Comparative Study on Prediction Algorithms for Power Grid System Access Failure Times

• Published in: IOP conference series. Earth and environmental science Vol. 252; no. 3; pp. 32183 - 32193
• Main Authors: Yan, Yi, Li, Bo, Xiao, Jianyi, Liang, Yunde, Shang, Yanwei, Zhou, Kaidong
• Format: Journal Article
• Language: English
• Published: Bristol IOP Publishing 09.07.2019
• Subjects: Algorithms; Comparative studies; Failure; Failure times; Mean square errors; Mean square values; Prediction models; Root-mean-square errors; Time series; Winter
• ISSN: 1755-1307; 1755-1315; 1755-1315
• DOI: 10.1088/1755-1315/252/3/032183

As functions of the power grid system gradually increase and its scale becomes larger, the frequency of access failure times will gradually increase. In order to prevent access failures and make countermeasures in advance, we perform model fitting analysis and fault risk prediction for the time series of power grid failure code. In this paper, we use the failure code 404 as the training data and use the SARIMA algorithm, Fbprophet algorithm, holt-winter algorithm, and GM algorithm respectively to construct a time series prediction model. According to the result of the model building and the calculating, we find that the root mean square error of SARIMA algorithm is 258.85, which is the lowest among these algorithms, and the root mean square errors of Prophet and holt-winter algorithm are 749.288 and 809.89, respectively. However, the root mean square error of GM algorithm reaches 1710.95, which is 6 times as many as the SARIMA algorithm. In conclusion, with algorithm analysis and the comparisons of these four algorithms, we recommend the SARIMA algorithm as a predictive model for the power grid system.