Integration testing is a crucial phase in software testing, and determining the sequence of class integration testing is a key issue within it. A reasonable class integration test sequence can reduce the cost of test stubs, thereby lowering the overall testing cost. When class integration test sequences are generated for large-scale programs, the particle swarm optimization (PSO) algorithm is prone to premature convergence. To address this, this paper proposes a method for generating class integration test sequences on the basis of a simulated annealing particle swarm optimization (SAPSO) algorithm. When particles fall into a local optimal region, the introduction of an annealing mechanism enables them to quickly escape this region and continue searching. The experimental results demonstrate that this method can effectively slow the convergence of the PSO algorithm, thereby yielding a better solution.