Optimize The K-HHT Method And Its Application In The GPS Data Processing

Bridge structural health monitoring is of great significance for normal use and security of lives and property of the bridge structure.GPS monitoring is an important means to bridge health monitoring.With the development of GPS technology,GPS monitoring has been able to achieve real-time dynamic monitoring recently.Therefore, GPS monitoring signal implies a more extensive structural health information to be excavated. In this thesis, on the basis of health monitoring system of Hedong Bridge,taking GPS monitoring signal as project,and for the purpose of identifying the natural frequencies of the bridge structure,the main research works are as followed:(1) The method of signal decomposition.Improved HHT method(Our previous research)take the predictive Kriging fitting instead of cubic spline fitting technique to HHT decomposition,which can effectively improve the existence of end effects and modal aliasing in HHT method. However, the effect of HHT depends largely on the initial value of the correlation model parameters in the process of Kriging fitting. In this respect,this thesis introduces the particle swarm algorithm(PSO) to optimize the value of the parameter,eliminating the initial value of the impact parameter analysis of the effect of improved HHT through the optimization process.The sine signal,time-varying signal superimposed Chirp is analyzed,the results show that the improved HHT method which increased parameter optimized process(The following referred to as optimized K-HHT, its EMD process was known as optimized K-EMD process) decompose the IMF components which tend to be real situations, and it can effectively control the end effect problem of HHT Method.(2) Separating the signal trend. If identifying the natural frequencies of the structure from GPS signals, it must remove the multi-path effects of GPS signal and the structure displacement by loads, that is trend of GPS signals.Using the least square method, wavelet transform, the optimized K-HHT method separate the trend of digital simulation signal.Considering the variance, average, correlation coefficient of signal before/after de-trending as well as trends in the separated signal and the true value of the root mean square error as the evaluation index compare three methods for the detached effect of the trend. The results shows that the optimized K-HHT have the best effect.(3) The signal noise reduction. Considering the root mean-square error(RMSE),normalized absolute error,signal-to-noise ratio and average systematic bias of signal before/after de-noising as the evaluation index compare the noise reduction effect of the trend by the optimized HHT method, Wavelet transform, optimized K-HHT-Wavelet methods. The simulation examples shows that optimized K-HHT-Wavelet method is better than other two methods for non-stationary signal de-noising effect, and the de-noising signal curve is more smoother.(4) On the basis of above results analysis the GPS monitoring signal of Hedong Bridge.Firstly using the optimized K-HHT method separate the signal trend,and then use optimized K-HHT-Wavelet method for noise reduction, that obtains the vibration displacement time-history of hedong bridge, so as to identify the natural frequencies of the bridge. The frequency and time-history analysis results of the acceleration with the calculated value is very close.