Research On Control Method Of Main Fan Switching Process Based On Model Predictive Control

When corrosive gas and water vapor in the underground are discharged from the wellhead,they gradually corrode the fan components,reduce their operating efficiency,and affect the safe production of the mine.In order to ensure continuous production in the mine,it is necessary to switch the fan and put the standby fan into use.According to the requirements of "Coal Mine Safety Regulations",it is necessary to complete a fan switch once a month to ensure safe production under the mine.However,the fluctuation of the underground air volume caused by the switching process of the main fan can also affect the gas concentration,creating potential safety hazards.Therefore,it is necessary to control the switching process,maintain the stability of the underground air volume fluctuation,and complete the switching process of the mine fan within ten minutes.Due to high reliability requirements for mine ventilation systems during main fan switching process effective control of underground air volume is required to provide a stable amount of underground air.However,fan switching system is complex with seven state variables including air volume of four air doors two ventilation pipes and underground air volume.Each state variable is coupled with each other and changes in one will affect others.In addition some system model parameters are related to operating conditions which affect dynamic characteristics making it difficult to measure accurately.During fan switching performance requirements for each component must also be considered imposing state constraints on system.Traditional control methods are not applicable so this paper proposes a model predictive control based mine fan switching system aiming at optimization control The main work of this paper can be summarized as follows:(1)First,analyze the single-segment ventilation duct and establish its mathematical model.Then,establish the network topology of the fan switching system through the analysis of the system.Finally,combine the mathematical model in the ventilation duct with its network structure to establish the dynamic mathematical model of the main fan switching system.(2)An approximate linear model is constructed using Taylor expansion,a prediction model is constructed,and an objective function is established based on optimization indicators,constraints are set,and then it is transformed into a quadratic programming problem.A quadratic programming solver based on a primal dual neural network is used to quickly complete the solution to ensure the real-time performance of the system.(3)By using a regularized incremental random weight network to estimate the unmodeled dynamics online and compensating the unmodeled dynamics into a linear model,a linear tracking error model for a wind turbine switching system with unmodeled dynamics can be obtained.Then,a model predictive control strategy is used to convert the air volume optimization problem of the fan switching system into a quadratic programming problem with constraints.Finally,the primal dual neural network is analyzed to illustrate its convergence and stability.(4)The problem is solved by using particle swarm optimization and PID algorithms,and the actual control effects of various methods in this paper are compared.The simulation results show that the model predictive controller has better performance than other methods in this paper,can meet all hardware constraints and ensure stable air volume underground.