Experimental Study of Location and Identification of Ferrous Spheroids Using a 'Smart' Total Field Magnetometer
Abstract
A microprocessor-controlled magnetometer which accurately locates and identifies compact ferrous objects in real-time is described. The person- portable instrument consists of a cart-mounted cesium magnetometer, optical encoder, microcontroller, interface and laptop computer. The instrument guides the operator to collect simultaneous magnetic field and position data in a horizontal plane above an object. Custom algorithms estimate location and dipole moment and use the latter to classify the object. Data collection takes 6 - 13 minutes, location and moment estimation 5 seconds, classification 30 seconds. Experiments using two ferrous spheroids and studies using magnetic total field and vertical component magnetic maps generated by a mathematical computer model are described. Limits of error in estimation of location and dipole moment, error in classification, and relative effects of sources of error are quantified. The rms error for location vector components was 0.019 m - 0.045 m compared to the average precision of 0.003 m - 0.005 m. The average magnitude of the difference between estimated and theoretical dipole moment vectors as a percentage of the theoretical dipole moment was 24.5 + or - 11.4% compared to precision of 0.51 - 8.21%. Pattern classification with a computer generated dipole moment design set is described. Deviation between experimental moment estimates and the computer model degraded performance but the misclassification rate for the two objects for which experimental measurements were made was 11. 1%. If an experimental design set were used, analysis shows that the limiting misclassification error for the present experimental precision should be between 2 and 5%.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1993
- Accession Number
- ADA263661
Entities
People
- John E. Mcfee
- Robert Ellingson
- Yogadhish Das