Abstract: Aiming at the change of power and energy consumption in micro-milling process, the processing power of the spindle system of micro-milling machine tool was collected. A BP neural network prediction model was established to predict the effect of three important cutting parameters, spindle speed, feed per to-oth and cutting depth, on the specific cutting energy (SCE). Through the 45# steel radial tire die micro-milling test, the test data samples were obtained totrain and detect the BP neural network, and the prediction of SCE under the combination of different cutting parameters was realized, and genetic algorithm (GA) was used to optimize the cutting parameters. The prediction and optimization results show that the minimum specific cutting energy can be obtained under thecombination of the maximum cutting parameters. Therefore, without considering the surface roughness and tool wear, a high level cutting parameters combinationcan obtain large material removal rate and relatively small specific cutting energy to improve processing efficiency and reduce processing energy consumption.