Top-up operation has been started since many years ago at Taiwan Light Source Storage Ring (TLS-SR). To realize the operation of the ring by top-up injection mode, a coherent oscillation of the stored beam is one of serious problem. This phenomena is exited in every injection period due to error in the injection pulse-magnets. For this operation it is important to reduce the beam injections should not excite the oscillation of stored beams. Artificial neural network (ANN) technology was used to analyze and optimize the injection pulse-magnets parameters of the storage ring. The results of this research are discussed in this study.