A Grid Fault Diagnosis Method Based on Stacking Algorithm

Abstract

Power grid dispatching is developing towards digitalization and intellectualization. Using the data generated in power grid operations to diagnose faults through artificial intelligence technology directly has become the development trend of the power grid. However, traditional machine learning algorithms often need help to achieve good diagnostic results in practical applications. Therefore, a fusion model based on the Stacking integration algorithm is proposed, and four commonly used machine learning algorithms are selected for fusion. Furthermore, the model is trained using the fault alarm information features extracted from the SCADA system. Then, by comparing and analyzing the diagnostic effect of four single algorithms and the Stacking model for various fault types, the conclusion that the Stacking model has a better overall effect than other algorithms in the diagnosis of various fault types is obtained, and the advantage of the stacking fusion algorithm is proved.

A multi-source heterogeneous data-driven method for the site selection of elderly care facilities

“…More importantly, the generalization ability of the model in this paper has been improved to a certain extent due to the inherent characteristics of the stacking algorithm, in practical applications, this model will have greater value. [16] Random Forest algorithm is superior to the XGBoost algorithm in 𝐴𝑈𝐶, 𝑅𝑒𝑐𝑎𝑙𝑙 and 𝑃𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛 values, so it will be used to rank the importance of features. The output is shown in the table.…”

“…By combining different base classification algorithms, the stacking algorithm has better generalization ability and higher accuracy. [16] The algorithm in this paper is based on stacking algorithm, as shown in the figure. First, the training set is divided into two parts, which are used for learning and fitting the two base classifiers (Base-leaner) of XGBoost algorithm and the random forest algorithm.…”