The invention discloses a hyperedge weighted hypergraph interference model and a distributed spectrum access method. The model is that microcellular access points in a network are regarded as vertexes, one hyperedge is limited to be able to comprise maximum vertexes, a hyperedge formation algorithm is used to construct a hypergraph interference model, and the number of interfered vertexes in eachhyperedge as a weight of the hyperedge. The method comprises the steps of modeling a spectrum access problem into a game model, wherein participants of the game are all the microcellular access pointsin the network; defining all the other access points in the hyperedge which includes the access point as neighbors thereof; randomly selecting one or multiple channels for access according to a needby all the access points, and calculating the currently suffered interference based on the channel access condition of each neighbor; and updating channel selection by the access point based on an improved space adaptive algorithm until the channel selection strategies of all the access points are unchanged. According to the hyperedge weighted hypergraph interference model and the distributed spectrum access method, the degree of interference during hypergraph conflict can be distinguished, and the interference at the interior of the network can be described more accurately.