混合粒子群优化算法优化前向神经网络结构和参数*

提出了综合利用粒子群优化算法（PSO）和离散粒子群优化算法（D-PSO）同时优化前向神经网络结构和参数的新方法。该算法使用离散粒子群优化算法优化神经网络连接结构,用多维空间中0或1取值的粒子来描述所有可能的神经网络连接,同时使用粒子群优化算法优化神经网络权值。将经过该算法训练的神经网络应用于故障诊断,能够有效消除冗余连接结构对网络诊断能力的影响。仿真试验的结果表明,相比遗传算法等其他算法,该算法能够有效改善神经网络结构和参数的优化效率,提高故障模式识别的准确率。

Optimizing structure and parameters of feedforward neural networks using hybrid particle swarm optimization

A new algorithm based on particle swarm optimization(PSO) and discrete particle swarm optimization (D PSO) was proposed to optimize the topology and connection weights of artificial neural network(ANN) simultaneously. The topology of neural network was optimized by D PSO and connection weights was optimized by PSO. The algorithm was successfully applied to fault diagnosis, it could eliminate some ill effects introduced by redundant structure of ANN. The experimental results show that the algorithm can obtain good diagnosis results. Compared with genetic algorithm, the algorithm can effectively improve the optimizing effectiveness for feedforward network structure and parameters.