Cosmic rays at energies 10^18 eV and above are known as Ultra High Energy Cosmic Rays (UHECR). UHECR are charged particles that are accelerated by the biggest accelerators in our universe. Candidate accelerators generating these UHECR are super novas, black holes and neutron stars. But where and what these intergalactic accelerators is at large still unknown. One of the experiments in the forefront of research in this eld is JEM-EUSO, a planed space based telescope for detecting UHECR particles as they enter Earth's atmosphere. Made possible by the advances in photon detectors and light weighted Fresnel lenses. A ground based path nder experiment was carried out in 2015 called EUSO-TA to test the optics and photomultiplier technologies. When the UHECR enters the atmosphere it collides with the atoms generating a number of secondary particles which in turn interacts with other atoms in the atmosphere generating a cascade of secondary particles. These trails are known as Extensive Air Showers (EAS). Mostly electrons are generated and in turn they excites the nitrogen atoms in the atmosphere which generate a isotropic characteristic uorescence light. The JEM-EUSO telescope is designed to detect and measure the photon ux. From the photon ux it will be able to estimate the energy of the initial UHECR. JEM-EUSO will cover the largest area of EAS search and increase statistics of UHECR data. This thesis describes the method and development of algorithms made for EAS analysis and detection based on EUSO-TA data. A simulation of EUSO-TA focal surface was developed, simulating background, stars and EAS. The algorithms developed involves a background subtracting lter, line detection using Hough transform and a neural network for decision making. The Hough transform is used in computer vision and is a method used to detect lines in the pictures. It successfully identi ed both simulated and captured UHECR incoming direction with small errors. Neural network are a machine learning method used classi cation and regression problems. With the use of know example data simulated or real captured data a neural network can without explicit programing it, adjust its parameters to t the data. Based on method called supervised learning. The algorithms was programed in Python and using ROOT software to build the neural network. The resulting algorithm was able to successfully detect simulated data. Test on the EUSO-TA captured data shows a promising result but has to be developed and tested further.
| Identifer | oai:union.ndltd.org:UPSALLA1/oai:DiVA.org:ltu-313 |
| Date | January 2016 |
| Creators | Viberg, Fredrik |
| Publisher | Luleå tekniska universitet, Institutionen för system- och rymdteknik |
| Source Sets | DiVA Archive at Upsalla University |
| Language | English |
| Detected Language | English |
| Type | Student thesis, info:eu-repo/semantics/bachelorThesis, text |
| Format | application/pdf |
| Rights | info:eu-repo/semantics/openAccess |
Page generated in 0.0026 seconds