The optimal value of the PPM-Shift related with the binary TH-PPM should be cautiously selected to make sure the finest performance of the PPM-TH-UWB system for a given pulse shape utilized in a specific submission. In addition, it has been demonstrated that the ideal choice of the PPM-Shift parameter denotes back to the autocorrelation characteristics of the pulse shape under examination. In this paper, artificial neural network (ANN) with Levenberg-Marquardt Learning Algorithm (LM) is proposed to optimize PPM-shift for the TH-UWB under Standard Gaussian Approximation (SGA) with multi-Rate system, this optimal PPM-Shift is a function of the data bit Rate (Rb) and Gaussian pulse and its derivatives. Theoretical analysis and simulation results show that our algorithm rapidly converges to the optimal PPM-Shift and has the advantages of low bit-error rate.