Gas leakage detection technology based on improved DRSN-CW
In order to solve the problem of environmental gas leakage detection,this paper proposes an improved method based on channel threshold deep residual network (DRSN-CW) to improve the detection accuracy of the GPLA-12 dataset. In this method,all convolutional layers are designed with larger and unequal convolutional kernel sizes to expand the receptive field during the process of extracting fault features. In addition,considering that environmental gas leakage datasets typically contain a large amount of environmental noise,the combination of DRSN-CW's soft threshold module and the designed kernel size reduces the impact of noise on the accuracy of environmental gas leakage detection. The experimental results show that the leakage detection technology based on improved DRSN-CW gas is superior to the commonly used technologies at present.
万方数据
维普
