A simulation study of the global orbit feedback system for Pohang Light Source

This paper describes the simulation of the global orbit feedback system usi ng the singular value decomposition (SVD) method, the error minimization me thod, and the neural network method. Instead of facing unacceptable correct ion result raised occasionally in the SVD method, we choose the error minim ization method for the global orbit feedback. This method provides minimum orbit errors while avoiding unacceptable corrections, and keeps the orbit w ithin the dynamic aperture of the storage ring. We simulate the Pohang Ligh t Sourer (PLS) storage ring using the Methodical Accelerator Design (MAD) c ode that generates the orbit distortions for the error minimization method and the learning data set for neural network method. In order to compare th e effectiveness of the neural network method with others, a neural network is trained by the learning algorithm using the learning data set. The globa l response matrix with a minimum error and the trained neural network are u sed to the global orbit feedback system. The simulation shows that a select ion of beam position monitors (BPMs) is very sensitive in the reduction of rms orbit distortions, and the random choice gives better results than any other cases.