Research On ACFM Intelligent Recognition Method And System For Underwater Structure Defects

With the implementing of maritime power strategy,the exploration and exploitation of offshore oil and gas have been putforward from shallow water to deepwater,which promotes the rapid development of various advanced marine technologies,engineering equipments and industries.The offshore structure serves in the seawater environment for a long time.It can be easily introduced corrosion and crack defects in the structure due to the corrosion medium,marine fouling organism,complex stress,external force damage and storm.Under the condition of underwater special working and severe environment,the classical ACFM theory is insufficient to the irregular crack and corrosion defects as the diverse defects and complex morphologies.The lift-off variations effect caused by the attachment brings various interfering signals.As a result,it is still a challenge to determine,recognize and evaluate defects in offshore structures using the nondestructive testing(NDT)method.This dissertation is funded by the Special National Key Research and Development Plan(Underwater Structure Defects Alternating Current Field Measurement Intelligent Recognition and Visualization Testing System)and the National Natural Science Foundation of China(Underwater Structure Defects High-precision Quantitative Recognition Method and Applied Research Based on Alternating Current Field Measurement).The key problems of automatic determination,intelligent recognition,visual evulation method and system are studied based on the alternating current field measurement(ACFM)technique.The research covers underwater ACFM theory,numerical simulation analysis,development of highly sensitive probes,automatic determination of characteristic signal,intelligent recognition and visual evulation of defects,development of underwater ACFM system.The research provides technical support for the intelligent testing,safety assessment and maintenance decision-making of underwater structure defects.Main works are summarized as follows:(1)Establishment of seawater environment ACFM theory model and simulation analysis On the basis of the conventional injected current for the detection of crack in the ACFM theory,the marine electromagnetism theory and leakage magnetic field equivalent magnetic-dipole model are introduced to set up the seawater environment ACFM theoretical model and finite element method(FEM)simulation model.Under the condition of leakage magnetic field,the distribution laws of electromagnetic field around the defect are analyzed.The distorted magnetic field characteristic signals of different defects are determined.The positive and negative evolution laws of excitated magnetic field-defect morphology-disturbed current-distorted magnetic field are revealed in the marine environment.It is an further improvement and supplement of ACFM throry model,which lays the theoretical foundation for automatic determination,intelligent recognition and visual evulation methods of underwater structure defects using ACFM technique.(2)Development of highly sensitive underwater ACFM probesThe key parameters,such as excitation frequency,are optimized by the underwater ACFM 3D finite element method(FEM)model,which determines the detection sensitivity of the probe.The excitation coil,conditioning circuit,highly sensitive tunneling magnetoresistance(TMR)magnetic sensor and underwater seal construction are designed.The highly sensitive ACFM single probe,weld detection probe and planar array probe are developed for the detection of underwater structure defects.The ACFM testing platform is set up to test the probe.The characteristic signals of defects are extracted and the C scan imaging is achieved by the planar array probe.It provides the probe and testing platform for the automatic determination,intelligent recognition and visual evulation of defects.(3)Automatic determination method for defects under the condition of lift-off variationsThe distortion mechanism of the interference signals caused by lift-off variations is investigated based on simulation and experimental results.The characteristic signal Bz is determined as the insensitive signal to the lift-off.The threshold determination and intelligent identification combined method is presented to determine the defect automatically.Firstly,the integration enhancement algorithm of characteristic signal Bz is proposed to improve the amplitude of the defect reponse signal which is determined by the threshold.Secondly,the convolutional neural networks(CNN)deep learning algorithm is presented to recognize the lift-off and defect butterfly plots respectively.In the end,the differential adaptive filtering determination algorithm based on characteristic signal Bz is proposed to determine the small defect.Thus all defects can be determined automatically in real-time under the condition of lift-off variations,which break through the existing bottleneck of defects determination under the condition of lift-off variations.(4)Intelligent recognition and visual evaluation methods for defectsOn the basis of the positive and negative evolution laws between characteristic signals and different defects,the Bz is set as the characteristic signal for imaging inversion of the defect surface profile.The defect surface profile imaging inversion algorithm based on the Bz image gradient field is presented to reconstruct the surface profile image of the crack,irregular crack and corrosion defects.The defect surface profile image database is developed by simulations and experiments.The CNN deep learning algorithm is proposed to achieve intelligent classification recognition of different kinds of defects.Based on the above,the two-step interpolation and segmentation interpolation algorithms are presented to achieve visual evaluation of the length,depth and 2D profile for the crack.The bidirectional gradient fusion algorithm is proposed to achieve visual evaluation of the arbitrary directional irregular crack surface profile.As the characteristic signal Bx is sensitive to the lift-off,the 3D morphology reconstruction algorithm based on the image segmentation technique is presented for the 3D morphology and arbitrary section visual evaluation of the corrosion defect.This work provides intelligent classification recognition and visual evaluation methods for the underwater defects and promotes the development of condition based management(CBM).(5)Development and testing of underwater ACFM intelligent testing systemBased on the intelligent recognition and visual evaluation algorithms,the intelligent recognition software is developed including online transmission and offline intelligent modes.The excitation module,signal processing module,acquisition module,signal transmission module,power control module,et al.are assembled in the underwater pressure hull.As a result,the 500 m underwater ACFM intelligent testing system is set up,including probe,underwater pressure hull and intelligent recognition software.The underwater ACFM intelligent testing system is tested with different defects.The results show that the underwater ACFM intelligent testing system can achieve automatic determination,intelligent recognition and visual evaluation of underwater structure defects,which provides system support for intelligent testing,safety assessment and maintenance decision-making for underwater structure defects and improves the intrinsic safety level of underwater structures.