Research On Key Technologies Of Coal-rock Cutting Pattern Recognition For Shearer

The unmanned or less-humanized fully mechanized coal mining working face is becoming extremely important for safety and high efficiency production.As one of key equipments in modern fully mechanized coal mining working face,the intelligent level of shearer directly affects the safty production and coaling mining efficiency of the entire coal mining working face.Furthermore,the identification of coal-rock cutting pattern is one of the necessary conditions to realize intelligent control of the shearer.Unfortunately,the cutting height and traction speed of the shearer cannot be adjusted adaptively according to coal-rock cutting pattern at present.In practical coal mining,frequent manual intervention is still unavoidable.Hence,it is necessary to research key technologies of the coal-rock cutting pattern recognition to improve the intelligent level of the shearer controlment.In this paper,the cutting sound signal is regarded as the souce signal,the quick and accuate recognition of coal-rock cutting pattern is treated as the target.Then methods and technologies focus on time-frequency characteristic analysis on the coal-rock cutting sound signa under strong background noise,adaptive enhancement and denoising,cutting pattern indentification under unsupervised condition and intelligent controlment are researched deeply.The main work in this paper can be summarized as follow:(1)On the basis of basic structure and working process of shearer,mathematical model of height and speed adjustment is establish by combining the functional requirements of the coal-rock cutting pattern recognition system.Overall framework of the coal-rock cutting pattern recognition system is constructed;main components and recognition flowof the system are analyzed.(2)Generation mechanism of the coal-rock cutting sound and time-frequnecy characteristic of the signal in different crush stage are analyzed;main sound source in fully-mechanized coal mining working face and the coupling relationship between different souces are studied;time-frequency distribution feature of sound signal in different cutting pattern and different traction speed is researched and key frequency band which could describe different cutting pattern is determined.(3)An adaptive enhancement algorithm for weak signal among strong background noise based on bistable stochastic resonance model is designed to enhance the cutting sound signal;the main source and type of noise in the cutting sound signal is analyzed and threshold denoising method based on improved fruit fly optimization algorithm is designed to improve the signal-to-noise ratio of the signal;an improved EEMD,where endpoint effect and pseudo IMF components are overcomed,is proposed to extracte feature vector from the cutting sound signal.(4)A division method on coal-rock cutting pattern is proposed and a combination of fuzzy C-means and modified bat search algorithm is proposed to realize mechine learning on accurate cutting pattern number and center under unsupervised condition.In this paper,coal-rock cutting pattern identification system of shearer is designed and researched,ground and underground experiment are conducted and the result indicates that: sound signal in different cutting pattern can be processed effectively,coal-rock cutting pattern can be identified accurately and the intelligent controlment can be realized according to the proposed system.