CAJ | 학술논문

小波神经网络（WNN）结合了小波良好的时频局域化性质和传统神经网络自学习功能,因而使得小波网络具有较强的逼近和容错能力,并具有良好的收敛性和鲁棒性。然而其网络收敛速度慢、搜索成功率低及易陷入局部极小值等缺点使得传统的小波神经网络难以得到广泛应用。本文介绍一种基于粒子群（PSO）算法的小波神经网络,其通过利用种群间信息共享进行寻优,以获得结构化的神经网络,克服传统小波网络的诸多缺点,取得了良好的效果。
소파신경망락（WNN）결합료소파량호적시빈국역화성질화전통신경망락자학습공능,인이사득소파망락구유교강적핍근화용착능력,병구유량호적수렴성화로봉성。연이기망락수렴속도만、수색성공솔저급역함입국부겁소치등결점사득전통적소파신경망락난이득도엄범응용。본문개소일충기우입자군（PSO）산법적소파신경망락,기통과이용충군간신식공향진행심우,이획득결구화적신경망락,극복전통소파망락적제다결점,취득료량호적효과。
Wavelet Neural Network（WNN） combined with wavelet good time-frequency localization and self-learning function of the traditional neural network,thus wavelet network has strong approximation and fault tolerance,and have good convergence and robustness.Yet,which it has slow speed of network convergence and low success rate and it＇s also easy to search into the local minimum value makes it difficult to be widely used.This paper introduces a kind algorithm of wavelet neural network based on particle swarm optimization,it has achieved a good effect that has already overcome many of the faults of the traditional wavelet network and get structured neural network through the use of information sharing among the population on the global optimal solution.