Application Of Data Mining Technology In The Data Processing Of Vibration Signal Of Bridge

The21st century, China’s bridge construction rapid development,Bridge span and number have jumped to the forefront of the world. Bridge as one of the lifeline engineering,, Of great significance for the national economy. Because the bridge is usually located in the important traffic trunk, Long-term under the continuous impact of the automobile, and the bridge structure long-term in the vibration state, The safety of the bridge structure is particularly important. Therefore, You need to build a bridge health monitoring system for real-time monitoring of the vibrational state of the bridge structure, And timely warning of the possible destruction. On the other hand, With the structure of the bridge is constantly updated and its largest span also increases, The regularity of the structure of the bridge is more and more complex. Particularly in the effect of earthquakes, typhoons, floods and other natural disasters, The regularity of the structure of the bridge is more complicated. Therefore, it is necessary to carry out the bridge model test to study the law of vibration response of bridges under the possible dynamic loads. Whether bridge real-time health monitoring system or bridge dynamic model test, the processing of the bridge structure’response data is particularly important. Load impact of the car, earthquakes, typhoons and floods, the structure of the bridge is not a static state of deformation, but changing the vibrational state, we need real-time bridge vibration signal acquisition.Because of bridge vibration occurred every moment, using only artificial means cannot process the enormous amount of vibration signal collected continuously timely and effective. In order to timely, effective and accurate to dig out the bridge structure status information contained in the measured vibration signal, we must reference data mining techniques for vibration signal analysis.This paper attempts to apply data mining technology to the massive bridge vibration signal analysis, find out the data analysis model of actual bridge vibration signal, acquire variation regularities of the parameters in the bridge structure, and dig potential correlations between the various of parameters of the bridge。 Finally, through a series of time-series analysis of the model to predict the possible vibrational state of the bridge structure.This paper for bridge vibration signal data mining technology research can also be applied to other large-scale structures, such as high-rise buildings, large-span stadium structure, with a wide range of practical engineering significance.Because of the vibration signals of the bridge need multiple signal conversion and transmission, and the internal force of bridge structure is complex,and the external environment conditions of bridge is poor. After the data conversion and transfer process,the resulting data inevitably mixed with many unnecessary ingredients. Therefore, this paper proposed should first get signal data preliminary treatment with fixed waveform distortion,eliminate mixed noise and interference in the signal.and weaken the excess content of signal. The paper is based on bridge vibration signals possible problems, which combined with theoretical deduction and data analysis methods are introduced in detail the corresponding signal pretreatment method(Calibration data of the physical unit、Eliminate trend item、Smooth processing vibration signal and vibration signal filter processing and so on).The paper takes the measured health monitoring data of a suspension bridge and mass vibration data obtained from the dynamic model test of a cable-stayed bridge as example, on the basis of vibration signal preprocessing, uses data mining techniques to create a variety of data mining models (clustering model, correlation model and time series models) bridge vibration signal analysis. Clustering model, to monitor anomalies measured bridge vibration signal monitoring. Association model, to sum up the bridge each part of the reaction and the external loads (including temperature) parameters associated law. And Time series model, to observe the trend of bridge vibration signal, to predict the parameter values, in order to determine the health status of the current bridge structure.