The Internet of Things (IoT) integrates a large number of distributed nodes to collect or transmit data. When the network scale increases, individuals use multiple sink nodes to construct the network. This increases the complexity of the network and leads to significant challenges in terms of the existing methods with respect to the aspect of data forwarding and collection. In order to address the issue, this paper proposes a Shortcut Addition strategy based on the Particle Swarm algorithm for multi-sink network (SAPS). It constructs network topology with multiple sinks based on a small-world network.In SAPS, we create a fitness function by combining the average path length and load of the sink node to evaluate the quality of a particle. Subsequently, crossover and mutation are used to update particles to determine the optimal solution. The simulation results indicate that SAPS is superior to the GMSW and LM-GAS in terms of the average path length, load balance, and number of added shortcuts.