Targeting Algorithm Based On Region Classification Using Numerical Model

It is important for military airport safety to detect the unexploded ordnance (UXO), besides, theability to do rapid ordnance clearance would become the key point for the war time recovery ofmilitary airport. The UXO can be detected rapidly, accurately and low costly using magnetic detector,however, the vector magnetic detection procedure is facing up to huge difficulty due to strong noise inthe process of which it brings in. A new algorithm which uses scalar quantity to solve the position ofattenuation source target and to estimate the influence of noise is brought up to deal with the problemmentioned above using numerical model and pattern identification theory.Firstly, the properties of two main kinds of noise which are brought in by magnetoresistivemagnetometer during the process of detection are analyzed with data from experiment. In this chapter,a formation of noise sensor and an algorithm of noise filtering are designed for foundational datapreprocessing in which a smoothing method is discussed and those work mentioned above lay thefoundation for pattern recognition method designed for the algorithm.Then, the formation classification algorithm is discussed with the exponential decay sourcemodel and the region with four sub-regions is chosen after discussion and comparison as the finalformation region. The formation of sources is calculated using measured data and neural networks andthis process is so called formation classification.At last, the basic ideas of two accuracy algorithms are elaborated and the one accuracy algorithmwhich uses error gradient descent procedures is discussed in detail. The relationship betweenmeasured data error and position error is analyzed on the basis of simulation results; however, theattempt to get the analytical expression of the relationship using parameter identification is failed eventhough it is still believed to be an essential way to keep going on research.The formation classification algorithm and the accuracy algorithm are able to locate the targetsand estimate the random noise effectively according to simulation results and the problems mentionedabove can also be solved in this way.