Based on the analysis of signal characteristics such as voltage and current of resistance spot welding, the cycle sequence of welding voltage, current effective value, dynamic resistance and heating power is extracted to construct input vectors, and BP and RBF neural network models are established to predict the size of the weld nugget of spot welding joints. The training and test results show that the network model trained by the overall normalized current and voltage effective value cycle sequence and the joint constructed input vector can effectively predict the size of the weld nugget of joints. The average verification error of the test samples is less than 6%, which is better than the input vectors such as dynamic resistance and heating power alone. The RBF network has a faster training speed than the BP network model, and the test verification error is generally slightly better than the BP network. It is a feasible and effective method for quality monitoring of resistance spot welding joints.
You Might Like
Recommended ContentMore
Open source project More
Popular Components
Searched by Users
Just Take a LookMore
Trending Downloads
Trending ArticlesMore