Low-light Coal-rock Image Enhancement Based On Hybrid Algorithm And Human-simulated Intelligent Control Of Shearer Mining Height

Intelligent and unmanned mining is the future of intelligent coal mine development,and the control of shearer mining height system is the core link of intelligent coal mine construction.The situation of coal mining face is complex,and the existing traditional control strategy is not enough to cope with a variety of complex conditions such as continuous coal-rock boundary,coal seam mutation and fault zone.In this paper,a new humanoid intelligent control strategy based on fuzzy reasoning and particle swarm optimization is proposed for the control of shearer height system.In addition,the identification of coal-rock boundary is the premise to realize the control of mining height system,but the low-quality coal-rock image will greatly improve the difficulty of coal-rock identification.According to Retinex theory and HSI principle,this paper proposes an improved image enhancement model for U-Net network based on the advantages of attention mechanism and residual network to improve the quality of coal and rock images.The main work is as follows :(1)By comprehensively analyzing the performance of common image enhancement algorithms such as histogram equalization,a new model is proposed.Firstly,Retinex theory is used to solve the problem of complex parameters in the process of image enhancement.Then,according to the principle of HSI,the problem of image conversion between RGB format and HSI format is solved,and the brightness component(I component)in the image is taken as the enhancement object.Then,based on the U-Net network,the attention mechanism is introduced and the residual network structure is referred to construct a new U-Net network model.After the model is built,it is trained on the general data set.According to the quantitative evaluation index of the synthetic low-illumination image and the non-reference evaluation index of the real low-illumination image,compared with other existing image enhancement algorithms,the model proposed in this paper has superior performance and strong generalization ability.On the trained model,the synthetic low-illumination coal-rock image is used for testing,and good enhancement effect is also achieved.(2)Through the comprehensive analysis of the performance of traditional control methods such as classical PID control,these algorithms can not deal with the complex working conditions of shearer mining height system.A human-simulated intelligent control model based on fuzzy logic to improve modal switching and particle swarm optimization to optimize parameters is proposed.The model extends the characteristic mode of phase plane of human-simulated intelligent control error to fuzzy set,selects the optimal control mode in real time by fuzzy logic reasoning following error and error variation,and adjusts the control mode parameters in real time by particle swarm optimization following error and error variation.The human-simulated intelligent control model of the hybrid algorithm is established and compared with the other three control models.The step response is used as the input signal to compare the performance of the four models.At the same time,the simulation is also carried out to verify the performance under natural curve and fault zone conditions.By analyzing the simulation results,each dynamic index of the model in this paper has been greatly improved,and its stability and robustness are also better than those of other models,which can better achieve the control purpose.