優(yōu)化K-HHT方法及其在GPS數(shù)據(jù)處理中的應(yīng)用
發(fā)布時間：2018-11-03 21:14
【摘要】：橋梁結(jié)構(gòu)健康監(jiān)測對于橋梁結(jié)構(gòu)的正常使用以及人民生命財產(chǎn)的安全具有重大意義。GPS監(jiān)測是橋梁健康監(jiān)測的重要手段。隨著GPS技術(shù)的發(fā)展,目前已經(jīng)能夠?qū)崿F(xiàn)實時動態(tài)監(jiān)測功能。因此,GPS監(jiān)測信號隱含了更加豐富的結(jié)構(gòu)健康信息有待挖掘。本文以鶴洞大橋健康監(jiān)測系統(tǒng)為依托,以GPS監(jiān)測信號為對象,以識別橋梁結(jié)構(gòu)的自振頻率為目的,展開以下研究工作:(1)信號分解方法。改進(jìn)HHT方法(課題組前期研究成果)采用具有預(yù)測性的Kriging擬合代替三次樣條擬合技術(shù)進(jìn)行HHT分解,能有效的改善HHT方法存在的端點效應(yīng)和模態(tài)混疊現(xiàn)象。但HHT分解效果很大程度上取決于Kriging擬合過程中相關(guān)模型參數(shù)?的初始取值。為此,本文采用尋優(yōu)能力較強(qiáng)的粒子群算法(PSO)對參數(shù)?的取值進(jìn)行尋優(yōu),通過尋優(yōu)過程消除參數(shù)?初始取值對改進(jìn)HHT分析效果的影響。對正弦、時變Chirp疊加信號進(jìn)行分析,結(jié)果表明,增加參數(shù)?優(yōu)化過程的改進(jìn)HHT方法(以下簡稱優(yōu)化K-HHT,其EMD過程稱為優(yōu)化K-EMD過程)分解出更趨于實際情況的IMF分量(固有模態(tài)函數(shù)),并且能夠有效的控制HHT方法的端點效應(yīng)問題。(2)信號趨勢項的分離。要從GPS信號中識別出結(jié)構(gòu)的自振頻率,必須分離GPS信號中的多路徑效應(yīng)和荷載作用下的結(jié)構(gòu)位移,即GPS信號的趨勢項。采用最小二乘法、小波變換、優(yōu)化K-HHT分離數(shù)字仿真信號中的趨勢項。以剔除趨勢前后信號的方差、均值、相關(guān)系數(shù)以及分離的趨勢與真值的均方根誤差為評價指標(biāo),對比三種方法對趨勢項的分離效果。結(jié)果表明優(yōu)化K-HHT效果最佳。(3)信號降噪處理。以均方根誤差、歸一化絕對誤差、信噪比以及系統(tǒng)平均偏差作為評價指標(biāo),對比優(yōu)化K-HHT方法、小波變換、優(yōu)化K-HHT-Wavelet三種方法的降噪效果。仿真算例表明,優(yōu)化K-HHT-Wavelet方法對非平穩(wěn)信號降噪的效果要優(yōu)于其他兩種方法,并且降噪后的信號曲線更加平滑。(4)基于上述成果分析鶴洞大橋GPS監(jiān)測信號。首先用優(yōu)化K-HHT分離趨勢項,然后利用優(yōu)化K-HHT-Wavelet方法進(jìn)行降噪,獲得鶴洞大橋振動位移時程,從而識別出橋梁的自振頻率。該頻率與理論計算值及加速度時程分析結(jié)果非常接近。
[Abstract]:The health monitoring of bridge structure is of great significance for the normal use of bridge structure and the safety of people's life and property. GPS monitoring is an important means of bridge health monitoring. With the development of GPS technology, real-time dynamic monitoring has been realized. Therefore, GPS monitoring signals imply more abundant structural health information to be mined. Based on the health monitoring system of Hedong Bridge, this paper takes the GPS monitoring signal as the object, and aims at identifying the natural vibration frequency of the bridge structure. The following research work is carried out: (1) signal decomposition method. Using predictive Kriging fitting instead of cubic spline fitting to decompose HHT, the improved HHT method can effectively improve the endpoint effect and modal aliasing in HHT method. But the effect of HHT decomposition largely depends on the model parameters in the process of Kriging fitting. The initial value of. In this paper, a particle swarm optimization algorithm, (PSO), is used to match the parameters. The parameters are eliminated by the optimization process. The effect of initial value on the effect of improved HHT analysis. The sinusoidal and time-varying Chirp superposition signals are analyzed. The results show that the parameters are increased. The improved HHT method for optimization process (hereinafter referred to as optimized K-HHT, whose EMD process is called optimized K-EMD process) decomposes the more practical IMF component (intrinsic mode function). And it can effectively control the endpoint effect of HHT method. (2) the separation of signal trend term. In order to identify the natural frequency of the structure from the GPS signal, it is necessary to separate the multipath effect in the GPS signal and the displacement of the structure under load, that is, the trend term of the GPS signal. Using the least square method and wavelet transform, the trend term of digital simulation signal separated by K-HHT is optimized. Taking the variance, mean value, correlation coefficient of the signal before and after the elimination of the trend, and the root mean square error of the separating trend and the true value as the evaluation indexes, the separation effect of the three methods on the trend term is compared. The results show that the optimization of K-HHT is the best. (3) signal noise reduction. The root-mean-square error, normalized absolute error, signal-to-noise ratio (SNR) and average deviation of the system are taken as evaluation indexes, and the noise reduction effects of the three methods, such as optimized K-HHT method, wavelet transform and K-HHT-Wavelet method, are compared and optimized. The simulation results show that the optimal K-HHT-Wavelet method is better than the other two methods in reducing the noise of non-stationary signals, and the signal curve is smoother after the noise reduction. (4) based on the above results, the GPS monitoring signals of Hedong Bridge are analyzed. The vibration displacement time history of Hedong Bridge is obtained by optimizing the K-HHT separation trend term, and then using the optimized K-HHT-Wavelet method to reduce the noise, so as to identify the natural vibration frequency of the bridge. The frequency is very close to the theoretical calculation value and the acceleration time history analysis result.
