基于改进LVQ算法的塔式起重机运行状态检验  

Operation status inspection of tower crane based on improved LVQ algorithm

作　　者：周庆辉 刘浩世 刘耀飞 李欣 谢贻东 ZHOU Qing-hui;LIU Hao-shi;LIU Yao-fei;LI Xin;XIE Yi-dong(School of Mechanical,Electrical and Vehicle Engineering,Beijing University of Civil Engineering and Architecture,Beijing 100044,China;Beijing Construction Safety Monitoring Engineering Technology Research Center,Beijing 100044,China;China Railway Construction Group Co.,Ltd.,Beijing 100040,China)

机构地区：[1]北京建筑大学机电与车辆工程学院,北京100044 [2]北京市建筑安全监测工程技术研究中心,北京100044 [3]中铁建设集团有限公司,北京100040

出　　处：《机电工程》2022年第11期1636-1642,共7页Journal of Mechanical & Electrical Engineering

基　　金：国家自然科学基金青年科学基金项目(51905028);住房与城乡建设部科技计划项目(2022 K 079)。

摘　　要：为了提高起重机运行安全检验结果的准确性,避免误判,并且提高塔式起重机检验的智能化水平,提出了一种基于改进的学习矢量量化(LVQ)人工神经网络模型,实现了对塔式起重机运行安全状态的智能检验。首先,根据近年来建筑工地塔式起重机的检验数据,建立了样本集,基于塔式起重机相关的安全技术标准和规范,将检验项目分解为最常见、最主要的15个因素,作为神经网络输入层的数目;然后,对290台塔式起重机的检验数据进行了统计(金属结构的连接、作业环境、主要零部件与机构,此3项不合格的频次较高);最后,在学习矢量量化(LVQ)算法基础上,改进了LVQ人工神经网络的检验评价模型,再运用优化的特征数据训练出了LVQ分类器,提出了改进的LVQ智能检验方法,对50个测试样本进行了分类识别实验。研究结果表明:改进后的LVQ人工神经网络算法提高了塔式起重机检验结果的正确率,在整机检验中合格率和不合格率均能达到100%,避免了误判,实现了对塔式起重机设备的安全智能检验。In order to improve the accuracy of safe operation of tower crane inspection results,avoid misjudgment,and improve the intelligence level of tower crane inspection,an improved learning vector quantization(LVQ)artificial neural network model was proposed to realize the intelligent inspection of safe operation of tower crane.Firstly,the randomly-selected test samples set was established for the whole equipment on the basis of the inspection samples of tower cranes on construction sites in recent years.Based on the safety technical standards and specifications of tower cranes,the inspection items were divided into the 15 common factors as the number of input layer of the neural network in the sample set.Then,the inspection data of 290 tower cranes were counted.For the connection of metal structure,working environment,main parts and mechanisms,the frequency of these three nonconformities was high.Finally,the conventional LVQ algorithm was improved on the evaluation model,and the LVQ classifier was trained by using the optimized characteristic data.Hence,based on the improved LVQ algorithm,an intelligent inspection was proposed.The classification and recognition experiments were carried out on 50 test samples.The research result shows that the improved LVQ algorithm can increase the accuracy of judgment,because both the qualified rate for qualified equipment and unqualified rate for the unqualified equipment can all reach 100%in the whole equipment inspection.Therefore,the improved LVQ algorithm can avoid misjudgment and realize the safe and intelligent inspection.

关 键 词：自行式起重机 运行安全状态 安全技术标准和规范 学习矢量量化 人工神经网络模型 LVQ分类器 

分 类 号：TH213.3[机械工程—机械制造及自动化] TU391[建筑科学—结构工程]