基于KPCA-RBFNN的激光拼焊性能预测

Performance Prediction of Laser Tailored Welding Based on KPCA-RBFNN

高精度的激光拼焊是激光技术应用中的研究重点,针对当前激光拼焊性能预测精度低的不足,提出一种基于KPCA-RBFNN的激光拼焊性能预测方法。首先对激光拼焊系统的工作原理进行分析,然后采用核主成分分析对焊接工艺参数进行优化和选择,得到RBF神经网络的输入向量,最后采用RBF神经网络建立激光拼焊性能预测模型。仿真测试结果表明,KPCA-RBFNN可以描述激光拼焊性能与对焊接工艺参数间的关系,获得了比其他模型更高的激光拼焊性能预测精度,预测结果可以为激光拼焊人员提供有价值的参考信息。

High precision of laser welding is the focus in the application of laser technology,and low precision of the performance prediction model for laser tailor welding,so this paper puts forward a prediction model based on the KPCA-RBFNN for laser tailor welding performance. Firstly,the laser spell welding system working principle were analyzed,and then the kernel principal component analysis is used to optimize and select the welding process parameters and were taken as input vector of the RBF neural network,Finally,laser welding performance prediction model is established by using RBF neural network. Simulation and test results show that the proposed model can describe the relationship between laser tailor welding performance and process parameters of welding,obtained higher prediction precision than other models laser welding performance prediction models,and the prediction results can provide valuable reference information for laser welding personnel.