多尺度特征的工业物联网隐性异常检测方法

A Hidden Anomaly Detection Method for Industrial Internet of Things Based on Multiscale Features

为了提高工业物联网隐性异常的检测效果,设计了基于多尺度特征的工业物联网隐性异常检测方法。根据工业物联网自身的特性,设定多个数据特征选择标准,并计算最终的特征选择结果。在此基础上,提取出多尺度特征,并计算不同特征的权重值,由此计算特征的重构误差和分类结果,完成对工业物联网数据分类器的设计,并通过计算不同尺度下的检测结果和权重值来实现对工业物联网隐性异常的检测。实验测试表明,所设计方法在实际应用中漏检率较低,检测效果较好。

In order to improve the detection effect of hidden anomalies in the industrial internet of things(IoT),a multi-scale feature based detection method for hidden anomalies in the industrial IoT was designed.Based on the characteristics of the industrial IoT,multiple data feature selection criteria are established and the final feature selection results are calculated.On this basis,multi-scale features are extracted and the weight values of different features are calculated.Furthermore,the reconstruction error and classification results of the features are calculated,and the design of an industrial IoT data classifier is completed.The detection of hidden anomalies in the industrial IoT is achieved by calculating the detection results and weight values at different scales.Experimental results have shown that the designed method has a low missed detection rate and good detection performance in practical applications.