Modeling and monitoring method of cutter abrasion loss on basis of residual error convolutional neural network

The invention provides a modeling and monitoring method of cutter abrasion loss on the basis of a residual error convolutional neural network. A residual error convolutional network is used as a self-adaptive model, adjustment of parameters of the model is performed on this basis, and finally a model of the cutter abrasion loss is established. The model adopts time-domain signals of different sensors as input, the cutter abrasion loss is used as output, and meanwhile in consideration of uncertainty in a machining process, the signal features in the machining process are extracted through the convolutional neural network theory; then the signal features of the different dimensions are used as input of full-joint neural network, and the influence weight of the feature components on the cutter abrasion loss is obtained automatically through an adam algorithm; and finally a model from signals to the abrasion loss is established according to a supervised learning method, and the predictionof the cutter abrasion lo