Research On Crack Quantification Of Ferromagnetic Materials Based On AMR Sensor

Nondestructive testing is an important part of quality inspection of industrial products,and determines the quality of a product.In industrial production,nondestructive testing methods are various,including magnetic particle testing,magnetic flux leakage testing,ultrasonic testing and so on.The principle of magnetic flux leakage testing is the most classical and widely used nondestructive testing technology scheme.It is widely used in metal profile testing represented by iron and steel.The scheme is simple and the cost is low.Low cost,high speed,high accuracy,and has excellent environmental adaptability,can be applied to various environments defect detection.In this paper,a series of researches are carried out around magnetic flux leakage testing.On the basis of studying the principle of magnetic flux leakage testing,a set of magnetic flux leakage testing device is designed to meet the actual needs.It can realize the quantitative detection of metal crack size and the accurate prediction of crack parameters.Firstly,the development and research status of magnetic flux leakage testing technology at home and abroad are studied,and the intuitive cognition of this technology is formed.The development trend of nondestructive testing is summarized and predicted,which lays a foundation for further design work.Secondly,the basic knowledge and detection principle are studied.Based on the magnetic dipole simulation method,the change law between rectangular crack and magnetic flux leakage is studied,and the key factors affecting the magnetic flux leakage are defined.According to the requirement of metal defect detection in actual production,a magnetic flux leakage detection system based on AMR sensor is designed and developed,and the system integration and testing are completed.Then,on the basis of the previous design,combined with the basic principle of neural network,a defect identification system based on BP neural network is proposed.This system is mainly used to analyze and process the detected magnetic flux leakage signal.BP neural network training was carried out based on 25 samples processed in real time.The BP neural network model after training is tested with selected test samples.The results show that the prediction error range of the model meets the requirements and verifies the accuracy of the method.Finally,a comprehensive summary of the work of this paper,and look forward to the follow-up work,put forward the next phase of the focus of work.