Aluminum alloys and copper alloys are typical low-resistance metals, which are widely used in the electrical industry. The adjustable-ring-mode (ARM) laser welding, with its advantages of small heat-affected zone, fast welding speed, and almost no spatter, is an ideal method for aluminum-copper dissimilar metal joining. Due to the differences in physical properties such as melting point and thermal conductivity between aluminum and copper, the seam width at the interface (SWI) tends to fluctuate significantly, which affects the electrical and mechanical properties of the joint. Therefore, monitoring SWI is an important method for evaluating the aluminum-copper dissimilar metal joint quality. In this study, a novel real-time aluminum-copper dissimilar metal SWI monitoring method based on optical coherence tomography (OCT) keyhole depth signals and plasma plume spectral signal in ARM laser welding is proposed. Based on the analysis of correlations of several features on the cross-section, a method for characterizing the SWI based on the upper and lower materials melted volumes and the penetration depth was proposed. Signal processing techniques are applied to denoise and analyze OCT and spectral signals, confirming the strong correlation between multiple signals and key features in the SWI characterizing model, which can be used for SWI prediction. Finally, based on multi-signal diagnosis, a backpropagation neural network (BPNN) model for SWI prediction is established. The results show that the average error of this method is only 10.7 μm, achieving high-precision prediction of SWI in aluminum-copper dissimilar metal ARM laser welding.