基于连续性遗传算法的接地网腐蚀诊断优化

Optimization of Corrosion Diagnosis of Grounding Grid Based on Continuous Genetic Algorithm

为了对接地网的导体腐蚀情况进行准确判断,建立了接地网腐蚀诊断的数学模型,基于连续性遗传算法编写了接地网腐蚀诊断程序,构建了适应度评价函数F,并利用多层减维技术对算法进行了优化。以重庆某110 k V实际接地网(41个节点、61条支路)为例仿真对比了优化前后的诊断效果。结果表明:优化前,诊断程序对真实故障支路的定位困难,出现漏诊、误诊的支路较多;利用多层减维技术进行优化后,不仅解决接地网腐蚀故障诊断中支路变量维数多、算法后期收敛速度慢的问题,使诊断程序能够准确地定位腐蚀支路,而且对腐蚀支路腐蚀程度的判断有相当高的精度,最大的诊断偏差也不超过10%,程序的单次运行时间约为15 min。对实际工程来说,程序的诊断效果和运行时间均满足要求。

In order to locate corrosion in grounding girds accurately, we established a mathematical model of the corrosion diagnosis and a diagnosis program based on the continuous genetic algorithm, as well as a fitness function, F. Specifically, the algorithm was optimized through multi-dimensional reduction. Taking a real grounding gird with 60 branches, 36 nodes in a 110 k V substationin（Chongqing, China） for an example, we tested and compared the effects on corrosion diagnose with/without the algorithm optimization. The results showed that, before being optimized, the program had trouble in locating fault branches while having relatively large number of misdiagnosis and omissions. However, the optimization of multi-dimensional reduction successfully solved the problems of excessive variable dimension number and slow convergence rate in the late iterative procedure, the program was capable of locating fault branches with high accuracy, namely no more than 10%, and high speed（a single run of the program takes 15 min approximately）. This indicates that the proposed program can meet the requirements of effect and speed of engineering applications.