You need JavaScript to view this

Development of a signal-analysis algorithm for the ZEUS transition-radiation detector under application of a neural network; Entwicklung eines Signalanalyse-Algorithmus fuer den ZEUS Uebergangsstrahlungs-Detektor unter Anwendung eines Neuronalen Netzwerks

Abstract

The aim of this thesis consisted in the development of a procedure for the analysis of the data of the transition-radiation detector at ZEUS. For this a neural network was applied and first studied, which results concerning the separation power between electron an pions can be reached by this procedure. It was shown that neural nets yield within the error limits as well results as standard algorithms (total charge, cluster analysis). At an electron efficiency of 90% pion contaminations in the range 1%-2% were reached. Furthermore it could be confirmed that neural networks can be considered for the here present application field as robust in relatively insensitive against external perturbations. For the application in the experiment beside the separation power also the time-behaviour is of importance. The requirement to keep dead-times small didn`t allow the application of standard method. By a simulation the time availabel for the signal analysis was estimated. For the testing of the processing time in a neural network subsequently the corresponding algorithm was implemented into an assembler code for the digital signal processor DSP56001. (orig./HSI) [Deutsch] Das Ziel dieser Arbeit bestand in der Entwicklung eines Verfahrens zur Analyse der Daten des Uebergangsstrahlungsdetektors bei Zeus. Dazu wurde  More>>
Authors:
Publication Date:
Jul 01, 1992
Product Type:
Thesis/Dissertation
Report Number:
BONN-IR-92-55
Reference Number:
SCA: 440104; 990300; PA: DEN-94:0F6567; EDB-94:077008; ERA-19:019436; NTS-94:023598; SN: 94001203492
Resource Relation:
Other Information: TH: Diplomarbeit; PBD: Jul 1992
Subject:
46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND TECHNOLOGY; 99 GENERAL AND MISCELLANEOUS//MATHEMATICS, COMPUTING, AND INFORMATION SCIENCE; DATA PROCESSING; TRANSITION RADIATION DETECTORS; HERA STORAGE RING; ELECTRON DETECTION; PARTICLE DISCRIMINATION; ALGORITHMS; COMPUTER CODES; NEURAL NETWORKS; DIGITAL COMPUTERS; PROCESS COMPUTERS; EFFICIENCY; DEAD TIME; PION DETECTION; 440104; 990300; HIGH ENERGY PHYSICS INSTRUMENTATION; INFORMATION HANDLING
OSTI ID:
10150576
Research Organizations:
Bonn Univ. (Germany). Physikalisches Inst.; Bonn Univ. (Germany). Mathematisch-Naturwissenschaftliche Fakultaet
Country of Origin:
Germany
Language:
German
Other Identifying Numbers:
Journal ID: ISSN 0172-8741; Other: ON: DE94766581; TRN: DE94F6567
Availability:
OSTI; NTIS (US Sales Only); INIS
Submitting Site:
DEN
Size:
82 p.
Announcement Date:
Jul 05, 2005

Citation Formats

Wollschlaeger, U. Development of a signal-analysis algorithm for the ZEUS transition-radiation detector under application of a neural network; Entwicklung eines Signalanalyse-Algorithmus fuer den ZEUS Uebergangsstrahlungs-Detektor unter Anwendung eines Neuronalen Netzwerks. Germany: N. p., 1992. Web.
Wollschlaeger, U. Development of a signal-analysis algorithm for the ZEUS transition-radiation detector under application of a neural network; Entwicklung eines Signalanalyse-Algorithmus fuer den ZEUS Uebergangsstrahlungs-Detektor unter Anwendung eines Neuronalen Netzwerks. Germany.
Wollschlaeger, U. 1992. "Development of a signal-analysis algorithm for the ZEUS transition-radiation detector under application of a neural network; Entwicklung eines Signalanalyse-Algorithmus fuer den ZEUS Uebergangsstrahlungs-Detektor unter Anwendung eines Neuronalen Netzwerks." Germany.
@misc{etde_10150576,
title = {Development of a signal-analysis algorithm for the ZEUS transition-radiation detector under application of a neural network; Entwicklung eines Signalanalyse-Algorithmus fuer den ZEUS Uebergangsstrahlungs-Detektor unter Anwendung eines Neuronalen Netzwerks}
author = {Wollschlaeger, U}
abstractNote = {The aim of this thesis consisted in the development of a procedure for the analysis of the data of the transition-radiation detector at ZEUS. For this a neural network was applied and first studied, which results concerning the separation power between electron an pions can be reached by this procedure. It was shown that neural nets yield within the error limits as well results as standard algorithms (total charge, cluster analysis). At an electron efficiency of 90% pion contaminations in the range 1%-2% were reached. Furthermore it could be confirmed that neural networks can be considered for the here present application field as robust in relatively insensitive against external perturbations. For the application in the experiment beside the separation power also the time-behaviour is of importance. The requirement to keep dead-times small didn`t allow the application of standard method. By a simulation the time availabel for the signal analysis was estimated. For the testing of the processing time in a neural network subsequently the corresponding algorithm was implemented into an assembler code for the digital signal processor DSP56001. (orig./HSI) [Deutsch] Das Ziel dieser Arbeit bestand in der Entwicklung eines Verfahrens zur Analyse der Daten des Uebergangsstrahlungsdetektors bei Zeus. Dazu wurde ein Neuronales Netzwerk eingesetzt und zunaechst untersucht, welche Ergebnisse im Hinblick auf die Separationsleistung zwischen Elektronen und Pionen mit diesem Verfahren erzielt werden konnten. Es zeigte sich, dass Neuronale Netze innerhalb der Fehlergrenzen ebensogute Resultate liefern, wie Standardalgorithmen (Gesamtladung, Clusteranalyse). Bei einer Elektroneffizienz von 90% wurden Pionkontaminationen im Bereich 1%-2% erreicht. Es konnte darueberhinaus sichergestellt werden, dass Neuronale Netze fuer das hier vorliegende Einsatzgebiet als robust und gegen aeussere Stoerungen relativ unempfindlich gelten koennen. Fuer den Einsatz im Experiment ist, neben der Trennleistung, auch das Zeitverhalten von Bedeutung. Die Forderung, Totzeiten klein zu halten, liess die Verwendung der Standardmethoden nicht zu. Durch eine Simulation wurde die fuer die Signalanalyse zur Verfuegung stehende Zeit abgeschaetzt. Zur Ueberpruefung der Verarbeitungsdauer bei einem Neuronalen Netzwerk wurde im Anschluss der entsprechende Algorithmus in ein Assemblerprogramm fuer den Digitalen Signalprozessor DSP56001 umgesetzt. (orig./HSI)}
place = {Germany}
year = {1992}
month = {Jul}
}