Radio frequency wireless power transfer is a promising technique for charging low-power devices without relying on wired infrastructure. However, achieving efficient multi-user energy beam focusing remains a significant challenge, particularly in the radiating near-field of extremely large antenna arrays (ELAA) and when considering practical nonlinear energy harvesting models. The spherical wavefront propagation in the near-field zone necessitates precise phase alignment across the array to maximize energy efficiency at the receiver location. In this paper, we propose a beam focusing framework comprising two complementary schemes. For line-of-sight dominant scenarios, we design a location-aware scheme based on a spherical wave channel model constructed from user position estimates. For position-agnostic conditions, we develop an iterative algorithm that operates using received power feedback. This dual-mode approach ensures adaptability and robustness across various operating environments and information constraints. The study is further extended to incorporate hardware constraints, specifically the use of finite-resolution phase shifters. Numerical results demonstrate that the proposed schemes are effective and can closely approach the performance of ideal continuous-phase implementations, highlighting their potential for practical applications.