n order to solve the problem of low detection probability in the traditional spectrum sensing algorithm under low SNR environment, a clustering cooperative spectrum sensing algorithm based on improved double-threshold energy detection was proposed. The algorithm divided the sensing nodes into clusters. “OR” rule was used to the information between the clusters for hard fusion. Inner the clusters, to the nodes outside dual-threshold, a 1 bit decision result was transmitted for hard fusion; and to the nodes inside dual-threshold, energy values and SNRs were transmitted to implement weighted soft fusion. The detection probability and the node weight coefficient function were constructed under weighted soft fusion, and particle swarm optimization algorithm with compression factor was used to further optimize the function and maximize the detection probability. The simulation results verified that the algorithm has good detection probability under low SNRs and different number of users.