本文選題：模式識(shí)別 + 機(jī)器學(xué)習(xí)　； 參考：《哈爾濱工業(yè)大學(xué)》2015年碩士論文

【摘要】：隨著交通量的迅速增大,現(xiàn)有橋梁結(jié)構(gòu)通常長(zhǎng)期在超負(fù)荷流量情況下服役,同時(shí)還要遭受地震、風(fēng)暴等自然災(zāi)害的侵襲,導(dǎo)致結(jié)構(gòu)承載力下降,影響其運(yùn)營(yíng)安全。所以通過(guò)在結(jié)構(gòu)上布設(shè)的實(shí)時(shí)監(jiān)測(cè)系統(tǒng),監(jiān)測(cè)結(jié)構(gòu)的動(dòng)力特性,從而對(duì)結(jié)構(gòu)進(jìn)行損傷識(shí)別和狀態(tài)評(píng)估,對(duì)于橋梁結(jié)構(gòu)的運(yùn)營(yíng)安全具有重要意義。但大量的研究表明,這種方法應(yīng)用于大型土木工程結(jié)構(gòu)時(shí),由于測(cè)量的噪聲和運(yùn)營(yíng)狀況的變化等,會(huì)使得識(shí)別出的模態(tài)參數(shù)產(chǎn)生相應(yīng)的變化,當(dāng)變化范圍大于由于結(jié)構(gòu)自身?yè)p傷而造成的模態(tài)參數(shù)變化時(shí),很難對(duì)結(jié)構(gòu)進(jìn)行準(zhǔn)確的損傷識(shí)別。一般情況下,造成模態(tài)參數(shù)變化的因素包括:1、環(huán)境條件(溫度的變化、地基狀態(tài)和濕度);2、運(yùn)行條件(車輛荷載狀況和其他外界激勵(lì));3、測(cè)試以及后處理存在的誤差。本文首先介紹了在環(huán)境因素變化條件下結(jié)構(gòu)模態(tài)參數(shù)研究的現(xiàn)狀,然后基于環(huán)境激勵(lì)下的模態(tài)參數(shù)識(shí)別技術(shù):頻域分解法(Frequency Domain Decomposition,FDD)和自然激勵(lì)(Natural Excitation Technique,NEx T)與最小特征系統(tǒng)實(shí)現(xiàn)(Eigen-system Realization Algorithms,ERA)的聯(lián)合識(shí)別法,對(duì)橋面加速度響應(yīng)數(shù)據(jù)進(jìn)行了分析,并且根據(jù)對(duì)監(jiān)測(cè)到的應(yīng)變數(shù)據(jù)進(jìn)行分析,選取合適的特征向量和模式數(shù)目,從而對(duì)車輛荷載進(jìn)行模式識(shí)別,提取輕車模式狀態(tài)下的加速度監(jiān)測(cè)數(shù)據(jù)和溫濕度監(jiān)測(cè)數(shù)據(jù),識(shí)別出模態(tài)參數(shù)并計(jì)算出相應(yīng)的溫濕度特征值,構(gòu)建樣本集合;最后運(yùn)用支持向量機(jī),直接利用現(xiàn)場(chǎng)實(shí)測(cè)數(shù)據(jù)和識(shí)別出的模態(tài)參數(shù)值建立了數(shù)據(jù)驅(qū)動(dòng)的特征頻率預(yù)測(cè)模型,并利用該模型反過(guò)來(lái)對(duì)特征頻率影響因素進(jìn)行了研究分析。本文首次將模式識(shí)別和機(jī)器學(xué)習(xí)相結(jié)合的方法應(yīng)用到橋梁結(jié)構(gòu)在變環(huán)境條件下的模態(tài)參數(shù)研究分析中,直接利用實(shí)測(cè)數(shù)據(jù)建立了特征頻率的預(yù)測(cè)模型,并且在利用該模型進(jìn)行特征頻率影響因素的研究分析中發(fā)現(xiàn)了一些新知識(shí),為環(huán)境變化條件下結(jié)構(gòu)模態(tài)參數(shù)的研究提供一種有效的方法。
[Abstract]:With the rapid increase of traffic volume, the existing bridge structures usually serve under the condition of overload and discharge for a long time. At the same time, they are also affected by natural disasters, such as earthquakes and storms, which lead to the decrease of the bearing capacity of the structure and affect its operation safety.Therefore, it is of great significance for the safety of the bridge structure to monitor the dynamic characteristics of the structure by the real-time monitoring system, so as to identify the damage and evaluate the state of the structure.However, a large number of studies have shown that, when this method is applied to large civil engineering structures, the measured noise and the change of operating conditions will cause the corresponding changes of the identified modal parameters.When the variation range is larger than the change of modal parameters caused by the damage of the structure itself, it is difficult to identify the damage of the structure accurately.In general, the factors causing modal parameter change include: 1, environmental conditions (temperature change, ground state and humidity), operating conditions (vehicle load and other external excitations), errors in testing and post-processing.This paper first introduces the research status of structural modal parameters under the change of environmental factors.Then, the acceleration response data of bridge deck are analyzed based on the modal parameter identification techniques under ambient excitation: frequency Domain decomposition (FDD) and Natural Excitation technique (NEX T) and minimum feature system implementation (Eigen-system Realization algorithm ERAA).Based on the analysis of the strain data, the appropriate eigenvector and the number of patterns are selected to identify the vehicle load, and the acceleration monitoring data and the temperature and humidity monitoring data are extracted.The modal parameters are identified, the corresponding temperature and humidity eigenvalues are calculated, and the sample set is constructed. Finally, the data-driven characteristic frequency prediction model is established by using the field measured data and the identified modal parameters directly by using the support vector machine.The influence factors of characteristic frequency are analyzed by using this model.In this paper, the method of pattern recognition and machine learning is first applied to the modal parameter analysis of bridge structure under the condition of variable environment, and the prediction model of characteristic frequency is established by using the measured data directly.Some new knowledge is found in the research and analysis of the influence factors of characteristic frequency by using this model, which provides an effective method for the study of structural modal parameters under the condition of environmental change.
【學(xué)位授予單位】：哈爾濱工業(yè)大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2015
【分類號(hào)】：U446

【參考文獻(xiàn)】

相關(guān)期刊論文 前2條

1 張啟偉;斜拉橋正常運(yùn)營(yíng)條件下振動(dòng)特性的變異性[J];同濟(jì)大學(xué)學(xué)報(bào)(自然科學(xué)版);2002年06期

2 歐進(jìn)萍,周智,武湛君,趙雪峰,莫淑華;黑龍江呼蘭河大橋的光纖光柵智能監(jiān)測(cè)技術(shù)[J];土木工程學(xué)報(bào);2004年01期

，

本文編號(hào)：1771424