The Studies On Damage Detection Methods Of Offshore Platform Structures Based On Wavelet Analysis

As the offshore platform structures service a long time in the harsh marineenvironment and suffer the interaction of a variety of loads, the structure is prone tovarious forms of damage, so that the carrying capacity of the structure will fell andsevere damage will lead to huge losses of lives and properties. Therefore, the damagedetection of offshore platform structure is very important in practice. The damagedetection method based on the vibration response of offshore platforms is a globaldamage detection method whose application is simple and low cost with a very broadapplication prospects, compared with the traditional damage detection method.The paper summarizes damage detection methods which are based on thestructural vibration response at home and abroad systematically, and discusses itspresent situation, research results and the problems comprehensively. Next the basictheory, developing process and application advantages in frequency analysis of thewavelet analysis is expounded. Combined with the preceding research results relatedto wavelet analysis, it has a further research on the application of wavelet analysis inthe field of damage detection.Based on wavelet analysis’s good characteristics of local analysis, damagedetection was investigated on an offshore platform which is modeled via finiteelement method, and four cases are considered, including damage detection fordifferent types of damaged members, damaged members located in different positions,different levels of noise pollution and vibration response signals in different directions.Useful conclusions could be obtained.In this paper, a method for damage detection of offshore platform based onwavelet detail signal node energy is proposed. The second layer detail signal nodes energy is choosed as the input of a probabilistic neural network because of differentreasons for example the second layer detail signal is sensitive to the damage of theoffshore platform during signal decomposition process by db wavelet. Detail signalnodes energys are got from acceleration signals of different position of the offshoreplatform via wavelet transform and integral in the time history. The input and outputof the probabilistic neural network input are wavelet detail signal node energy and thenearest leg numbers. In order to verify the validity of the proposed method, anoffshore platform was modeled via finite element method. And the numerical andphysical experiments are carried out. Their results show that the method can beapplied to damage detection of offshore platform structures. Next damage detection indifferent levels of the noise, diffferent size and direction of the driving forces arestudied, finding that the method has a certain degree of noise immunity; the size of thedriving force do not affect the results of damage identification and the direction of thedriving force have a large impact. So it requires recording the direction of the forcesduring the test of actual offshore platforms without the specific size of the drivingforces.