Modelling of civilian ships' ferromagnetic signatures

In FFI project 1321 Teknologisk risikoreduksjon for fremtidig minesveip, the goal is to support the acquisition of future mine countermeasure capability for the Royal Norwegian Navy. A part of this support is recommending performance requirements for future influence sweep systems. We have available a large data set of measured civilian ship signature data from the EDA project SIRAMIS. To determine the necessary performance for new influence sweep systems an analysis of this data set should be performed. The measurement are largely targets of opportunity where ships pass the sensor arrays at different speeds and distances. In order to do an ”apples-to-apples“ comparison we therefore want to produce models of the measured ship signatures, which can then be evaluated under the same conditions. This report details a study evaluating different methods for fitting a model to the magnetic signature data. The Prolate Spheroidal Harmonic (PSH) model is a mathematical model that describes the magnetic field of a ship as the cumulative effect of a collection of magnetic multipoles placed in a prolate spheroidal coordinate system. In order to get a good description of the magnetic field of a ship using the PSH model, the coefficients determining the strength of each multipole must be found through some fitting procedure. It is important to avoid overfitting when optimising the model coefficients. The degree of overfitting in a optimised model can be found by evaluating the model’s predictive ability. Many different linear regression methods for optimising the model coefficients have been evaluated based on the predictive ability of the models they produce. The Lasso LARS method was found to give the least amount of overfitting and was chosen to produce models based on the data set. In addition a non-linear method for optimising the size of the prolate spheroidal coordinate system was used. Magnetic models were produced for the civilian ship signature measurements. Analysis of these ship models will form the basis for future recommendations of influence sweep performance requirements.