基于神经网络和粒子群算法的切削参数现场实时优化

Using Network and Particle Swarm Optimizer to Realize Optimization of Cutting Parameter Real Time in Workplace

为了使金属切削加工中,能实现切削参数的实时优化,保证产品质量和设备效率,提出了一种新的切削参数最优化方法,引入加工时间、加工精度、加工成本三个目标控制量,建立了多目标非线性规划模型。并用惩罚函数法将多目标非线性约束规划问题转换成无约束非线性单目标规划问题。通过对神经网络和粒子群算法的有机结合,并充分利用了粒子群算法和BP网络各自具有的优点,对模型进行了求解。数值试验表明该方法能较好地解决切削参数的优化问题。

In metal cutting process, in order to improve the quality of products and raise the efficiency of equipment, new optimizations of cutting parameters method were put forward. Using three targets control parameters, machining time, machining accuracy and machining cost, multiple targets nonlinear programming model was established. The multiple targets nonlinear constraint program problem was transformed to single targets nonlinear no constraint program problem by using penalization function. With the combination of network and particle swarm optimizer, and making good use of the respective virtues, the model was solved. The experiment shows the mixes algorithm can realize optimization of cutting parameter real time in workplace.