Link prediction for weapon system-of-systems (WSoS) combat networks, a fundamental problem in network science, strives to exploit implicit information for heterogeneous combat networks (HCNs), with significant military applications in network reconfiguration, key node identification, operation plan design, and so on. In this regard, a meta-path based HCN link prediction approach is proposed from the perspective of system-of-systems. More specifically, a WSoS is first calibrated as an HCN by abstracting heterogeneous weapons and interrelations among them as nodes and edges of different types respectively. Next, various metapaths are classified with the description of physical meanings according to their specified semantic information in HCNs. Then, a meta-path based WSoS combat network link prediction model is established using a node similarity index that is presented based on the metapath redundancy. Finally, extensive experiments are conducted on a real-world combat network case to demonstrate the feasibility and effectiveness of the proposed approach, which can achieve very good performance and substantially outperform the baseline methods.