【摘要】 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。更多还原

【Abstract】 Dilution of cladding layer is very important factor to influence the performance of forming metal parts and success of subsequent handling in process of laser cladding forming.A new method based evolutionary neural networks is proposed,and a model of calculating the dilution of cladding layer is established on basis of the improved learned arithmetic.The model combines the global optimization searching performance of the genetic algorithm and the localization of the BP neural networks.The experimental and simulant results show that the combination of BP and GA,can effectively overcome the problems of BP neural networks,and higher predicted accuracy is achieved.更多还原