Research on Dynamic Detection of Peripheral Thermostatic Network Intrusion in Box-Type Substation
Substation peripheral thermostatic network nodes are weakly associated and vulnerable to external intrusion.For this reason,a dynamic detection method of peripheral thermostatic network intrusion in box-type substation based on deep belief network(DBN)is proposed.Firstly,type conversion and normalization are implemented on the thermostatic network data,and the Nominal type is converted to Numeric type to avoid the converted data dimension from increasing.Then,non-linear iterative partial least squares method is utilized to implement feature extraction and classification processing on thermostatic network data to enhance the clustering of intrusion signals.Finally,combining the DBN model and utilizing the kernel extreme learning machine(KELM)classification algorithm,the intrusion data of the thermostatic network is implemented for detection.The experimental results show that the model training accuracy of the method is up to 90%;the training time is 10.35 s when the number of iterations reaches 200;the network intrusion detection efficiency is high and the error score of the detection results is small,with an average error score of 1%.The method has a stable network intrusion detection rate for each sample,which is conducive to the practical application of network intrusion detection.
Box-type substationsThermostatic networkNetwork intrusionDeep belief network(DBN)Data type conversionData dimension
NETL
NSTL
万方数据
