Intelligent Control And Experiment Studies On Tandem Skew Rolling Equipment Produced Steel Tube

Tandem Skew Rolling(TSR)is a new technology of short process for the production of seamless steel tube,which integrates piercing with rolling.The billet after heating is pierced in the piercing section,and then is rolled in the rolling section.The continuous rolling relationship is formed in two groups.Tandem Skew Rolling equipment has the advantages of simple structure,convenient adjustment,production cost reduction,production efficiency increase,etc.Because the temperature drop of the piece is very small in the piercing and rolling process,the process can roll the metal of small temperature area or difficult deformation,and the key of the implementation of the new process is:(1)how to process the metal of temperature range narrow or difficult deformation and realize continuous forming of piercing and rolling for the pipe;(2)how to improve quality,reduce cost,satisfy the goals of steel pipe size precision and internal organization.The effective means and ways to solve the key problems are to establish the models of Tandem Skew Rolling process and research the control method of the system.According to the characteristics of complex nonlinear,dynamic multivariable,continuous rolling,strong coupling,the speed setting model,rolling force setting model,the tandem rolling tension model,etc are studied by the rolling theory,intelligent control and predictive control in Tandem Skew Rolling process.The main research results are as follows:(1)For the speed control system of Tandem Skew Rolling,the speed is demanded to response quickly,small dynamic downhill,high synchronicity,etc.This paper proposes a method of the speed synchronization control based on memristor PID.On the basis of single neuron PID control algorithm,combined with the resistance change characteristics of memristor,the memristor PID controller is established.The algorithm has simple structure and is closer to the biological neuron learning characteristics.Finally,the speed synchronization control based on memristor PID is realized by the deviation of coupling synchronization control strategy.The tests show that the control method has theadvantages of strong real-time performance,high synchronization control accuracy and response speed,etc.(2)Because the tension control system has the traits of complicated nonlinear,the influence of uncertain disturbance,the paper establishes Dynamic Matrix Control prediction algorithm of the speed and tension system.First the dynamic mathematical model of the speed and tension system is established according to the theory of skew rolling and tandem rolling as well as the dynamic characteristics of the motor.Because the model has the characteristics of nonlinear,strong coupling,the traditional PID control can not achieve good control effect.The model accuracy demand of Dynamic Matrix Control prediction algorithm is not high,and the algorithm has strong robustness for the uncertainty factors such as environmental interference.Under the constraint conditions of corresponding rolling state,the optimal control value is calculated by the rolling optimization of the objective function using Particle Swarm Optimization(PSO).The method solves the nonlinear optimization problem under constraint conditions.The simulation results show that this method inhibits effectively the sensitivity of model parameter change,the influence of uncertain disturbance.The micro tension control is realized.(3)The paper puts forward the forecast model of steel thickness based on the STDP-Spiking Neural network algorithm.On the basis that Spiking Neural network(Spiking Neural Networks,SNN)is analyzed,STDP-Spiking Neural network algorithm is proposed combining with Spike Timing Dependent Plasticity(STDP)learning mechanism of biological.The network uses single synaptic connections between neurons.In the weights learning,the influence for the synapse of the pulse distribution lag of neurons before and after is considered on the basis of error feedback algorithm.The network structure not only is simplified,but also overcomes that the weights only can take positive value in the traditional SpikeProp algorithm.In the test,the main factors influencing the thickness changes of seamless pipe are analyzed.Further the prediction model of steel tube wall thickness is trained using actual data as training samples,and finally the actual values are compared with the estimated.The experiments show that the high prediction accuracy of STDP-Spiking neural network can meet the requirements of prediction for the wall thickness of the steel.(4)According to the texture theory of metal deformation,on the basis of the theoretical of predecessors' research,this paper establishes the microstructure prediction model of TSR equipment.Due to TSR process belongs to the deformation above the recrystallization temperature,the rolling process is given priority to dynamic recrystallization.The sizes of austenite and ferrite grain are predicted.The calculated values are compared with experimental values through metallographic experiment.The results show that the model can be used to predict the microstructure of TSR.