尝试将重置算法应用于神经网络结构的优化,研究了重置算法中最佳重置时间的性质,并提出一种基于重置的L-M(Levernberg-Marquardt)变结构BP神经网络.对比实验表明,重置算法的引入有效地解决了神经网络结构优化的问题,优化后的网络具有较好的收敛性. This paper attempts to apply the reset algorithm to the optimization of the neural network structure, studies the nature of the optimal reset time in the reset algorithm, and proposes a reconfigurable LM (Levernberg-Marquardt) variable structure BP neural network. The introduction of reset algorithm effectively solves the problem of neural network structure optimization, and the optimized network has better convergence.