基于卷积神经网络的调度交换网缺陷智能研判

Intelligent Diagnosis of Defects in Dispatching Switching Networks Based on Convolutional Neural Network

为达到调度交换网在电网中的高标准要求,提出了调度交换网缺陷智能研判技术,设计了调度交换网缺陷智能研判系统。该系统包括硬件部分和软件部分。硬件部分利用集中控制方法,将系统分为公共控制部分和电话控制部分,实现研判系统与每个调度交换网节点之间的通信。软件部分使用8031汇编程序和VC++进行系统间的操作,实现对调度交换网缺陷智能研判系统自身的实时监控。利用卷积神经网络(CNN)算法对调度交换网故障进行判断,以更准确地判断调度交换网的故障类型。为了验证该系统的可行性,进行了相关试验。试验结果表明:在对系统研判的可靠性进行测试时,该系统的可靠性高达96%;对判断故障类型进行准确度测试时,该系统的准确度高达98%。

To meet the high standard requirements of the dispatching switching network in the power grid,the dispatching switching network defect intelligent research and judgement technology is proposed,and the dispatching switching network defect intelligent research and judgement system is designed.The system includes a hardware part and a software part.The hardware part uses a centralized control method to divide the system into a public control part and a telephone control part to realize the communication between the research and judgement system and each dispatching exchange network node.The software part uses 8031 assembly program and VC++to carry out inter-system operations and realize real-time monitoring of the dispatching exchange network defect intelligent research and judgement system itself.Convolutional neural network(CNN)algorithm is used to judge the dispatching switching network faults,so as to achieve a more accurate judgement of the dispatching switching network with a more accurate judgement of the type of faults.In order to verify the feasibility of the system,trial experiments are conducted.The results of the experiments show that the reliability of the system is 96%when tested for reliability and 98%when tested for accuracy in determining fault types.