To address the prominent issues of poor voltage quality and high active power losses in remote mountainous power grids caused by limited investment and long power supply distances， an innovative source-grid-load-storage coordinated planning method for distribution networks is proposed. Firstly， the characteristics of mountainous power grids are systematically analyzed from six dimensions such as geographical environment， economic conditions， and load characteristics. Secondly， a bi-level optimization model is constructed to optimize the siting and sizing of photovoltaic power stations， reactive power compensation equipment， and energy storage systems， while also formulating optimal operational dispatch strategies. Finally， in terms of algorithm design， the upper level employs an improved artificial lemming optimization algorithm， and the lower level is solved using an improved non-dominated sorting genetic algorithm combined with a fuzzy membership degree method. The results demonstrate that the proposed model， under investment constraints， effectively improves the voltage qualification rate， enhances voltage quality， and significantly reduces the line loss rate. Furthermore， the proposed solution method shows marked improvements in both accuracy and computational efficiency compared to traditional methods， providing new technical support for the planning of power grids in remote mountainous areas.