Research On Coal Gangue Image Recognition Method Based On CNN-ELM Hybrid Model

Coal gangue separation is an effective way to improve coal quality.The traditional separation method has complex process flow,low separation efficiency and easy to pollute the environment.It is no longer suitable for the rapid development of industrial production.Exploring more advantageous sorting methods has become the focus of research.In view of this,the deep learning technology is applied to the separation of coal gangue,and the coal gangue image recognition model of CNN-ELM is designed.Firstly,the coal gangue image acquisition device is built.The device takes S3C2410(ARM9)as the core controller.The hardware platform includes USB camera,LCD display,memory and necessary communication interfaces.Then,write the driver,transplant it to the Linux operating system,and finally realize the functions of image acquisition,display and storage.A series of preprocessing is carried out on the collected images to produce similar but different training samples,and the coal gangue image data set is constructed.As a classical classifier,the full connection layer of convolutional neural network is trained according to the traditional gradient descent method,and its generalization ability is limited.To solve this problem,a hybrid model combining convolutional neural network and limit learning machine is proposed to be applied to the field of coal gangue image classification.Convolutional neural network is used to extract features from the input image.The feature map will eventually be encoded into a one-dimensional vector and sent to the limit learning machine for classification.The detailed design of CNN-ELM model is given,including parameter design,structural analysis and the derivation of backpropagation algorithm in the iterative process.The cnn-elm model has input layer,5convolution layers,3 pooling layers,2 standardization layers and elm classification layer.The activation function used is Relu.CNN-ELM model is used to recognize coal gangue image.At the same time,three groups of comparative experiments are designed to analyze the performance and advantages of CNN-ELM model in terms of accuracy and training time.The upper computer interface of coal gangue sorting system is designed,including login interface,main interface,setting interface and operation interface.The upper computer receives the collected coal gangue image,and calls the algorithm cnn-elm through the interface operation to identify and sort the coal gangue.Figure[50] Table [9] Reference[84]...