利用Alopex改进的粒子群优化算法及其在软测量建模中的应用

Improved Particle Swarm Optimization Algorithms by Alopex and Its Application in Soft Sensor Modeling

提出了一种利用A lopex算法改进的粒子群优化算法,并将其应用于神经网络的建模中。改进的粒子群优化算法改善了粒子群优化算法摆脱局部极小点的能力,对典型函数的测试和基于神经网络的软测量建模表明:改进算法的全局搜索能力有了显著提高,特别是对多峰函数能够有效地避免早熟收敛问题。

Particle swarm optimization is a simple stochastic global optimization technique. Its significant feature is simpler expression and less parameters, but it is easily slumped local minima. A particle swarm optimization algorithm improved by Alopex is brought forward. The proposed algorithm sustains diversity in population efficiently and improves the ability of breaking away from local minima. At last the improved algorithm is used to model the soft sensor based on artificial neural networks. The experiment results demonstrate that the proposed algorithm is superior to the original particle swarm optimization algorithm, especially multi-apices function.