Research On Nondestructive Detection Of Electromagnetic Fusion Based On BP Neural Network

Detection based on a single physical field has limited application scenarios and cannot balance efficiency and accuracy.The electromagnetic multi-field coupling detection method can obtain multi-physics signals under a certain stress or defect state of the test piece,which overcomes the limitations of low accuracy,large interference,and low efficiency of a single physical field.In this paper,a non-destructive testing method for electromagnetic fusion with Barkhausen signals and eddy current signals is studied.It turns out that the electromagnetic fusion detection method can effectively improve the detection accuracy.This paper first introduces the research background and the advantages and disadvantages of traditional non-destructive testing,and puts forward the basic idea of electromagnetic fusion detection.It introduces the research status of eddy current testing and Barkhausen inspection in detail.Then the eddy current detection,Barkhausen detection detection principle is introduced,and the two detection signal analysis methods and feature extraction methods.are summed up The basic principle of electromagnetic fusion and its implementation method are elaborated in detail.The eddy current signal and Barkhausen signal fusion detection method based on BP neural network was studied.The neural network electromagnetic fusion detection model was established.Then,the software system and hardware system of electromagnetic fusion detection system are designed.Including signal generator,power amplifier circuit,signal conditioning circuit design and data acquisition device selection;software system related functional modules.Finally,the developed electromagnetic fusion detection system was used to detect and analyze the stress and typical defects of equal strength beams,including a single Barkhausen signal analysis and a single eddy current signal analysis,Barkhausen and eddy current fusion analysis.Comparing the results of the three analysis methods,the study found that in addition to the quantitative analysis of defect width,the accuracy of the training network for all features of electromagnetic fusion is higher than the single feature value of the detection signal.Electromagnetic fusion detection of stress and depth of defects have achieved good results,indicating that the electromagnetic fusion method can effectively improve the accuracy of electromagnetic non-destructive testing.