基于MLP模型的建筑结构损伤检测与抗震性能分析

Damage Detection and Seismic Performance Analysis of Building Structures Based on MLP Model

针对建筑结构损伤检测和抗震可靠性评估的问题,提出一种基于多层感知器(MLP)神经网络的分析方法 .首先,构建一种钢结构建筑模型,定义多种柱截面积的损伤场景,将加载地震数据时各层最大相对位移作为MLP模型的训练数据.然后,利用MLP模型建立结构响应(输入)与结构刚度(输出)之间的关系,用来进行结构损伤检测.另外,在考虑各层的柱截面积、弹性模量和重力载荷因素下,构建MLP抗震性能评估模型.在一个三层的钢结构上进行了仿真实验,结果表明提出的方法能够准确检测出损伤,并对结构给出了可靠的抗震性能评估.

For the issues of the damage detection and seismic reliability evaluation of building structures,an analysis method based on multi-layer perceptron（MLP）neural network is proposed.Firstly,a steel structure building model is constructed and the damage scenarios of a variety of column cross-sections are defined.The maximum relative displacement of each layer when loading seismic data is used as the training data of the MLP model.Then,the relationship between the structure response（input）and structural stiffness（output）is established by using the MLP model,which is used to detect structural damage.In addition,the MLP seismic performance evaluation model is constructed considering the column cross-sectional area,elastic modulus and gravity load.Simulation results on a three-story steel structure show that the proposed method can accurately detect the damage and provide a reliable assessment of the seismic performance of the structure.