This paper addresses the critical guidance problem for launch vehicles performing trans-lunar injection missions with low thrust-to-weight ratios, where the extended flight duration and substantial multi-source uncertainties present formidable challenges to conventional closed-loop guidance methods. A novel hybrid guidance approach integrating second-order cone programming with neighboring optimal control is developed to overcome these challenges. The proposed methodology begins with establishing a comprehensive trans-lunar injection problem model that explicitly accounts for multi-source uncertainties. A hybrid guidance strategy based on real-time trajectory planning is formulated to handle these uncertainties effectively. The main innovation is the application of a real-time trajectory planning technique that uses second-order cone programming in conjunction with Gauss pseudospectral discretization. By synergistically applying the costate mapping theorem and exploiting the dual output of second-order cone programming, the method achieves high-precision costate estimation, which provides the essential foundation for the neighboring optimal control component. A trajectory correction method based on neighboring optimal control is developed to address situations where uncertainties cause significant trajectory deviations. This method generates high-quality initial references for the second-order cone programming solver, substantially improving convergence characteristics and computational efficiency. Extensive numerical experiments demonstrate the superiority of the proposed guidance method in trans-lunar injection missions. The computational performance shows remarkable improvement, with the average computation time reduced to 75.353 milliseconds per guidance cycle, representing a 30.67% reduction. In comparison to current methods, this hybrid guidance method’s successful demonstration offers a novel and feasible technical pathway for trans-lunar injection missions.