论文部分内容阅读
在激光熔覆成形金属制件工艺中,熔覆层稀释率大小对成形制件的性能以及后续工序的处理有至关重要的影响。设计了基于进化神经网络的学习算法,建立了熔覆层稀释率随工艺参数变化的预测模型,该模型结合了基因遗传算法的全局搜索能力和BP神经网络良好的局部性质。实验和模拟结果表明,基于进化计算的神经网络不仅可以克服单纯使用BP神经网络易陷入局部极小值等问题,而且预测精度较高,具有一定的实用价值。
In the laser clad metal parts manufacturing process, the cladding layer dilution rate of the shape of the workpiece performance and follow-up process has a crucial impact. A learning algorithm based on evolutionary neural network is designed and a prediction model of the dilution rate of the cladding layer with the process parameters is established. The model combines the global search ability of genetic genetic algorithm with the good local properties of BP neural network. The experimental and simulation results show that the neural network based on evolutionary computation can not only overcome the problems such as easy to fall into local minima by simple BP neural network, but also have high prediction accuracy and have certain practical value.