ASCo, Clayton, Victoria
Argonne National Laboratory, Office of Naval Research Project, Argonne
Monash University, Faculty of Science, Monash University
SLAC, Menlo Park, California
This paper describes the implementation of a neural network hybrid controller for energy and bunch length stabilization. The structure of the controller consists of a neural network (NNET) feed forward control, augmented by a conventional Proportional-Integral (PI) feedback controller to ensure stability of the system. The system is provided with past states of the machine in order to predict its future state, and therefore apply appropriate feed forward control. Experiments performed at the Australian Synchrotron showed the ability of the NNET to cancel multiple frequency energy jitter and the successful augmentation of the system by a PI algorithm. The LCLS experiments showed that the system can be expended to predict and correct coupled energy-bunch length deviations, and showed the improved jitter attenuation by the NNET system in comparison to the PI algorithm alone. Focus is also made on the machine response that needs to be accurately known to best operate the correction. When machine settings are modified, the response is re-calculated with the help of a model, and slight adjustments are made to optimize the energy jitter reduction as the control is operating.
