Abstract: Based on the high cost of multi-variety and small-batch parts processing, a numerical control machine tool energy consumption model was constructed based on artificial fish swarm genetic algorithm (AFSA-GA) to reduce the energy consumption of parts processing. Firstly, the power of CNC machine tool is divided into power model of each process, and the energy consumption model of machine tool is obtained based on the relationship between power model and working time. Combined with the product surface roughness model, the energy consumption model of each process and the overall roughness are normalized to form the overall energy consumption model. Secondly, taking energy consumption and roughness as the objective function, AFSA-GA algorithm is established, and the most appropriate machine power and its corresponding energy consumption and surface roughness are obtained by solving the energy consumption of each process. Finally, according to the optimal power obtained, the optimization results are verified, and a solution is provided for the actual processing of the five-axis machine tool.