微型电脑应用

为了降低智能变电站改扩建工程实施过程中母线保护变电站配置描述(SCD)文件修改的错误率，提出一种基于卷积神经网络的智能变电站改扩建工程母线保护SCD文件智能校核技术。对母线保护与不同类型间隔设备的虚端子交互进行分析。介绍基于卷积神经网络的SCD文件智能校核技术实现过程，并给出不同类型间隔不同施工方式情况下训练得到的9类神经网络结构参数。以主变扩建SCD校核神经网络为例，对训练得到的网络参数进行分析，验证该方法给出的神经网络结构参数及学习参数的优越性及合理性。

In order to reduce the error rate of bus protection substation configuration description(SCD) file modification during the implementation of intelligent substation reconstruction and expansion project, an intelligent verification technology of bus protection SCD file in intelligent substation reconstruction and expansion project based on convolutional neural network is proposed. The interaction between bus protection and virtual terminals of different types of interval devices are analyzed. It focuses on the realization process of SCD file intelligent verification technology based on convolutional neural network, and gives 9 kinds of neural network structure parameters trained under different types of intervals and different construction methods. Taken the main transformer extension SCD check neural network as an example, the trained network parameters are analyzed to verify the superiority and rationality of the neural network structure parameters and learning parameters given by this method.