[电力机车论文] 摘 要：针对BP算法存在的缺陷，如训练速度慢，易收敛于局部极小点及全局搜索能力弱等，利用遗传算法能够进行全局最优化搜索这一特点，在改进的自适应遗传算法的基础上，提出了一种新的用于BP网络训练的混合算法，即改进的自适应遗传算法与BP算法相结合的混合训练方法。将所提出的混合训练方法应用于神经网络式距离保护中，利用ATP仿真计算的结果进行训练及检验,结果表明：所提出的算法与单一的BP算法相比，不仅可避免陷入局部极小点，而且提高了网络的训练速度，同时满足对距离保护的速度和精度的要求。关键词：遗传算法；
BP算法； 距离保护
A HYBRID LEARNING ALGORITHM FOR
NEURAL NETWORK BASED DISTANCE PROTECTION
Abstract：For
avoiding the shortcomings existing in BP algorithm, such as being slow in
training speed, convergence to the local minimum, and weakness in global search,
a new hybrid algorithm based on modified adaptive genetic algorithm for BP
network is presented. Then the hybrid algorithm is applied to neural network
based distance protection. The training and testing results by ATP simulation
show that the hybrid algorithm can not only avoid convergence to the local
minimum, but also improve the training speed for the neural network, as compared
with the simple BP algorithm. At the same time, it satisfies the demands of
speed and accura……
<<<<<全文未完，本文约5471个中文字，未计算英文字母、数字>>>>>
|
|
