Cable Damage Identification Based On Wavelet Packet Analysis And Neural Network

Cables are widely used in existing bridge structures.Due to the external environment,traffic load and fatigue,the cables will inevitably be damaged,resulting in the decrease of the overall stiffness of the bridge and the decrease of carrying capacity.The timely and accurate identification of cable damage is an important and urgent task to ensure the safe operation of the bridge.The application field of cable is very extensive.This paper mainly focuses on the research of the cables of two types of bridges,external prestressed concrete bridges and cable-stayed bridges.Among them,it is difficult to solve the problem of in vitro cable injury using frequency domain analysis.The cable damage identification belongs to the analysis of highly nonlinear systems.Most of the current researches use the mathematical model or mechanical parameter identification,so it is difficult to obtain a good recognition effect.Therefore,this study aims at the above problems,combining experimental research and theoretical analysis,establishing a method of cable damage identification based on wavelet packet analysis and neural network,which provides a theoretical basis for cable safety assessment and has theoretical and practical significance.The main work are as follows:(1)According to the damage characteristics of the external prestressed cable,the cable vibration tests with different damage were designed by cutting the cable to simulate the cable damage.By directly cutting the cable to simulate the damage of the cable,the cable vibration signals are obtained at different damage levels,different cable forces and different cable lengths.(2)Study on the result of identifing the cable damage with constant cable force,using the energy ratio deviation ERVD and energy ratio of variance ERVV based on wavelet packet analysis.The cable damage identification index based on wavelet packet energy spectrum,RES,is established.Using index RES to build BP neural network model.The health status of external cables used for bridge reinforcement was evaluated.(3)Aiming at the damage characteristics of stay cables,a cable-stayed bridge with a main span of 340 meters was used as the engineering background.The method of reducing elastic modulus was used to simulate the cable damage.The damage identification method of cable based on BP neural network was studied.The main conclusions are as follows:(1)The characteristics of the cable acceleration signal in the time-frequency domain are obtained.In the time domain,the vibration decay time decreases;in the frequency domain,the spectrum appears rough and multiple peak points appear;and the total energy of the wavelet packet decreases in the energy domain.(2)The structural damage pre-warning index based on wavelet packet analysis shows that the energy-ratio deviation ERVD and the energy-ratio variance ERVV are less effective in identifying the cable tension-invariant cable damage.Wavelet packet energy spectrum based cable damage identification index Wavelet packet total energy rate of change indicator RES can well identify the damage.Combined with the BP neural network model established by the RES parameter,it can theoretically predict the damage index value of any cable force,cable length and damage degree,and can be used to identify the cable damage of the cable structure such as prestressed reinforcement.(3)The BP neural network model was established based on the change rate of cable force,which can effectively identify the damage position and damage degree of the cable.The maximum error is less than 6%.The method of early identification of cable damage of cable-stayed bridge is established.