中文摘要：

神经网络作为机器智能分支中一种优秀的分类算法，在图像分类、人脸识别等领域中都有非常广泛的应用。但由于其参数过多，所以容易陷入局部最优解。针对BP神经网络易陷入局部最优的问题，提出了一种粒子群算法和聚类算法结合的优化神经网络权值的方法。该方法通过把神经网络的权值作为粒子群算法的初始粒子并利用粒子群算法的随机性全局搜索神经网络的待选初始权值，然后利用C均值算法找出包含权值较多的那一类，并把其聚类中心作为BP神经网络的初始权值。仿真结果表明，利用这种新的融合算法在防止BP神经网络易陷入局部最优的问题上能比普通的粒子群算法更加优秀。

英文摘要：

Neural network is a kind of excellent classification algorithm in the branch of machine intelligence, which has a wide range of applications in the field of image classification, face recognition and so on. However, because of its excessive parameters, it is easy to fall into the local optimal solution. According to this problem, a method combining particle swarm algorithm and clustering algorithm to opti- mize the weights of neural networks is proposed, which takes the neural network weights as the initial particle of particle swarm algorithm and uses the random of particle swarm algorithm to search the initial weights of neural network. Then the class contains more weight is found using C-means algorithm and its clustering center is regarded as the initial weights of BP neural network. The simulation results show that it is more excellent than the conventional particle swarm optimization algorithm in preventing the BP neural network from fall- ing into local optimum.