CAJ | 학술논문

传统的蚁群算法在迭代过程中产生逆转变异，新的结点与链路也可能在任意时刻加入到云中，给电网系统云数据的云计算和故障数据预测检测带来很大难度，出现拥塞控制，导致聚类效果不好。结合云计算处理数据的特点，对传统的蚁群算法进行改进，提出一种改进的蚁群引导电网系统云数据聚类和故障检测算法，根据基因位随机数大小决定输出概率的精度，更新状态类别充分统计量，得到故障特征观测概率和初始概率，执行聚类中心更新规则。搭建的Hadoop集群云计算原型系统，在开源的云计算平台框架和HBase电网系统数据库下进行数据采集和算法实现。仿真结果表明，算法在数据聚类和故障检测中具有较好的应用性能。
전통적의군산법재질대과정중산생역전변이，신적결점여련로야가능재임의시각가입도운중，급전망계통운수거적운계산화고장수거예측검측대래흔대난도，출현옹새공제，도치취류효과불호。결합운계산처리수거적특점，대전통적의군산법진행개진，제출일충개진적의군인도전망계통운수거취류화고장검측산법，근거기인위수궤수대소결정수출개솔적정도，경신상태유별충분통계량，득도고장특정관측개솔화초시개솔，집행취류중심경신규칙。탑건적Hadoop집군운계산원형계통，재개원적운계산평태광가화HBase전망계통수거고하진행수거채집화산법실현。방진결과표명，산법재수거취류화고장검측중구유교호적응용성능。
Traditional ant colony algorithm produced reverse mutation in the iterative process, node and link would be add-ed to the cloud at any time, it brought great difficulty to cloud computing and fault data grid system cloud data prediction, and it had congestion control, leading to the clustering effect was not good. Processing data characteristics of cloud was used, improved ant colony algorithm was proposed, and an improved ant colony system guided grid cloud data clustering and fault detection algorithm was proposed, the gene bit random number determined the output probability of accuracy, up-dated status categories sufficient statistic, fault feature observation probability and initial probability were calculated. Clus-tering center update rule was obtained. Hadoop cluster cloud computing prototype system was built, computing platform framework and HBase database were established for algorithm realization. The simulation results show that, the algorithm has the better application performance in data clustering and fault detection.