Improved collaborative optimization based on support vector regression and particle swarm optimization algorithm was researched. The basic principle of collaborative optimization and support vector regression was represented, and in order to resolve the difficulty in system-level coordination, improve convergence performance and efficiency, approximate models of constraint conditions in system-level were constructed using support vector regression, and particle swarm optimization algorithm was introduced to the system-level optimization and disciplinary-level optimization. Simulation results show that the improved collaborative optimization can effectively resolve multidisciplinary design optimization problems, and compared to standard collaborative optimization, optimization accuracy is higher, system-level iterative operation is less, and the stability is better. All those can provide theoretical reference for the research of multidisciplinary design optimization.