To address power grid disturbances and power quality degradation caused by suboptimal charging facility deployment and disordered charging behaviors, this study establishes a user equilibrium-based bi-level joint planning model for diversified charging infrastructure. The upper-level model determines the optimal joint siting scheme by minimizing the total societal cost of traffic and power distribution networks, while the lower-level model refines the siting decisions through iterative feedback of user equilibrium states, forming a bidirectional optimization mechanism. To mitigate local optima limitations in mixed-integer surrogate optimization, a Radial Basis Function (RBF)-enhanced algorithm is proposed, significantly improving computational efficiency. Validations on a Sioux-Falls network coupled with a power distribution system demonstrate the method's capability to coordinate charging infrastructure deployment with grid operations, achieving substantial reductions in system-wide costs and reliable maintenance of power quality standards.