The annihilation of dark-matter particles may lead to the production of monochromatic gamma-ray emission. In this contribution, the search for spectral lines in the gamma-ray spectrum, using eight years of data collected with the space-borne Dark Matter Particle Explorer is presented. To improve the event selection, two machine-learning algorithms were developed and proved to outperform all the standard methods. No line signal was found between 5 GeV and 280 GeV in the considered region of interest. The constraints on the velocity-averaged cross section of the neutralino was estimated for the Einasto dark-matter density profile and compared to the one obtained with Fermi-LAT and H.E.S.S. data.