Magnetic Memory Quantitative Recognition Model For Defect Grade Of Submarine Pipeline Based On Dynamic Immune Fuzzy Clustering

Influenced by the external environment,self-structure,and conveying media,submarine pipeline will produce stress concentration,which will lead to submarine pipeline cracking and leakage accidents.This will not only cause a large number of resource losses but also pose a threat to the marine environment.In this paper,the degree of defect damage can not be accurately judged based on a single metal magnetic memory(MMM)characteristic parameter due to external environment impact and interference in the process of submarine pipeline detection.Due to the influence and interference of the external environment in the process of submarine pipeline detection,the acquired magnetic memory data has the characteristics of dispersion and fuzzy uncertainty.Only based on a single MMM characteristic parameter can not accurately judge the degree of defect damage.Therefore,it is necessary to comprehensively consider the multi-dimensional MMM characteristic parameters,and further take effective classification optimization algorithm to identify the MMM characteristic parameters classification.Therefore,a fuzzy dynamic clustering model based on immune optimization is presented to quantitatively identify MMM signals with different degrees of damage.In order to study the change rule of MMM signal under different internal pressure,flow velocity,depth,and diameter of defects in the submarine pipeline.In this paper,by using the submarine pipeline specimens prefabricated with different defects,the MMM testing experiments are carried out in the laboratory environment.The MMM characterization laws are obtained for the submarine pipelines.The experimental results show that different depth and diameter of defects have more influence on the MMM signals than internal pressure and flow velocities.Under the same internal pressure and flow velocity,with the increase of defect depth and diameter,the tangential magnetic field component changes larger than the normal magnetic field component,and the tangential gradient changes smaller than the normal gradient.Moreover,the MMM signals under different damage degrees are nonlinear,fuzzy and uncertain.Therefore,it is necessary to introduce multi-dimensional MMM parameters and an effective classification algorithm to identify the critical features of defects.The change rate of magnetic field intensity?Hpx/?x,magnetic gradient K_wand gradient limit state coefficient M are extracted as critical characteristic parameters of the MMM signal.In this paper,the Modified B31G standard is used to evaluate the defect pipeline.By analyzing the MMM signals of submarine pipeline specimens with different depths and diameters,it is found that the different depths of defects have an obvious influence on the MMM signals.By extracting the MMM characteristic parameters as training samples,the fuzzy dynamic clustering algorithm is carried out and the initial fuzzy clustering is acquired based on the output threshold?.To further consider the problem that the fuzzy dynamic clustering algorithm is easy to fall into local solution,an immune algorithm with the ability of global searching is used to optimize the Euclidean distance between samples to approach the fuzzy similarity.According to the affinity between antibodies,the optimal threshold?is obtained,and the optimal classification is output.Finally,a fuzzy dynamic clustering model with immune optimization is established.The validation results show that the prediction classification accuracy is 87.5%,which provides a new idea for the quantitative evaluation of MMM of submarine pipeline defect grade.