The application of evolutionary programming (EP) in power load prediction in simulated evolutionary optimization methods was discussed, and the gene coding, fitness function and variability value were studied, and the optimization combination prediction model was introduced, which further expanded the application prospect of evolutionary planning in power system. Example calculations show that this method not only overcomes the shortcomings that gradient optimization may be trapped in the local optimal solution or difficult to converge when it is close to the optimal solution, but also eliminates the large, complex and cumbersome calculations of conventional methods, and has the advantages of global optimization, parallel processing characteristics, good versatility and strong stability. So it’s effective.