainties related to the prediction of bridge structural parameters and earthquake loading. Adaptive control schemes based on the simple adaptive control and the neuro-fuzzy adaptive control theoretical basis are proposed to attenuate the seismic responses of the benchmark for cable-stayed bridges, considering different parametric scenarios, as well as site conditions and seismic characteristics of the central US region. The performances of the adaptive schemes are compared to non-adaptive control before and after two parametric variations are introduced to the bridge, considering earthquakes matched to the American Association of State Highways and Transportation Officials' most recent design spectra. Probability density functions are developed in order to capture the controlled performance representative of the earthquake suite. The adaptive methods performances are compared to uncontrolled and passive schemes before and after two parametric variations are considered. The bridge controlled by the passive-on case presents a satisfactory performance for the nominal structure. However, once the parameters are changed, the performance of the control scheme deteriorates. The passive-off scheme sustains performance well; however, the control scheme does not reduce overall responses significantly. The neuro-fuzzy control displays improved performance in comparison to the passive cases. The simple adaptive control scheme gives an overall successful reduction in both peak and normed responses, sustaining performance for most criteria and shows improved performance and robustness compared to the other schemes.