Tunneling function method is an effective method for global optimization problems. Its ability to jump out of the local minimizer is deeply affected by the properties of the tunneling function. With the complexity of practical optimization problems, the corresponding forms of tunneling functions become more complex. Therefore, constructing tunneling functions with simple form and good properties is one of the main research objectives of the tunneling function method. In order to improve the efficiency of the tunneling function method for solving multi-modal functions, a new tunneling function is constructed. Its minimizers are not only the better feasible points of the objective function than the current local minimizer, but also the better local minimizers, i.e., the  tunneling function and the objective function have the same local minimizers. Thus, the better local minimizer of the objective function can be obtained directly by minimizing the tunneling function. Based on this feature, a new tunneling function algorithm is designed. It changes the frame of conventional tunneling function methods that objective function and tunneling function are minimized alternately, and can effectively reduce the iterations of local optimization and accelerate the speed of global optimization. Theoretical analysis and numerical experiments exhibit the feasibility and effectiveness of the algorithm.