一种新的遗传算法及其在变压器故障诊断中的应用

来源期刊:中南大学学报(自然科学版)2005年第3期

论文作者:邓宏贵 曹建 罗安

文章页码:481 - 485

关键词:遗传算法;变压器;故障诊断

Key words:genetic algorithm; transformer; fault diagnosis

摘    要:提出了一种新的遗传算法,其基本思想是:以网络权重和偏差的实数形式作为基因构成染色体向量,采用基因多点交叉和动态变异进行种群最优选择。研究结果表明,这种新的遗传算法是一种随机优化算法,克服了梯度下降法的不足,能够自动调节网络参数、网络的连接权重和偏差。在此基础上设计出一种基于遗传算法和溶解气体分析的变压器故障在线诊断系统。该系统只要将传感器测出的变压器中线圈电流、特征气体的含量作为输入参数,就能对信息进行融合分析,得到输入变量(线圈电流、溶解气体含量)与输出结果(故障类型、程度和部位)的复杂对应关系;能有效地减少输入层神经元的个数,改进网络内部结构,提高神经网络模型的学习效率和诊断的准确率,诊断精度高,漏报少,无误报现象。

Abstract: A novel genetic algorithm is presented. The algorithm is intended to select the individuals with a smaller unfitness value in virtue of multi-point crossover and mutation in which gene vectors are composed of connection weights and bias terms of the neural networks. The results show that the system can ascertain complex corresponding relationship between input parameters (winding current, dissolved gas content) and outcom e(fault type, severity, position) in virtue of windings current of power transformer, dissolved gas content from sensor. The random optimized algorithm overcomes the deficiency of grads descend algorithm and can automatically tune the network parameters, connection weights and bias terms of the neural networks, an online diagnosis system is set up based on the genetic algorithm and dissovled gas analysis (DGA). The algorithm can decrease the number of the network input nerve cells effectively and ameliorate network inner structural and improve the study efficiency and veracity. So the system brings about accurate diagnosis, less leaked diagnosis, and diagnosis without mistake.

基金信息:国家计委自动化高新技术专项基金资助项目
国家自然科学基金资助项目

有色金属在线官网  |   会议  |   在线投稿  |   购买纸书  |   科技图书馆

中南大学出版社 技术支持 版权声明   电话:0731-88830515 88830516   传真:0731-88710482   Email:administrator@cnnmol.com

互联网出版许可证:(署)网出证(京)字第342号   京ICP备17050991号-6      京公网安备11010802042557号