基于BP神经网络的紫外光源自适应固化控制研究

为解决紫外光动态固化技术中的固化不充分或反固化反应等问题,提出一种基于BP算法的LED紫外光源多参数自适应控制方法。利用神经网络优异的非线性逼近能力,并结合优化后的BP算法构建一个3输入2输出的网络预测模型。通过与多元线性回归和多元非线性回归模型的对比显示,BP神经网络算法有更高的拟合度。最后将57组数据导入训练好的模型进行验证。实验表明:该BP神经网络模型预测结果较好,且稳健性强,2输出参数预测值误差分别为1.86%和2.35%,可灵活应用于多种紫外光固化场合。

Research on adaptive curing control of ultraviolet light driver based on BP neural network

In order to solve the problem of insufficient curing or anti curing reaction in UV dynamic curing technology,a multi-parameter adaptive control method of LED UV light source based on the BP algorithm is proposed.A 3-input-2-output network prediction model is built based on the excellent non-linear approximation ability of the neural network and the optimized BP algorithm.Compared with the multiple linear regression model and multiple non-linear regression model,BP neural network algorithm has a higher fitting degree.Finally,57 groups of data are imported into the trained model for verification.The experimental results show that the BP neural network model has good prediction results and strong robustness.The two output parameters'prediction errors are 1.86%and 2.35%respectively,which can be flexibly applied to a variety of UV curing occasions.