| As an important raw material for iron and steel smelting,the drying and preheating process of pellets in the grate makes a big impact on the subsequent handling.The equilibrium and stable control of the temperature field in the grate can provide an important environment for the pellet drying.In this paper,the grate down-draft drying section(DDD)is taken as the research object,based on the system identification method,a spatio-temporal coupling mathematical model is built.The decoupling of the space-time coupling mathematical model was studied and realized using output dimension reduction decoupling and feedforward compensation decoupling methods.Based on the decoupling model,the fuzzy neural net-PID control method was selected,and the fuzzy neural net-PID controller was designed.The spatiotemporal coupling control model of temperature field was established,and the decoupling and control effect were analyzed by simulation.Finally,the hardware and software of the control system are designed and built.Through experimental research,the control precision of the temperature field in DDD is improved.The main research contents and conclusions of this paper include:(1)According to idea of data-driven,the spatio-temporal coupling model of the temperature field in DDD was deduced and established using the input and output data of the test system.Firstly,the test system of DDD is made up of 12 temperature sensor array to monitor the temperature field,and system identification is carried out by controlling the input heat system of the input air temperature,hot air blower fan wind speed and wind speed of exhauster.The state space model,transfer function model and process model were used respectively,and the state space model identification effect is best.The identification accuracy of the state space model is the highest in all temperature sensors,so the time dimension model is established based on the state space to reflect the temperature change of each temperature sensor with time under the action of three inputs.Secondly,the spatial dimension model of temperature field was established based on RBF neural network,the network input is the temperature at each sensor,and the output is the temperature at any position of the temperature field.Through the simulation analysis of the temperature field by FLUENT software,the temperature field in the best working condition of pellets production is found.Based on the temperature field,the parameter identification training of RBF neural network is carried out.Compared with the model established by biharmonic spline interpolation method,the established spatial dimension model based on RBF neural network has less identification error in the case of steady-state temperature field identification.Finally,the time dimension model was substituted into the spatial dimension model to obtain the spatial-temporal coupling mathematical model of temperature field in DDD.(2)Decoupling of input and output of the spatio-temporal coupling model of temperature field in DDD.Because feed-forward compensation decoupling is only applicable to the input number is equal to the number of output,this paper proposes a descending dimension decoupling method of output based on principal component analysis(PCA),first of all to debasing dimension of outputs,after the dimension reduction system input number is equal to the output number,and then using feed-forward compensation decoupling method for decoupling between system inputs and outputs.Through the simulation of decoupling between inputs and outputs of spatio-temporal coupling temperature field model of DDD,the results show that the method can effectively realize decoupling.The temperature field system becomes three simple control channel after decoupling to facilitate the follow-up temperature control of DDD temperature field.(3)A fuzzy neural network-PID control strategy is proposed to control the temperature field of DDD,and its control method are designed for each of the three temperature control channels of the temperature field in the decoupled DDD.The fuzzy neural network topology structure is designed based on the fuzzy self-learning ideas.The PID control parameters are adjusted by the output of the network,and the PID parameters are self-tuning for the three fuzzy neural networks.It can be found by simulation that,the proposed fuzzy neural network-PID control strategy has higher control precision,better control robustness and stronger anti-interference ability than fuzzy PID algorithm,and it can realize the temperature field control of DDD.The fuzzy neural net-PID control model of temperature field was established to simulate the temperature field of DDD.The maximum control error of temperature field is2.3?,which meet control requirement.(4)The temperature field spatio-temporal coupling control test equipment was built,and the corresponding software and control program were written.The temperature field spatio-temporal coupling model verification experiment and temperature field control experiment were carried out in DDD.The experimental results show that the maximum error of the temperature field spatial-temporal coupling model is 5.8?,and the maximum error of temperature field control is 8.1?.The experimental results are in good agreement with the simulation results,which proves the feasibility of the spatio-temporal coupling model of the spatio-temporal temperature field of DDD and the fuzzy neural networked-PID control strategy.Compared with the temperature control error of 20? in the actual grate operation,the accuracy of temperature field control in DDD is improved to some extent.The research in this paper provides a theoretical basis for the equilibrium and stable control of the temperature field of the grate and has a reference value for improving the drying effect of pellets. |
PDF Full Text Request