Research On Nonlinear Multi-Agent Collaborative Control Strategy For Mine Inspection Robot

In mine production activities,underground safety inspection is an important task to ensure safe mine production.However,due to the complex underground environmental conditions in most mines and the difficulty of applying new technologies,manual inspections are basically adopted for underground safety inspections.However,manual inspection often takes a long time,and the information obtained will be mixed with some subjective factors of inspectors,which brings problems such as low information acquisition efficiency,untimeliness,and low credibility.This is also the main cause of many mine accidents and casualties.It can be seen that the inspection results of manual inspection method is not only low reliability,but also exist hidden dangers on 'the safety of the inspection personnel themselves.According to the goal of mine unmanned construction in the future,the application of mine inspection robot will be an inevitable trend.With the maturity of multi-agent systems,it becomes possible to implement complex functions with the cooperation of multiple agents.Therefore,based on the cooperative control of multi-agent,the application of mine inspection robot is studied in this thesis.The specific work is as follows:Aiming at the problem of path planning and obstacle avoidance for the multi-agent of mine inspection robot to realize the given inspection function,a deep reinforcement learning algorithm based on the fusion of neural network and Q-learning algorithm is used to complete the obstacle avoidance of the agent on the basis of realizing the self-adaptive path planning of multi-agent inspection robot,to solve the disaster of traditional dynamic planning dimension,and the convergence and feasibility of the algorithm are verified by simulation.Finally,the information sharing network is established under the inspiration of the organism flora cooperation mechanism to optimize the collaborative control of the multi-agent system of the mine inspection robot.Aiming at the abrupt change in the state of the environment under the mine,the inspection robots in the area need to have the ability to cope with unexpected conditions,and the energy conservation issues of the inspection robots also need to be considered.On the basis of making full use of existing mining equipment,the adaptive control law is designed through RBF neural network and system dynamics equations,and the stability of multi-agent systems is studied by using Lyapulov's stability principle and graph theory.Finally,the consistency of multi-agent system of mine inspection robot with leader is studied by simulation,and the rationality of control law design is verified.Aiming at the difference of monitoring information obtained by multi-agent in underground mine and the requirements of each type of information on the information transmission network,a "fiber + Wifi+LoRa" mode integrated transmission network was established to reduce the energy consumption of agents while ensuring the normal information transmission requirements of multiple agents.At the same time,under the condition that the communication topology of the multi-agent system of mine inspection robot changes due to obstacles or accidents,the research method of switching topology consistency control is proposed to achieve the consistency of the multi-agent system under the dynamic topology structure,and the local non service computing function of edge computing is used to realize the information collection,continued transmission,local storage and self-determination of the robot agent Policy and other functions.In view of the transmission delay existing in the information transmission between multi-agent,the transmission delay is classified,and the corresponding control laws are designed according to the delay categories.The stability of the system is verified by the Lyapunov stability judgment principle.Finally,the consistency of the system is verified by simulation.