The volume of data has experienced explosive growth which poses high requirements for computing power. The traditional centralized computing power scheduling model has many drawbacks, including low resource utilization, single-point failures, and lack of trust. Cross-domain collaborative computing power scheduling is applied between the supply and demand of computing power. This paper aims to combine the blockchain technology with federated learning for cross-domain collaborative computing power scheduling. Blockchain technology builds a trusted trading environment while federated learning helps the optimal resource allocation of the entire cross-domain computing power scheduling system. The experiment shows the design has high accuracy in task assignment. It also provides fast verification times and security protection.