Detection and localisation of magnetic objects

Abstract

Detection and localisation of hidden magnetic objects is studied. The concomitant disturbance of the earth magnetic field is exploited. Magnetic fields and or gradients are measured with a number of sensors. Two point models for magnetic fields are selected for the analysis: the magnetic dipole and its quadrupole extension. It is shown how to apply least squares analysis of the data in order to detect and localise magnetised objects. The theoretical framework is tested in simulations, including noise. Two methods to do the minimisation are chosen. The formalism performs well in the simulations. Detection and localisation of the model objects is successful. No decisive differences in performance of the two minimisation methods are found. Results for the position parameters are reliable in any case. We have evaluated the goodness of fit in our simulations. Theoretical and simulated chi-square distributions are compared for different fit models and noise distributions with satisfactory results. The obtained parameter distributions are Gaussian. Measurements are performed to test the algorithms on real data collected for three ferromagnetic objects: a magnet, pipe and block. The magnet is easily localised, the pipe reasonably well. The block cannot be localised: its field variations cannot be distinguished from background field variations. Subsequent simulations show that using gradiometers improves the performances of the algorithms. The effects of a number of controllable parameters are evaluated. © The Institution of Engineering and Technology 2007.