【學(xué)位授予單位】：廣州大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2015
【分類號】：U446
本文編號：2309057
[Abstract]:The health monitoring of bridge structure is of great significance for the normal use of bridge structure and the safety of people's life and property. GPS monitoring is an important means of bridge health monitoring. With the development of GPS technology, real-time dynamic monitoring has been realized. Therefore, GPS monitoring signals imply more abundant structural health information to be mined. Based on the health monitoring system of Hedong Bridge, this paper takes the GPS monitoring signal as the object, and aims at identifying the natural vibration frequency of the bridge structure. The following research work is carried out: (1) signal decomposition method. Using predictive Kriging fitting instead of cubic spline fitting to decompose HHT, the improved HHT method can effectively improve the endpoint effect and modal aliasing in HHT method. But the effect of HHT decomposition largely depends on the model parameters in the process of Kriging fitting. The initial value of. In this paper, a particle swarm optimization algorithm, (PSO), is used to match the parameters. The parameters are eliminated by the optimization process. The effect of initial value on the effect of improved HHT analysis. The sinusoidal and time-varying Chirp superposition signals are analyzed. The results show that the parameters are increased. The improved HHT method for optimization process (hereinafter referred to as optimized K-HHT, whose EMD process is called optimized K-EMD process) decomposes the more practical IMF component (intrinsic mode function). And it can effectively control the endpoint effect of HHT method. (2) the separation of signal trend term. In order to identify the natural frequency of the structure from the GPS signal, it is necessary to separate the multipath effect in the GPS signal and the displacement of the structure under load, that is, the trend term of the GPS signal. Using the least square method and wavelet transform, the trend term of digital simulation signal separated by K-HHT is optimized. Taking the variance, mean value, correlation coefficient of the signal before and after the elimination of the trend, and the root mean square error of the separating trend and the true value as the evaluation indexes, the separation effect of the three methods on the trend term is compared. The results show that the optimization of K-HHT is the best. (3) signal noise reduction. The root-mean-square error, normalized absolute error, signal-to-noise ratio (SNR) and average deviation of the system are taken as evaluation indexes, and the noise reduction effects of the three methods, such as optimized K-HHT method, wavelet transform and K-HHT-Wavelet method, are compared and optimized. The simulation results show that the optimal K-HHT-Wavelet method is better than the other two methods in reducing the noise of non-stationary signals, and the signal curve is smoother after the noise reduction. (4) based on the above results, the GPS monitoring signals of Hedong Bridge are analyzed. The vibration displacement time history of Hedong Bridge is obtained by optimizing the K-HHT separation trend term, and then using the optimized K-HHT-Wavelet method to reduce the noise, so as to identify the natural vibration frequency of the bridge. The frequency is very close to the theoretical calculation value and the acceleration time history analysis result.
【學(xué)位授予單位】：廣州大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2015
【分類號】：U446
【參考文獻(xiàn)】
相關(guān)期刊論文 前2條
1 韓建平;李林;王洪濤;錢炯;;基于Hilbert-Huang變換和隨機(jī)減量技術(shù)的模態(tài)參數(shù)識別[J];世界地震工程;2011年01期
2 王超;沈斐敏;;小波變換在探地雷達(dá)弱信號去噪中的研究[J];物探與化探;2015年02期
相關(guān)碩士學(xué)位論文 前1條
1 趙寶新;車載運輸系統(tǒng)激勵載荷分析處理技術(shù)研究[D];中北大學(xué);2009年
，本文編號：2309057
本文鏈接：http://sikaile.net/kejilunwen/daoluqiaoliang/2309057.html
教材專著
