Design and Implementation of a Real-Time System Survey Functionality for Beam Loss Monitoring Systems

This PhD thesis describes the design and implementation of a real-time system survey functionality for beam loss monitoring (BLM) systems, with a particular focus on signal processing aspects. In Part I, I introduce the basic concepts fundamental to the project. I give an overview of the European Organization for Nuclear Research (CERN), where the bulk of this research work was carried out. I introduce the basic principles of BLM systems, followed by details of the new BLM system under installation and commissioning targeted by the project. I also present a survey of the related literature, along with the first measurement and simulation results related to the identification of the BLM system. The switching operation of the high voltage power supplies results in a parasitic signal in the acquisition chain of the new beam loss monitoring system. Part II presents a modification of the adaptive Fourier analyzer aimed at detecting these spurious components, with a view to realizing a noninvasive system survey. I introduce a novel frequency adaptation method, universally usable for resource-efficient frequency adaptation in a Fourier analyzer. Part III presents another approach to surveying the new BLM system, relying on an external excitation applied to the detectors. The current best method and the suggested new solution are described, the latter complemented by implementational considerations and some operational experience. Part IV introduces methods intended to aid detecting the presence of an arbitrary predefined pattern repeated periodically in a digitized data stream. I present several novel modifications to the Fourier analyzer facilitating pattern detection, with some also making it possible to detect additive disturbances in the frequency bands covered by the pattern. I show how a modified Fourier analyzer can be used to realize a lightweight time-variant cross-correlation filter, which produces the peak cross-correlation value in every time step when properly synchronized to the input signal. All the structures I present need to be synchronized to the input signal, for which I also introduce two algorithms.