Unattended systems have received extensive attention in both military and civilian fields as an important application of autonomous technology. Currently, unattended systems are mainly used for the auxiliary monitoring of important areas, but lack intelligent disposition for the monitored results. Furthermore, traditional target assignment model for unattended systems has adaptability and interpretability concerns. To address those issues, this paper proposes an intelligent target assignment method based on fuzzy cognitive map (FCM) for unattended system. Initially, a situational map is constructed for the description of situation entities and their relationships for unattended systems. By integrating the dynamic inferential characteristics of FCM, an intelligent target assignment model for unattended systems based on FCM is established, achieving the knowledge modeling of the target assignment model. Subsequently, a hybrid learning algorithm based on particle swarm optimization (PSO) and gradient descent algorithm (GDA) is designed to improve the performance of the target assignment model. The introduction of flag operator, dynamic adjustment strategy, and adaptive moment estimation (Adam) algorithm further optimize the hybrid learning algorithm, accomplishing the data optimization of the proposed target assignment model. Experimental results demonstrate that the proposed intelligent model achieves promising results, highlighting its feasibility and potential practical applications.