Abstract: In view of problems of existing positioning algorithms for mine proximity detection that ranging error is large due to non-line-of-sight and other factors, and solution of ranging equation set does not converge under complex conditions, a robust positioning algorithm for mine proximity detection was proposed, namely weighted LM algorithm, which combines Gauss-Newton algorithm, weighted least square algorithm and Levenberg-Marquardt algorithm. The algorithm adds ranging error information to nonlinear iterative solution by adding weight, and adds damping coefficient in iterative process, which greatly improves positioning stability and robustness under precondition of ensuring convergence speed of iteration. Test results show that the weighted LM algorithm has high positioning efficiency and accuracy.