Research On Damage Identification Technology Of Pipeline Support Structure Based On GA-BP Neural Network And Data Fusion

High-temperature and high-pressure steam pipelines,widely used in industrial production and daily life,are common structures in heating systems.In the service stage,the pipelines are generally used in a harsh environment,where they are often damaged by pressure,external force,corrosion,etc.If health monitoring of the pipeline system is not carried out,once the pipeline bursts,it will inevitably lead to a serious accident.Therefore,it is of great necessity and practical engineering significance to carry on the research on structural damage identification of pipeline system.Based on BP neural network,combined with genetic algorithm and D-S evidence theory and taking a high temperature and high pressure steam pipeline system as an engineering background,the structural damage identification technology of the pipeline system was studied.The main contents of the research are as follows:(1)The finite element model of the piping system structure was established with ABAQUS and the modal analysis was carried out.The measuring points of acceleration sensors on the pipeline system were selected and optimized by motion modal energy method.By comparing the experimental results with the numerical simulation results,the consistency of the experimental model and the finite element model in modal analysis was proved.(2)Based on the basic theory of BP neural network,the structure of BP neural network,learning algorithm,the determination of the number of hidden layers and the main parameters of the algorithm were studied.The weight updating formula of BP neural network was deduced,and the BP neural network model was established.The main components of genetic algorithm and the operation parameters of the algorithm were studied.The algorithm steps of BP neural network optimized by genetic algorithm were summarized,and the GA-BP neural network model for structural damage identification of pipeline system was constructed.(3)Based on GA-BP neural network model,three modal parameters ─ natural frequency,mode shape and curvature mode ─ were separately used to identify the damage degree and damage location of the pipeline system structure.It can be seen from the identification results that the curvature mode ratio and modal shape were sensitive to the damage degree of the structure,and the natural frequency and mode shape had good identification effect on structural damage location.By comparing the network performance of BP neural network and GA-BP neural network,it is clear that genetic algorithm can not only improve the accuracy of structural damage identification,but also improve the operating efficiency of the algorithm.(4)Based on D-S evidence theory,the results of the BP neural networks in structural damage degree identification and structural damage location were subjected to decision-level data fusion.By comparing the results before and after the fusion,it can be seen that compared with the identification result of a single modal parameter,the damage misjudgment of each modal parameter can be corrected by data fusion,and the robustness of structural damage identification system was improved.