Research On Classification Of Non-ferrous Metals Based On Electromagnetic Eddy Current Detection

The widespread promotion of the concept of circular economy has placed higher demands on the separation and recycling of metals.A key step in recovering metal sorting systems is the development of real-time non-contact sorting methods to identify class differences in metals on the conveyor belt.Non-ferrous metals cannot be sorted using magnetic deflection forces in the recycling process like steel,but different types of non-ferrous metals have different electrical conductivity properties.Eddy current detection,as a non-contact,economical and fast response non-destructive testing technology,provides a new development direction for the classification of non-magnetic metals.In this paper,we study the classification method of nonferrous metals based on electromagnetic eddy current detection technology,and the main work is as follows.In view of the current classification methods of nonferrous metals and the current status of electromagnetic eddy current detection technology and research,the development potential of electromagnetic eddy current detection in the classification technology of nonferrous metals is analyzed,and the influence of the lift-off effect of the phase spectrum of eddy current sensors is summarized and explained by combining the derivation process of the analytical solution of the double coil-metal plate model.A new method of nonferrous metal conductivity classification is proposed for five nonferrous metals with respect to the change of tilt angle that may occur in the process of metal classification.The method obtains the characteristic slope by fitting a straight line to the peak point of the amplitude of the mutual inductance trajectory at different tilt angles,and based on this,a metal classification method based on the minimum error of the characteristic slope is proposed.Test results show that the technique can accurately classify five nonmagnetic metals with tilt angles within 16.7°,and the accuracy can reach 96.4% when the tilt angle is extended to 45°.The classification system uses only eddy current sensors and has a fast response time for single-frequency eddy current detection,which will enable future industrial systems for realtime metal classificationThe classification method of non-ferrous metals at different lifting heights was studied.By normalizing the mutual sensing trajectory based on the maximum value of the real part,and based on the fact that the mutual sensing data of different metals are separated from each other and do not intersect during the change of lifting height when the normalized real part is 0.5,the K-mean clustering algorithm was used to successfully separate 26×5 groups of data points and obtain the clustering centers of each type of data.The Y-values of each metal clustering center coordinate are separated from each other and do not intersect in the change range,thus proposing a classification method for non-ferrous metals at different lifting heights.The test results show that this method can accurately classify five metals in the range of 2.5mm-5mm lifting heights,and the classification rate can reach 98.5% in the range of 2mm-5mm lifting heights.The proposed method can still classify effectively under the condition of vibration in the actual transmission,and the mean value point is used to compensate for the mutual inductance change caused by tilt,so that this classification method can be used to successfully classify the non-ferrous metals whose lifting height and tilt angle both change.