摘要
依据模糊均值聚类故障检测原理,对舰船电气系统初始状态数据样本归一化后,构建舰船电气系统状态数据模糊相似矩阵,依据该矩阵分析系统状态数据亲疏度;利用亲疏度结果获取状态数据的最佳权值指数,并对状态数据样本进行加权处理,降低系统状态数据聚类稀疏度;将最佳权值指数赋予到状态数据中,同时更新模糊C-均值聚类目标函数。通过设置聚类类别数量、停止阈值和迭代次数后,求解更新模糊C-均值聚类目标函数并输出聚类隶属度和聚类中心,完成舰船电气系统初始状态数据聚类,得到舰船电气系统故障类别。实验结果表明,模糊C-均值聚类钻粉在舰船电气系统故障类型检测过程中具备较低的稀疏度,可有效检测舰船电气系统故障类型的同时其检测结果贴近度数值最高接近1.0。
Based on the fault detection principle of fuzzy C-means clustering, the initial state data samples of ship electrical system were normalized, and the fuzzy similarity matrix of ship electrical system state data was constructed, and the affinity degree of system state data was analyzed according to the matrix. The optimal weight index of the state data was obtained by using the results of affinity and affinity, and the state data samples were weighted to reduce the clustering sparsity of the system state data. The optimal weight index was assigned to the state data, and the fuzzy C-mean clustering objective function was updated. After setting the number of clustering categories, stop threshold and iteration times, the fuzzy C-mean clustering objective function was solved and updated, and the clustering membership degree and clustering center were output. The initial state data clustering of ship electrical system was completed, and the fault categories of ship electrical system were obtained. The experimental results show that the fuzzy C-means clustering drill powder has a low sparsity in the detection process of warship electrical system fault type, and can effectively detect the warship electrical system fault type, while the highest value of the detection result proximity is close to 1.0.
作者
林航
方宁
LIN Hang;FANG Ning(Qingdao Branch of China Classification Society,Qingdao 266034,China;Shandong Wide Area Technology Co.,Ltd.,Dongying 257000,China)
出处
《舰船科学技术》
北大核心
2022年第4期156-160,共5页
Ship Science and Technology
关键词
模糊聚类算法
舰船电气系统
故障远程检测
贴近度
故障类型
加权处理
fuzzy clustering algorithm
ship electrical system
remote fault detection
closeness
fault type
weighted processing
