Abstract:Due to the limitation of magnetic induction intensity of the traditional magnetite separation method, the separation accuracy of large-grained ore is low. In this paper, the magnetic induction signal of magnetite with a grain size greater than 20mm is analyzed, and an improved one-dimensional convolutional neural network (1D-CNN) magnetite recognition classification model is established. First, the DC component of magnetic induction signal is removed. Then, the signal is used as an input to the 1D-CNN model, and the signal characteristics are extracted adaptively through the convolution layer and the pooling layer. Finally, the output layer uses Softmax logic regression to realize the classification of magnetite magnetic induction signals. The model is compared with the classic 1D-CNN, and the confusion matrix is used to evaluate the feature recognition effect of the model. The experimental results show that the recognition accuracy of the model has reached 94%, and the convergence speed has increased by about 28%.