Multi-objective Non-intrusive Load Monitoring Based on Improved Unet
Non-intrusive load monitoring is considered to be a key issue in energy monitoring and manage-ment.Aiming at the problem that the traditional NILM model has low accuracy in training multi-applicances can only train a single appliance.A multi-objective non-intrusive load monitoring model based on improved Unet was proposed,Firstly,a convolutional block attention module on the basis of Unet(CBAM)was intro-duced,CBAM combined channel and spatial attention to enhance the feature extraction ability of the model.Then the activation function ReLU was replaced by GELU by combining the residual connection to prevent network degradation and gradient explosion.Meanwhile the multi-channel output was realized by utilizing the feature extraction ability of different channels of the neural network,and the five appliances that were used the most on the UK-DALE dataset were trained at the same time.Finally,the experimental results showed that the model in this paper could realize multi-objective monitoring and had high accuracy with low-er number of network parameters compared to the other existing NILM models.
NETL
NSTL
万方数据
