Abstract: Basic particle swarm optimization (PSO) algorithm, which is a global and parallel optimization of high performance, simplicity, robustness, no problem specific information, etc., has been widely used in computer science, optimization of scheduling, function optimization and other fields. However, the basic PSO algorithm has the defects of premature convergence, stagnation phenomenon and slow convergence speed in the later evolution period for complex optimization problems. In order to overcome the premature convergence problem of basic PSO algorithm, using idea of multi-subpopulation and self-adaptive for reference, a novel multi-subpopulation adaptive polymorphic crossbreeding particle swarm optimization immune algorithm (MAPCPSOI) based on two-layer model is proposed. Through the bottom layer adaptive polymorphic crossbreeding PSO operation of several subpopulations, the MAPCPSOI algorithm, firstly, could ameliorate diversity of subpopulation distribution and effectively suppress premature and stagnation behavior of the convergence process. Secondly, the MAPCPSOI algorithm, by the top layer immune clonal selection operation of several subpopulations, could significantly improve the global optimization performance and further enhance the convergence precision. Compared with other improved PSO algorithms, simulated results of function optimization show that the MAPCPSOI algorithm, especially suitable for solving high-dimension and multimodal optimization problems, has rapider convergence speed and higher solution precision.