Abstract: A novel Hybrid Particle Swarm Optimization（HPSO） algorithm combined with the ideal of adaptive population mutation and individual annealing operation is proposed to solve the problem of slow searching speed and low computational precision of basic particle swarm optimization to blind source separation.Compared with Simulated Annealing（SA） and basic Particle Swarm Optimization（PSO），the HPSO is almost as simple for implementation as standard particle swarm optimization，but can carry on mutation adaptively and improve the abilities of seeking the global excellent result and evolution speed.The simulated results show that the effect of separation with the proposed approach is very well，and the blind source separation has not only the character of fast convergence speed but of stable performance.

Key words: acoustic signal separation, particle swarm algorithm, simulated annealing, adaptive mutation, negentropy, blind source separation

摘要： 为了解决基本粒子群盲分离算法收敛速度慢、优化精度低的问题，提出用基于群体自适应变异和个体退火操作的混合粒子群优化算法（HPSO）来实现听觉信号盲分离。与模拟退火算法（SA）和基本粒子群算法（PSO）相比，该算法保持了基本粒子群算法简单、容易实现的特点，又能进行自适应变异，改善了其摆脱局部极值点的能力。仿真对比结果表明，基于该改进算法的盲分离效果良好，具有收敛速度快、性能稳定等特点。

关键词: 听觉信号分离, 粒子群算法, 模拟退火, 自适应变异, 负熵, 盲分离