The identification of influential nodes in complex networks has attracted much attention due to their significant theoretical significance and wide applicability. When designing an identification method for an unweighted network, existing methods also consider edges. In this paper, we propose a new improved K-Shell algorithm based on weight for complex networks, short for CNW-IKS algorithm. We use edge load-bearing and edge influence factor measure the local features of nodes, turning the problem of unweighted network to a weighted network. With the help of SIR information dissemination model to verify the validity and accuracy of CNW-IKS algorithm, the real social network simulation results show that the CNWIKS algorithm is more accurate for the division granularity to influence the size of the nodes. This method can provide theoretical support for the application of public opinion control and advertising marketing in complex networks.