Research On Cooperative Control Decision-Making Method Of Various Devices On Intelligent Fully Mechanized Mining Face

The intellectualization of coal mine is core technological support for high-quality development of coal industry and is Inevitable direction for technological revolution and upgrading of coal industry.The intellectualization of comprehensive mechanized coal mining(referred to as comprehensive mining)is one of the key technologies in intellectualization of coal mine,and its realization will greatly promote the development of intelligent coal mine improving the level of safe,efficient and green mining powerfully.The equipment for comprehensive mining mainly includes hydraulic support,coal shearer,scraper conveyor,transfer machine,crusher and equipment train,etc.,and hydraulic support,Shearer,and scraper conveyor in them can support each other,work divided and operate synergetic,which are responsible for the task of support,destroying and transporting coal on comprehensive mining face supporting each other,their respective division of labor,collaborative work.Therefore,realizing cooperative control to "three machines" during comprehensive mining is the key to realize intellectualization on comprehensive mechanized coal mining face.This dissertation,which relies on sub-project "study on control mechanism of multi-equipment cooperative operation in large mining height working face"(Grant No.2017YFC0804310)of National Key Research and Development Program "safety technology and equipment research and development of intelligent coal mining",takes the three fully mechanized coal mining machines as the research object and the cooperative control decision-making method of the three fully mechanized coal mining machines as the research main line to solve the problems of state data preprocessing,cooperative control strategy learning and intelligent decision-making in the three fully mechanized coal mining machines,verifying the cooperative control decision-making model of the three fully mechanized coal mining machines with laboratory and field data.The specific contents are as follows:(1)In view of lacking decision-making model of "three machines" collaborative control on intelligent fully mechanized coal mining face,through analyzing cooperative operation relationship and manual collaborative control process,determining the basic idea of collaborative control,and establishing the model and framework of collaborative control system for the three fully mechanized coal mining machines,the key technologies and solutions of the decision-making model are given,which lays the foundation for subsequent research.(2)In view of noise pollution and redundancy of characteristic parameters in the status data in "three machines" in fully mechanized coal mining,a data cleaning method based on dynamic fusion of local abnormal factors and stacking denoising self-encoder(DFLOF-SDAE)is designed to improve the data quality.Based on the rough finite state machine model of fully mechanized "three machines" cooperative operation process,the original decision data table of multi-source data association for fully mechanized "three machines" cooperative control is established.The multi-label adaptive fuzzy neighborhood rough set and its characteristic parameter selection algorithm are proposed to select characteristic parameters for the decision data table of fully mechanized mining "three machines" with multi-label mixed data type.Finally,the accuracy and reliability of this method are verified by comparative experiments.(3)In view of the problem of learning decision-making strategy of cooperative control of fully mechanized coal mining "three machines" with nonlinear,strong coupling and spatio-temporal correlation data,firstly,a new fuzzy rough width neural network model based on width architecture is proposed,which can extract the hidden features of strongly coupled nonlinear data.Then,a decision strategy learning method for cooperative control of fully mechanized coal mining "three machines" based on fuzzy rough width neural network is constructed to obtain the time sequence feature information of input parameters and the spatial association feature information among multiple input parameters realizing the learning of cooperative control decision strategy of fully mechanized coal mining "three machines".The verification of laboratory data of fully mechanized coal mining "three machines" shows that compared with ELM,FRNN,BLS and LSTM methods,SL-FRBNN method has higher accuracy and reliability in the behavior decision of fully mechanized coal mining "three machines",the height adjustment of shearer drum,and the coordinated speed regulation of scraper conveyor and shearer.(4)In view of the decision-making problem of "three machines" cooperative control in fully mechanized coal mining under dynamic unstable working conditions,a decision-making model of "three machines" cooperative control in fully mechanized coal mining based on incremental selective integration(ISEDFM-FRBNN&KELM)is proposed.The model is based on the idea of integrated decision-making.Firstly,Bootstrap resampling method is used to generate multiple training sets.Then,several binary mixed different base decision-maker models OIL-KELM&FRBNN with self-learning ability are constructed.Finally,the base decision-makers are integrated by the selective integration method of double error measures,to realize the online decision-making of "three machines" in fully mechanized mining.The method of ISEDFM is verified by laboratory data,and the results show that the proposed ISEDFM-FRBNN&KELM method can effectively recognize the behavior pattern of "three machines" in fully mechanized coal mining,predict the height of shearer drum and accurately predict the traction speed of shearer under the load change of scraper conveyor.(5)To further verify the rationality and effectiveness of the decision-making model of"three-machine" cooperative control in fully mechanized coal mining,the algorithm is verified by the operation data of "three-machine" in 43101 fully mechanized coal mining face of Yujialiang Coal Mine.The results show that the decision-making method of"three-machine" cooperative control in fully mechanized coal mining is correct and feasible,and the cooperative and stable operation of "three-machine" is realized by applying the proposed characteristic parameter selection algorithm,cooperative control decision strategy learning algorithm and dynamic selective integrated decision-making method.The decision-making method of multi-equipment cooperative control in intelligent fully mechanized coal mining face proposed in this dissertation provides a new solution for the autonomous cooperative control of "three machines" in fully mechanized coal mining face,makes a beneficial exploration for the intelligent development of fully mechanized coal mining face,and lays a foundation for realizing low-carbonization,intelligent and unmanned mining in coal mines.In this dissertation,the decision-making method of multi-equipment cooperative control on comprehensive mining face is proposed,which provides a new solution for the autonomous cooperative operation of key equipment on comprehensive mining face and lays the foundation of low carbon,intellectualization and without worker on comprehensive mining face.