Comfort Evaluation For Footbridge Vibration Based On Neural Network

Modern passenger foot-bridge is developed towards a simple and large-span tendency, which architecture quality and rigidity is decreased to result in diminished self-vibration frequency to foot-bridge, with pedestrian load causing more easily foot-bridge’s vibration, which may be insufficient to cause structural safety problem in one aspect, but could bring uncomfortableness or even fears and confusions to pedestrians. Hence, the amplitude of vibration of the foot-bridge should be taken into consideration and analyzed during the design stage for the foot-bridge, together with the assessment of the comfortableness which may bring to pedestrians. Currently, the evaluation method and index for comfortableness arising from human-made vibration on the foot-bridge are different for domestic and overseas conclusion and there has no confirmed one yet. Thus, this paper has an analysis of the various foot-bridges in different countries and is referenced with the one at square section in Qingquan Road as an example to start research on the comfortableness caused by foot-bridge vibration. The main content of this research is included with the following aspects:Firstly, the current research status and unresolved issues maintained in both domestic and foreign industry are analyzed for foot-bridge vibration. Among these issues, 3 points are mainly included. The first is the study on pedestrian load, which is divided into the single load and the crowd load, with this paper mainly emphasized in modeling method of crowd load and issue simplification; the second is the study on walk parameters including the average weight, step length, stride frequency, leg speed and other parameters for the Chinese people; the third is the study on evaluation method and index for comfortableness, which specifically introduces the five types of the specifications which are widely-adopted internationally on pedestrian bridge; besides, both advantages and disadvantages are analyzed as well for the specifications adopted in different countries; in addition, the evaluation method and index to foot-bridge is confirmed in this paper based on that, that is, to combine the avoidance for the sensitive frequency method with dynamic response restriction method, which is adopted to mainly judge comfortableness according to peak acceleration.Secondly, BP neural network is adopted to predict peak acceleration. Currently, finite element software is widely used in China for analysis purpose and it has also the disadvantages such as difficult pedestrian load modeling and poor universal property. This paper is based on a large amount of the effective research data obtained from the great on-the-spot survey conducted on the foot-bridge at square section in Qingquan Road, Shenzhen; Besides, BP neural network which can be used to predict peak acceleration is also successfully established.Lastly, the current established network can be used to predict this foot-bridge’s peak acceleration under various working conditions; additionally, modal test can be combined to verify the model’s effectiveness, so as to provide references to assess the comfortableness of a foot-bridge in the future.