由于绿色装配式建筑受环境变化、设备故障等因素影响,使得能耗数据存在较多噪声,无法准确定位异常数据,导致对其修正难度较大.为了有效解决能耗数据存在异常和缺失等问题,提出一种绿色装配式建筑能耗异常数据快速修正方法.结合小波阈值去噪方法和卡尔曼滤波方法,去噪处理绿色装配式建筑能耗数据.利用基于密度的空间聚类算法(Spatial-Augmented Density-Based Spatial Clustering of Applications with Noise,SA-DBSCAN)算法,检测绿色装配式建筑能耗异常数据,引入反向传播(Backpropagation,BP)神经网络对其修正.经过大量仿真分析表明,所提方法可以准确检测绿色装配式建筑能耗异常数据,且修正精度较高,误报率仅为 0.15%、检测率高达98.09%,修正仅耗时12.8ms,可以为绿色装配式建筑能耗数据的可靠性和完整性提供有效支持.
Reliable Correction Method of Abnormal Energy Consumption Data of Green Prefabricated Buildings
At present,green prefabricated buildings are affected by environmental factors,so there is more noise in energy consumption data,and abnormal data cannot be accurately located.In order to effectively solve the problems of abnormal and missing energy consumption data,a fast correction method for abnormal energy consumption data of green prefabricated buildings was proposed.Firstly,the wavelet threshold denoising method was combined with the Kalman filter method to denoise the energy consumption data of green prefabricated buildings.Secondly,the Spatial-Augmented Density-Based Spatial Clustering of Applications with Noise(SA-DBSCAN)algorithm was used to detect the abnormal energy consumption data of green prefabricated buildings.Finally,a Backpropagation(BP)neural net-work was introduced to correct the data.Simulation analyses show that the proposed method can accurately detect the abnormal energy consumption data of green prefabricated buildings,and the correction accuracy is high.In addition,the false positive rate is only 0.15%,the detection rate is up to 98.09%,and the correction time is only 12.8ms,which can provide effective support for the reliability and integrity of the energy consumption data of green prefabrica-ted buildings.
Green prefabricated buildingEnergy consumptionAbnormal dataQuick correction
NETL
NSTL
万方数据
