Combined with laser-induced breakdown spectroscopy (LIBS) technology, the laser cleaning online monitoring system is designed to monitor the quality of laser cleaning in real time. The fiber laser uses in the experiment can be processed and applied in a multidimensional space. First, we determine the laser cleaning speed value and study the change law of LIBS with laser single pulse energy density, which can characterize the cleaning effect of carbon fiber composite materials. Then, during data analysis processing, the method of removing the background by mean smoothing is used to process the continuous background of the envelope-like spectrum. The density-based spatial clustering of applications with noise algorithm is used to realize the separation of spectral noise and effective data. The Pearson coefficient analysis method determines the best ablation times for laser cleaning, and provides a basis for the automatic optimization control of the laser cleaning process. Finally, a scanning electron microscope is used to analyze the surface morphology of the carbon fiber, which confirmes the effectiveness of LIBS technology to monitor the laser cleaning effect online.