基于BP神经网络的抛浆冲击后金属基体表面粗糙度控制研究

Study on surface roughness control of metal matrix after slurry blasting based on BP neural network

针对抛浆冲击后基体表面粗糙度缺乏精确控制模型的问题,基于BP神经网络,构建了抛浆工艺参数对基体表面粗糙度的BP神经网络预测模型,并以304不锈钢为研究对象进行了抛浆冲击试验,通过对样本数据进行方差分析检验,分析了各工艺参数对表面粗糙度的影响程度,由此确定了网络模型的输入和输出参数,以进行网络模型的训练和验证。结果表明,网络模型抛射速度的预测值和总样本的最大相对误差为4.80%;通过网格模型结果给定抛射速度试验得到的粗糙度值与目标值最大相对误差为2.40%。说明该神经网络模型能够指导工艺参数设定,实现冲击后钢板表面粗糙度的精准控制。

Aiming at the problem of the lacking of accurate control model for substrate surface roughness after slurry blasting,the BP neural network prediction model of the slurry blasting process parameters on substrate surface roughness was constructed based on BP neural networks.Taking 304 stainless steel as the research object for the slurry blasting impact tests,the influence degree of each process parameter on the surface roughness was analyzed by analysis of variance test on the sample data,from which the input and output parameters of the network model were determined for the training and validation of the network model.The results show that the maximum relative error between the predicted value of the network model blasting velocity and the total sample is 4.80%;the maximum relative error between the roughness value obtained experimentally for the given blasting velocity by the network model result and the target value is 2.40%.It demonstrates that the neural network model can guide the process parameter setting and realize the accurate control of surface roughness of steel plate after impacting.