In　this　article,　we　propose　a　continuous　monitoring　method　to　enhance　disaster　warning　capabilities　and　address　the　persistent　coverage　problem　in　multi-agent　systems.　To　enable　the　agent　to　continuously　perform　tasks,　we　transform　the　optimization　objective　function　under　energy　constraints　into　constraints　on　control　inputs.　Additionally,　the　energy　of　the　agent　is　constrained　through　the　control　barrier　function　(CBF),　and　a　corresponding　control　framework　is　designed　to　achieve　persistent　task　execution.　We　apply　the　proposed　method　to　landslide　early　warning　and　validate　its　effectiveness　and　superiority　in　persistent　coverage　through　numerical　simulation.