Perturbed Lambert Problem forms the fundamental basis for tasks such as spacecraft rendezvous and on-orbit servicing. Due to the lack of an analytical solution for the perturbed Lambert problem， numerical solutions can only be obtained through iterative computations. Consequently， existing methods primarily focus on improving iteration convergence and computational efficiency. Building upon current homotopic iteration methods， this paper introduces the general concept of the A* algorithm for graph search and proposes a pruned homotopic iteration approach. Firstly， by combining the gradient direction of the state transition matrix from the terminal position error to the initial velocity increment with the target direction， an iteration direction is derived， and a homotopic mapping based on the A* algorithm is designed. Secondly， utilizing Taylor expansion， a linear perturbation matrix is devised， enabling pruning of divergent end-position paths. This addresses the issue of ineffective iterations due to initial velocity divergence that conventional methods such as Newton’s shooting method and quasi-linearization fail to exclude. Simulation results demonstrate that while ensuring the attainment of optimal solutions， our proposed method improves computational efficiency by over 25%， compared to existing homotopic methods， featuring a broader convergence range. Moreover， the method exhibits favorable characteristics regarding initial value convergence. In the case of orbit transfer in the Earth-Moon three-body system， it achieves more than a 30% increase in computational efficiency compared to quasi-linearization techniques.