Extremum seeking for parameter identification, implementation for electron beam property prediction

We report on an experiment performed at the Facility for Advanced Accelerator Experimental Tests (FACET), which is the first two kilometers of the Stanford Linear Accelerator Center (SLAC) linear accelerator, in which a new form of extremum seeking, one with known, bounded update rates, despite operating on an analytically unknown cost function, was utilized in order to provide a real time bunch length estimate of the electron beam. The approach was to simultaneously tune fourteen parameters, such as arbitrary klystron phase shifts and electron bunch energy, in order to match a simulated (LiTrack) bunch energy spread spectrum with the real time wiggler/scintillating YAG crystal signal. The simple adaptive scheme was digitally implemented using Matlab and the Experimental Physics and Industrial Control System (EPICS). The main result is the development of a non-intrusive, non-destructive real-time diagnostic scheme for prediction of bunch length, as well as other beam parameters, the precise control of which is very important for the plasma acceleration scheme being explored at FACET.