| 杨光,栾博文,李波,孙锦,朱赵辉,张建伟.基于原型试验和封装式策略的PCCP断丝信号有效特征挖掘[J].水利学报,2025,56(2):170-180 |
| 基于原型试验和封装式策略的PCCP断丝信号有效特征挖掘 |
| Effective parameters mining algorithm for broken wire signals of prestressed concrete cylinder pipes based on prototype experiment and wrapper strategy |
| 投稿时间：2024-03-08 |
| DOI：10.13243/j.cnki.slxb.20240128 |
| 中文关键词: 预应力钢筒混凝土管 断丝实时监测 原型试验 有效特征挖掘 封装式建模 |
| 英文关键词: prestressed concrete cylinder pipe real-time broken wire monitoring prototype experiment effective parameter mining wrapper modelling |
| 基金项目:国家自然科学基金项目(52109155，52279133，42401319) |
|
| 摘要点击次数: 628 |
| 全文下载次数: 407 |
| 中文摘要: |
| 为提升分布式光纤监测系统对预应力钢筒混凝土管(PCCP)断丝信号的实时探测能力，基于封装式策略构建分析模型，挖掘利于振动模式识别的有效特征参量。应用相位敏感光时域反射(Φ-OTDR)型分布式光纤监测系统采集振动信号，针对时域、频域以及时频域建立原始特征参量集合;引入非线性收敛因子、非线性自适应权值、高斯变异扰动和Tent混沌扰动，对鲸鱼优化算法(WOA)进行改进，在此基础上，借助改进的WOA (IWOA)和层次聚类(HC)，提出IWOA-HC有效特征挖掘模型。案例分析结果显示：Fisher、BPSO-HC、WOA-HC和IWOA-HC分别挖掘出61、35、32和27个有效特征参量，利用4组有效特征参量进行振动模式识别，准确率分别为71.11%、86.67%、89.44%和97.22%，IWOA-HC挖掘出的有效特征参量具备最佳的模式识别能力。本研究可为在役PCCP工程的健康诊断及灾害预警提供技术支持。 |
| 英文摘要: |
| To enhance the capability of distributed optical fiber monitoring system on real-time broken wire signal detection of prestressed concrete cylinder pipe (PCCP),this study proposes a wrapper theory-based model for mining the effective feature parameters which are conducive to identifying PCCP broken wire signals.Phase sensitive optical time domain reflectometry (Φ-OTDR) monitoring system is used to obtain vibration signals,and the original feature parameter set is established by considering time domain,frequency domain,and time-frequency domain.The improvement strategies for whale optimization algorithm (WOA) are established by employing the nonlinear adaptive weights,the nonlinear convergence factor,the Gaussian mutation perturbation,and the Tent chaotic perturbation.By integrating the improved WOA (IWOA) and the hierarchical clustering (HC),the IWOA-HC model for effective feature mining is constructed.The following results are drawn.Fisher,BPSO-HC,WOA-HC,and IWOA-HC mine 61,35,32,and 27 effective feature parameters,respectively.Using the above four feature parameter sets for vibration pattern recognition,the accuracy rates are 71.11%,86.67%,89.44%,and 97.22%,respectively.The effective feature parameters mined by IWOA-HC have the best pattern recognition performance.This research provides technical support for health diagnosis and disaster warning of in-service PCCP engineering. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|