CAJ | 학술논문

摊铺机是复杂的机电液系统，因受环境因素影响，容易出现故障，影响施工质量，需建立摊铺机的故障诊断系统。在建立摊铺机的故障征兆网络后，对采样信号进行3层小波包分解，用小波系数模极大值法提取故障的特征状态信息，用Pajek软件分析9个子系统故障的度和聚类系数，得到故障的可能部位，降低系统维数，减少后续计算量。用PSO优化神经网络算法进行状态样本数据学习，训练网络结构并优化，进而得到故障部位信息。样本测试表明，和神经网络相比，系统的迭代次数有效减少，实际输出结果和期望区间相符。
탄포궤시복잡적궤전액계통，인수배경인소영향，용역출현고장，영향시공질량，수건립탄포궤적고장진단계통。재건립탄포궤적고장정조망락후，대채양신호진행3층소파포분해，용소파계수모겁대치법제취고장적특정상태신식，용Pajek연건분석9개자계통고장적도화취류계수，득도고장적가능부위，강저계통유수，감소후속계산량。용PSO우화신경망락산법진행상태양본수거학습，훈련망락결구병우화，진이득도고장부위신식。양본측시표명，화신경망락상비，계통적질대차수유효감소，실제수출결과화기망구간상부。
The paver is a complex machinery-electro-hydraulic system .Due to environmental factors , the paver is prone to failure , affecting the quality of construction and resulting in property damage .Therefore, a diagnosis system for paver needs to be established .Firstly, the sampling signal is decomposed by three-layer wavelet packet , and fault feature status information is extracted by the coefficient modulus maxima method af -ter the establishment of fault symptoms network for the paver .Then failure degree and clustering coefficient of nine sub-system is analyzed by Pajek software to obtain the possible site of the fault , and reduce the system di-mension , and the amount of subsequent calculations .Finally, the state sample data is trained by PSO optimiza-tion neural network algorithm , then the structure of the network and fault information are achieved .The test da-ta show that comparing with the neural networks , the system is effective in reducing the number of iterations , and the actual output corresponds with the desired line interval .