A Modified Decision Tree and its Application to Assess Variable Importance

Bbosa, Francis Fuller; Wesonga, Ronald; Nabende, Peter; Nabukenya, Josephine

doi:10.1145/3478905.3479245

Cited by 1 publication

(1 citation statement)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The decision tree algorithm can recursively select the optimal features of the sample, and segment the training data according to the feature, so that each subset of the data has a best classification result [44,45]. Moreover, its calculation is relatively simple, and no parameter assumptions are required.…”

Section: Construction Of the Secondary Data Sets And Fusion Modelsmentioning

confidence: 99%

Single-Phase Grounding Fault Types Identification Based on Multi-Feature Transformation and Fusion

Min

Xia

Meng

et al. 2022

Sensors

View full text Add to dashboard Cite

The frequent occurrence of single-phase grounding faults affects the reliable operation of power systems. When a single-phase grounding fault occurs, it is difficult to accurately identify the fault type because of the weak characterization and subtle distinction between different fault types. Therefore, this paper proposes a single-phase grounding fault type identification method based on the multi-feature transformation and fusion. Firstly, the Hilbert–Huang transform (HHT) was used to preprocess the fault recorded wave data to highlight the characteristics between different fault types. Secondly, the deep learning model ResNet18 and the long short-term memory (LSTM) are designed to extract the complex abstract features and time-series correlation features from the preprocessed data set separately. Finally, it designs a fusion model to combine the advantages of heterogeneous models to identify the type of single-phase grounding fault. Experiments validate that the method is good at fully mining the characteristics of the fault types contained in the fault recorded wave data, so it can identify multiple types of faults with strong robustness and provide a reliable basis for the subsequent formulation of targeted fault-handling measures.

show abstract

Section: Construction Of the Secondary Data Sets And Fusion Modelsmentioning

confidence: 99%