| Safety is a nonnegligible and significant topic for mine production.In order to avoid the dangerous and enhance safety as much as possible for mine production,the management rules need to be required strictly as well as the latest technologies,which includes cyber physical system(CPS),fog computing and Reinforcement learning etc.,are necessary to be relied on.CPS provides a new research idea for safety production and disaster warning in underground mine.Along with the construction advancement of smart mine,the number of edge perception and control terminals,which is responsible for production data perception in mine,is also increasing daily.These terminals are scattered and independent.It is necessary to intergrate the data resource of independent production subsystems,then strengthen the information interaction by relying on mine CPS so that each subsystem can work sequentially and coordinately.As a Consequence,the intelligent work of unattended and less participated can be realized.Nevertheless,the volume of mine production data demonstrates an explosive growth trend,and the current distribution of data computing resource can not satisfy the property of real-time data analysis in the intelligent process of underground mine.In addition,the poor communication environment in underground mine leads to that the wireless transmission performance is inferior to the ground,and the increasing number of wireless edge perception and control terminals in underground mine also aggravate the problem of channel congestion,which further affects the reliability of data transmission.Finally,for battery-powered wireless edge perception and control terminals,improving transmitting power can improve transmission performance,whereas shorten its lifetime as well.Such problems are the keys that must be solved in the construction of smart mine.Based on the problems shown above,considering the particularity of mine production environment as well as the requirements of real-time data computing and reliable transmission,computing resource allocation,channel allocation,and power control in perception and execution layer of mine CPS is studied in this dissertation.The main research results are summarized as follows.1)The theory of computing resource allocation based on fog-edge cooperation in mine is studied,and the dynamic mechanism of mine computing resource scheduling is constructed.A resource scheduling algorithm driven by computing resource value stream is proposed based on the demand and supply of computing resource.In view of the optimization problem on computing resource allocation based on fog-edge coorperation in mine CPS.The theory of mine computing resource allocation is proposed according to the thought of resource value,and the flow model of computing resource value based on fog-edge coorperation is established in mine CPS.Then,hardware resources are quantified into computing resources by virtualizing and then treated as resource pool of fog computing nodes.The flow of resource value is realized by computing resource allocation between each fog computing node according to the computing demand of edge perception and control terminals.A resource allocation algorithm driven by value flow of computing resource is proposed to achieve the dynamic balance between the demand and supply of computing resources in mine CPS.Numerical simulation results show that the proposed algorithm can effectively improve the utilization of computing resource.2)A channel allocation algorithm for underground mine is studied.Target-oriented system model of throughput maximization in channel reusing mode is established,and channel clustering is carried out based on transmission performance analysis.The channel allocation mechanism is studied under the constraint of SINR.For the specific communication scenarios in mine,the data transmission model in channel reusing mode is established.Firstly,according to the results of transmission performance test and topology analysis in the underground mine,an optimal node deployment scheme is obtained.Then,a system model aimed at maximizing the system throughput is established based on the efficient constraint mechanism of channel allocation.After that,the distributed power control solution is used to configure the transmit power of cognitive users(CUs),and clustering of potential users is accomplished by combining the performance requirements of communication quality for cognitive network when reusing channels.Lastly,the actual users of the channel corresponding to each cluster are determined by the SINR weighted efficiency mechanism of potential users.Simulation experiments demonstrate that the proposed algorithm can guarantee the transmission performance of wireless edge perception and control terminals in mine CPS.3)A power control strategy is studied for CUs in underground mine.The power control algorithm based on deep Q network is proposed by discretizing the continuous transmission power threshold and establishing markov decision process model.The continuous threshold of transmit power for CUs is obtained by SINR constraint,and the mathematical model of arithmetic discretization in continuous threshold space is established which ensures the timely and effective convergence of the DQN.Then,the power control for CUs is rebuilded with MDP,and the corresponding state space,action space and reward function are given.Next,the features of multidimensional state space are extracted and the dimension of state is reduced to promote the convergence performance of DQN training.Numerical simulation results demonstrate that the states of CUs can converge fastly and effectively in any initial states with the assistance of proposed algorithm.Lastly,the feasibility of above three methods is verified by field experiment combining with the actual mine production environment.The differences of results and corresponding reasons of simulation experiments and field ones are analyzed as well.According to the result analysis,the proposed three methods in this dissertation are conducive to improve the performance of data computing and transmission in perception and execution layer of mine CPS.Figure[46]Table[2 7]Reference[157]... |
PDF Full Text Request