粒子群算法优化的人工神经网络预测Ni-Fe合金镀层的性能
|
摘要
建立了粒子群算法优化的人工神经网络预测模型。以工艺参数为输入变量,以单因素试验得到的Ni-Fe合金镀层的性能指标为输出变量,将粒子群算法优化的人工神经网络预测模型的预测结果与传统BP神经网络预测模型的预测结果进行了比较。结果表明:粒子群算法优化的人工神经网络预测模型具有更高的预测精度。通过建立模型得到了各个工艺参数对Ni-Fe合金镀层性能指标的评价指标权重。当电流密度为1.0～1.5A/dm2、镀液温度为45℃、搅拌速率为1 000～1 200r/min时,Ni-Fe合金镀层的表面粗糙度和腐蚀速率均处于较低水平。
Particle swarm optimization optimized artificial neural network prediction model(PSO Model) was established. The process parameters were used as input variable,and the performance index of Ni-Fe alloy coatings obtained through the single-factor experiment was used as output variable.The prediction results of PSO Model was compared with that of conventional BP neural network prediction model.The results showed that PSO Model has higher prediction accuracy,and the evaluation index weight of each process parameter to the performance index of Ni-Fe alloy coatings was obtained through the establishment of models.It was determined that under the conditions of curren density 1.0～1.5 A/dm2,temperature 45 ℃ and stirring rate1 000～1 200r/min,the surface roughness and corrosion rate of Ni-Fe alloy coating were both at a low level.
Particle swarm optimization optimized artificial neural network prediction model(PSO Model) was established. The process parameters were used as input variable,and the performance index of Ni-Fe alloy coatings obtained through the single-factor experiment was used as output variable.The prediction results of PSO Model was compared with that of conventional BP neural network prediction model.The results showed that PSO Model has higher prediction accuracy,and the evaluation index weight of each process parameter to the performance index of Ni-Fe alloy coatings was obtained through the establishment of models.It was determined that under the conditions of curren density 1.0～1.5 A/dm2,temperature 45 ℃ and stirring rate1 000～1 200r/min,the surface roughness and corrosion rate of Ni-Fe alloy coating were both at a low level.
引文
[1]周春桂,王剑,谢石林,等.基于神经网络的复杂系统杂交建模研究[J].测试技术学报,2007,21(1):75-78.
[2]李兴远,夏法锋,陈达.BP神经网络预测脉冲电沉积Cu-SiC镀层镀速的研究[J].兵器材料科学与工程,2013,36(3):99-101.
[3]王金东,赵岩,高媛媛,等.基于BP神经网络的Ni-Al2O3镀层粒子复合量预测研究[J].功能材料,2016,47(1):1226-1228.
[4]李远会,郭忠诚,万明攀,等.用BP神经网络和遗传算法优化电沉积Cu-W的工艺参数[J].电镀与环保,2015,35(2):12-13.
[5]贾丽会,张修如.BP算法分析与改进[J].计算机技术与发展,2006,16(10):101-103.
[2]李兴远,夏法锋,陈达.BP神经网络预测脉冲电沉积Cu-SiC镀层镀速的研究[J].兵器材料科学与工程,2013,36(3):99-101.
[3]王金东,赵岩,高媛媛,等.基于BP神经网络的Ni-Al2O3镀层粒子复合量预测研究[J].功能材料,2016,47(1):1226-1228.
[4]李远会,郭忠诚,万明攀,等.用BP神经网络和遗传算法优化电沉积Cu-W的工艺参数[J].电镀与环保,2015,35(2):12-13.
[5]贾丽会,张修如.BP算法分析与改进[J].计算机技术与发展,2006,16(10):101-103.