<p>For global optimization problems with multi-peaks, an ultimate challenge of global optimization algorithms is how to avoid local minimizers for improving solution precision. The filled function algorithm can successfully transition from one local minimizer to a better one with the help of a filled function. However, the performance of such algorithms is compromised when the filled function contains parameters, ill-conditioned terms, or exhibits discontinulity and non-differentiability. A new continuously differentiable one-parameter filled function with specific properties is constructed to improve the efficiency of the filled function algorithm in numerical computation. Based on these properties, a new filled function algorithm is given, which only needs to iteratively minimize the filled function after obtaining a local minimizer of the objective function in the first iteration. Furthermore, the local optimal solution of this filled function can be searched in the entire search space without being limited to the neighborhood of a local minimizer. Comparative experiments on some global optimization problems are provided to illustrate the efficiency and feasibility of the new filled function algorithm.</p>