*The invention discloses an elevator equipment safety early warning method based on limited non-negative matrix factorization multi-view clustering. The method comprises the following steps that an elevator equipment multi-view data set Xv belongs to Rd (v) * N and a label constraint matrix A belongs to Rn * (n + c + 1) are input; semi-supervised clustering is achieved on the basis of multi-view clustering in combination with limited non-negative matrix factorization, meanwhile, low-rank representation is introduced to solve the noise problem, and finally a target function of the elevator equipment safety early warning method based on limited non-negative matrix factorization multi-view clustering is constructed; solving the obtained objective function by using an alternating iterative optimization strategy; after a consistent constraint representation matrix S * is obtained, spectral clustering is carried out by using a k-means clustering algorithm to obtain a clustering result and a clustering result visualization graph; and after the clustering result visualization graph is obtained, the safety levels of different clusters in the clustering result visualization graph are determined according to the safety levels of part of the marked elevator equipment. The problem of noise is solved by fusing information of data from different sources and introducing low-rank representation.
本发明公开了一种基于受限非负矩阵分解多视图聚类的电梯设备安全预警方法,包括以下步骤:输入电梯设备多视图数据集Xv∈Rd(v)×N,和标签约束矩阵A∈Rn×(n+c+l);基于多视图聚类并结合受限非负矩阵分解实现半监督聚类,同时引入低秩表示解决噪声问题,最终构造一种基于受限非负矩阵分解多视图聚类的电梯设备安全预警方法的目标函数;使用交替迭代优化策略对得到的目标函数进行求解;获得一致约束表示矩阵S*后,利用k‑means聚类算法进行谱聚类得到聚类结果和聚类结果可视化图;得到聚类结果可视化图后,通过部分被标记的电梯设备的安全等级从而确定聚类结果可视化图中不同簇的安全等级。通过融合不同来源的数据的信息,并且引入低秩表示解决噪声的问题。
Elevator equipment safety early warning method based on limited non-negative matrix factorization multi-view clustering
一种基于受限非负矩阵分解多视图聚类的电梯设备安全预警方法
2023-01-03
Patent
Elektronische Ressource
Chinesisch
Elevator detection early warning safety control system
Europäisches Patentamt | 2022
|Elevator safety early warning monitoring system and judgment method thereof
Europäisches Patentamt | 2021
|