基于模式识别的桥梁故障检测  

An Identification Method for Bridge Based on Pattern Recognition

作　　者：周波[1] 杨小帆[1] 王磊[1] 

机构地区：[1]重庆大学计算机学院

出　　处：《计算机与数字工程》2010年第4期173-175,共3页Computer & Digital Engineering

摘　　要：文章主要任务是研究如何从挠度数据出发,综合运用模式识别技术和数据分析技术,能有效检测桥梁中某些类型的损伤和隐患。将模式识别中的近邻算法与K-均值算法相结合,应用到桥梁检测中,提出了一种新的桥梁异常检测方法。首先,用K均值算法对数据进行聚类分析,寻找数据之间的相似性,识别出模式,将桥梁挠度数据归类;然后,在分析阶段运用改进的近邻算法进行分析检测。由于最近邻和k-近邻的交替使用,保证了检测结果的准确率。将上述方法应用于重庆菜园坝大桥的模拟检测,取得了较好的效果,并具有一定理论价值和潜在应用价值。This paper addresses the issue of detecting the abnormal behaviors of a bridge by interpreting the deflection data.By combining the near-neighborhood method and the K-mean method,a new bridge detection method is proposed,which is composed of two phases.In the first phase,the historical deflection data are classified by using the K-mean algorithm,while the second phase performs detection by employing a modified near-neighborhood method.The rate of correct detection is guaranteed by alternating applying these two pattern recognition techniques.The proposed method is applied to the simulation detection of the Chongqing Caiyuanba Grand Bridge.The experimental results justify the new method.As compared to the model-based methods,our detection method is very time-saving and,hence,adapts to the real-time detection sinario.

关 键 词：桥梁 健康检测 挠度 模式识别 

分 类 号：TP399[自动化与计算机技术—计算机应用技术]