7075 aluminum alloy has been widely applied in the field of aerospace and marine sheet metal because of its protruding mechanical and corrosion resistance. In this paper, the problem of selecting optimal process parameters to optimize multiple processing variables had been studied in precision manufacturing. Multi-objective particle swarm optimized neural networks system was put forward to determine the optimal cutting conditions with multi-objective particle swarm algorithm and multiple neural networks as prediction models of machining variables. Precision parts manufacturing of 7075 aluminum alloy would go through two operations of material removal and surface forming. Firstly, optimal cutting conditions were determined to minimize tool wear while maximizing metal removal rate in material removal stage. Secondly, it was significant and meaningful to select optimal cutting conditions corresponding to the best surface quality and minimum root mean square of tool vibration in surface forming stage. Orthogonal experiments had been carried out to observe the relationship between machining-related variables and cutting parameters in detail. Multiple neural networks were trained to establish predictive models of cutting process from orthogonal experimental and statistical data. Maximum deviation theory sorted the Pareto solutions searched by optimization process of neural networks driven by multi-objective particle swarm algorithm. The top ranking Pareto solutions had been determined as the optimal cutting parameters combination for material removal and surface forming stages, respectively. Finally, the proposed optimization system can also be used to optimize the processing of other difficult-to-machine materials.

7075铝合金因其突出的机械性能和耐腐蚀性而被广泛应用于航空航天和钣金领域。本文研究了在精密制造中选择最佳工艺参数以优化多个工艺变量的问题。提出了多目标粒子群优化神经网络系统，以多目标粒子群算法和多神经网络作为加工变量的预测模型，确定最优切削条件。7075铝合金的精密零件制造将经历材料去除和表面成型的两个操作。首先，确定最佳切削条件以最大程度地减少工具磨损，同时在材料去除阶段最大化金属去除率。其次，在表面成形阶段选择与最佳表面质量和最小刀具振动均方根相对应的最佳切削条件，具有重要意义。已经进行了正交实验以详细观察加工相关变量与切削参数之间的关系。训练了多个神经网络，以根据正交实验和统计数据建立切削过程的预测模型。最大偏差理论对由多目标粒子群算法驱动的神经网络优化过程搜索的Pareto解进行排序。排名最高的帕累托解决方案已确定为分别用于材料去除和表面成型阶段的最佳切削参数组合。最后，