Most community detection methods rely on a single method to measure the connection strength between nodes, and lead to not perfect effects, and they are unable to adapt to different types of Network, either. In order to solve above problems, a novel community detection algorithm based on multi-attribute fusion strategy was used to investigate community detection. Firstly, from the view of multi-attribute fusion strategy, the proposed algorithm comprehensively measured multiple factors which affecting community detection, and introduced the modularity to determine weights of attributes to get more objective reference for detection and achieve more accurate effects. Secondly, in order to decrease excessive iteration times and improve efficiency of traditional methods, the algorithm further used artificial immune network with powerful parallel computing ability to find the core nodes of the community rapidly, and improved convergence rate and solved the local optimal problem by using dynamic operators, immune detection operators and opposition-learning. The experimental results show that the proposed algorithm gets better results on three different datasets of Zachary、Dolphin and College Football compared with traditional methods. It is concluded that the proposed algorithm achieves more accurate division effects in shorter time and has greater efficiency.