Research on Dynamic Rule Mining Method for Incremental Small Sample Experimental Data
Weiguang Bai, Jie Li Sun, Yun Yang · 2023
In terms of fault diagnosis and prediction, new equipment mainly has the characteristics of small data samples and increasing sample size. In the rulemaking process, there are problems such as incomplete data, insufficient rules support, low accuracy data and information completeness, the current method is difficult to automatically correct forward rules based on continuously accumulated data. There is a lack of basic models and algorithms for rule correction, This method establishes an initial rule library after modeling system faults based on knowledge and data fusion. When a system fault occurs, it will exhibit various abnormal phenomena, and the collection of different types of abnormal phenomena (i.e. fault phenomena) corresponds to typical faults.