Open AccessResearch on anomaly data mining method of new energy field stations based on improved Adaboost algorithm
Author(s) -
Nan Wang,
Yanzhuo Wang,
Yan Cheng,
Ti Guan,
Qiang Ma,
Shumin Sun,
Yifei Guan,
Yuejiao Wang,
Shibo Wang
Publication year - 2021
Publication title -
iop conference series. earth and environmental science
Language(s) - English
Resource type - Journals
eISSN - 1755-1307
pISSN - 1755-1315
DOI - 10.1088/1755-1315/680/1/012017
Subject(s) - adaboost , data mining , computer science , field (mathematics) , anomaly detection , algorithm , anomaly (physics) , energy (signal processing) , artificial neural network , pattern recognition (psychology) , artificial intelligence , mathematics , support vector machine , statistics , physics , pure mathematics , condensed matter physics
Traditional anomalous data mining methods require a lot of prior knowledge, which leads to low data mining integrity and efficiency. For this reason, a new energy field abnormal data mining method based on improved Adaboost algorithm is proposed. After pre-processing the new energy field data, the algorithm is improved by introducing dynamic weight parameters to address the shortcomings of the Adaboost algorithm. After calculating the degree of data anomaly using the direct push belief machine, the neural network is used to reduce the error value of the Adaboost algorithm, and finally the output of the Adaboost algorithm is used to realize abnormal data mining. The simulation experiment proves that the researched abnormal data mining method has high data integrity and high efficiency.