Abstract: In the processing information system, the particle swarm optimization algorithm is applied for the minimum attribute reduction, which is slow and easy to fall into local optimum. Accordingly, this paper proposes a quantum-behaved particle swarm optimization algorithm combined with cloud model（CQPSO） to reduce the number of attributes in data set. First, the speed of convergence is accelerated by using a quantum behavior of QPSO algorithm; and the cloud model is introduced into QPSO to control different particle swarms in different states; then, the attribute reduction mathematical model is constructed according to property dependency and other properties; finally, the CQPSO algorithm is used to solve the problem and achieve the reduction results. In this experiment, the CQPSO algorithm is simulated and compared by the standard test function, which shows that the CQPSO algorithm performance is better than the quantum-behaved PSO algorithm. And the UCI standard database is used to perform attribute reduction tests. The results show that the proposed attribute reduction method is superior to the existing reduction method, and its calculation speed is fast and the recognition precision is high.

Key words: attribute reduction, particle swarm optimization algorithm, quantum behavior, cloud model

摘要： 针对粒子群优化算法在处理信息系统中属性约简收敛速度慢、早熟的问题,提出了一种结合云模型的量子粒子群优化算法（CQPSO）的属性约简方法。改进量子粒子群优化算法，即利用量子粒子群算法的量子行为来加快收敛速度；引入云模型控制粒子种群在不同状态下进行寻优；根据属性依赖度等性质构造属性约简数学模型；采用CQPSO算法对其进行求解，得到约简结果。实验中采用标准测试函数对CQPSO算法进行仿真对比，验证了CQPSO算法性能优于量子PSO算法；采用UCI标准数据库的典型例子进行属性约简测试，结果表明提出的属性约简方法优于现有约简方法，其计算速度快、识别精度高。

关键词: 属性约简, 粒子群优化算法, 量子行为, 云模型