基于核主元分析和贝叶斯推理的建设工程项目质量状态在线诊断模型及其仿真  

On-line Diagnosis Model of Construction Project Quality Status and its Simulation based on KPCA and Bayesian Inference

作　　者：侯学良[1] 汪勇[1] 

机构地区：[1]华北电力大学经济与管理学院,北京102206

出　　处：《系统管理学报》2015年第4期472-479,共8页Journal of Systems & Management

基　　金：国家自然科学基金资助项目(71171081);教育部新世纪优秀人才项目(NCET-11-0633)

摘　　要：针对建设工程项目质量控制过程的随机变化性,运用核主元分析和贝叶斯推理的方法,建立了工程质量状态在线诊断的数学模型;借鉴现代工业控制技术与手段,通过实时连续采集质量信息、提取信息特征、识别特征异常以及估计异常模式,实现了对质量异常状态的在线监测与模式识别。利用Matlab对某工程实例进行了常态下的仿真试验。分析结果表明,该模型不仅可以快速响应突发性质量异常,而且能够提前预警累积性质量异常,还能以概率形式诊断出异常状态所属的模式类型。For the randomness in construction project quality control process,we establish an on-line diagnosis model for construction quality using kernel principal component analysis（KPCA）and Bayesian inference.Then,using modern industrial control technology and method,on-line monitoring and pattern recognition of abnormal quality are implemented by continually acquiring quality information,extracting information features,identifying abnormal features and estimating abnormal patterns. Simulation experiments with normal data and abnormal data are used to test the on-line diagnosis model using Matlab.The simulation results show that this model is capable of not only quickly responding to sudden abnormal quality,but also to early warning cumulatively abnormal quality,and effectively diagnosing the mode type of abnormal quality with a probability.

关 键 词：工程质量 在线诊断 核主元分析 贝叶斯推理 

分 类 号：F203[经济管理—国民经济]