Effects Of Structural Design Parameters And Ground Motion On Dynamic Response Of High-rise Residential Buildings

Shear wall structures are widely used in residential design because of their advantages such as good integrity,strength and good resistance to lateral deformation.This type of building structure design is short and tasking,with less time for detailed analysis and greater limitations in the parameters that can be changed,mainly through variations in the structural plan layout,the choice of different beam heights and the concrete strength levels.This paper therefore analyses the effect of changes in these parameters on the structural dynamic response of the largest number of practical structures encountered in daily work and which are typically representative of engineering,which is important for efficient analysis of structures.According to structural dynamics theory,the dynamic response of a structure depends on the dynamic characteristics of the structure itself and the spectral characteristics of the input ground vibrations,where the structural dynamic characteristics mainly include frequency(period),vibration pattern and damping.It is proposed to classify the dynamic characteristics of the structure and the spectral characteristics of the input seismic waves by means of cluster analysis techniques.Then,based on the knowledge of structural dynamics,the response of the structure should have some regular and characteristic corresponding dynamic response based on the dynamic characteristics of the structure and the spectral characteristics of the input seismic waves,so we carried out a lot of dynamic analysis on this basis to obtain the dynamic response characteristics with certain types of specific dynamic characteristics under the corresponding ground vibration input.Based on these structural dynamic characteristics and ground vibration input characteristics,a neural network-based inference machine is studied for rapid structural response assessment and structural design parameter optimisation.The main research elements are as follows:(1)Feature extraction of structural dynamic characteristics and the influence of design parameters on the dynamic characteristics.Parameter variations are carried out for a 22-storey residential structure: different structural plan arrangements,uniform adjustment of beam heights as well as concrete strengths are used to establish different kinds of computational models,which are subsequently analysed modally.The self-oscillation period curves were significantly different when the structural plan was varied,with a large difference in amplitude,whereas the self-oscillation period curves did not change much when the beam height and concrete strength were adjusted for the structure.The self-oscillation period curve can reflect the influence on the structural dynamic characteristics when the structural design parameters are changed,so it is feasible to use the self-oscillation period data as the extraction of structural dynamic characteristics.(2)The structural dynamic characteristics as well as the input ground vibration characteristics were analysed by clustering.Firstly,a 22-storey residential structure was subjected to parameter changes: different structural plan layouts,uniform adjustment of beam heights and concrete strengths,whereby different kinds of computational models were established,followed by modal analysis,and the self-oscillation cycle data sets of 180 models were collected.k-mean cluster analysis was applied to the collected self-oscillation cycle data sets of 180 models to classify them into five classes of cycles(i.e.classes A,B The wavelet transformed images were then classified into 6 categories(i.e.,1,2,3,4,5,6),and each category of wavelet transformed images had obvious features.(3)Construction and application of neural network inference machine.Based on the classification data after clustering analysis,the inference machine of the neural network was built.The model converged after nearly 650 steps of algorithm,the training curve was smooth and the convergence speed was fast,indicating that the established neural network worked well and the comprehensive accuracy of the model reached 94.7%.Finally,the inference machine is applied to realise the optimisation of the structure when the design parameters change,and the fast prediction of the classification to which the maximum displacement of the top layer of the structure belongs under different ground vibrations and different structural models.