Differential evolution (DE) is one kind of evolution algorithm based on difference of individuals. DE has exhibited good performance on optimization. However, for the high dimension and perplexed function, the algorithm is apt to fall into premature convergence, its performance is strongly influenced by the value of each strategy parameter including scale factor. Therefore, a modified differential evolution algorithm (MDE) was proposed to solve the optimization problems. First, the scale factor was randomly initialized and calculated by chaos each generation, which decreases the participation of user and balances the convergency speed and global optimal capability. Next, disaster factor was introduced to eliminate the individual of a small probability, along with a new individual generating, which can increase the diversity of population and global optimal capability. Simulated results and engineering optimization design example showed that MDE outperforms standard DE in global optimal capability. 