By taking a structure of the moving arm of hydraulic excavator as the research object, the mathematical model of optimal design taking performance into consideration of static strength, stiffness and lightweight was built. Based on the software ANSYS software as tools, the design and calculation for lightweight on the moving arm of hydraulic excavator were completed. The model of the neural network was built by aiming at the nonlinearity characteristics of the lightweight design variable in process, so as to verify the optimization results by ANSYS software, and the weight of the moving arm is significantly reduced. Results indicate that, this method combined with ANSYS and neural network is reasonable and feasible for lightweight design, and it has a certain guiding significance for lightweight design on other machineries.