Signal Detection Based On Eddy Current Effect And Its Application In Metal Detection

Metal detection(MD)technology has been widely utilized since the development of industrialization.Nowadays,the applications of this technology not only benefit industry,but also ensure people's safety.In some important occasions,articles that endanger public safety are prohibited to be carried,especially metal products.It is necessary to inspect and control the related personnel's belongings in advance,and the most common detection tool is the metal detection device.A large amount of existing metal detection devices not only have low detection sensitivity and cost a lot of labor for assistance,but also have poor system stability and high false alarm rate.Therefore,in response to such a bulky metal detection system that exists in name only,one modern metal detection technology is requisite to provide support to reduce the incidence of safety hazards.Therefore,based on this application scenario,further study is made in metal detection technology.First of all,based on the eddy current effect,this thesis clarifies the theories of eddy current testing(ECT),understands the distribution characteristics of eddy current,analyzeds the five main factors affecting eddy current testing,and establishes the geometric circuit model of eddy current testing.On this basis the metal detection process is summarized.According to the requirements,the metal detection system is designed and the structure of eddy current sensor is determined.By the use of digital signal processing(DSP)method,the generation and extraction process of signals from transmitter to receiver are achieved.The functions and algorithms of each module in the system are described in accordance with the detection process.Followed by this metal detection technology,the test environment is set up and the test procedures in sensitivity of the system detection are under a national standard.The least square(LS)method is utilized to estimate the slope of inductive signal,which provides a possibility for the classification of metal materials.Then,on the basis of this system,the system is optimized by taking precautions after analyzing the existing defects in the system design combined with practical applications.These optimization measures include making a use of comb notch filter to eliminate the power frequency interference,taking advantage of an all pass filter in a way to shift phase in order to reduce phase disturbance,and adopting differencial amplitude detection method to suppress the temperature drift.The improved system has a good performance in detection capability and stability.Finally,the inductive signals are preprocessed mainly based on endpoint detection to generate sample sets,and the artificial neural network(ANN)model is built to discriminate categories of metal objects.The neural network is debugged and analyzed by cross validation,and the network parameters are determined.The classification ability of the neural network is tested and compared with the amplitude-phase method.The results indicate that the classification technology of metal objects with neural network is effective.