基于正态分布模型的无线机房用电异常预测方法

Prediction Method of Power Consumption Abnormity in Wireless Computer Room Based on Normal Distribution Model

无线机房的环境特殊性使其用电异常预测操作困难,因此研究基于正态分布模型的无线机房用电异常预测方法。使用正态分布模型提取用电异常因素,发现多与节假日和环境因素相关。处理数据并使用拉依达准则对异常样本进行异常定位与修正,利用组合方法,将数据预处理和数据预测结合起来完成用电异常预测。经实验验证,正态分布模型的用电异常预测方法有92.56%的准确率,具有实用性。

Because of the special environment of the wireless computer room,it is difficult to predict the power consumption abnormity of the wireless computer room.The normal distribution model is used to extract the abnormal factors of electricity consumption,and it is found that most of them are related to holidays and environmental factors.The data are processed and the abnormal samples are located and corrected by laida criterion.Using the combination method,the data pre-processing and the data forecast are combined to complete the electricity consumption abnormal forecast.The experimental results show that the power consumption anomaly prediction method based on the normal distribution model has an accuracy of 92.56%and is practical.