Predictive Control Of Friction Stir Welding Temperature

In the process of friction stir welding,the welded metal will be plasticized because of friction heat with the stirring head,and the plasticized degree of the welded metal often has a decisive influence on the quality of the weld.Welding temperature is an important sign to observe the plasticized degree of welded metal.Effective control of the temperature around the friction stir head is of great significance to improve the welding quality.In this paper,the genetic BP neural network algorithm is used to establish the temperature prediction model of Friction Stir Welding forward direction.The training data and test data used in the neural network model are randomly taken from the data of FSW temperature experiment.Using the same training data,the BP neural network is used to establish the temperature prediction model.By comparison,the temperature prediction model established by genetic BP neural network has less error and better fitting with the actual temperature value,which can be applied in practice.Based on the temperature prediction model established by genetic BP neural network,the temperature control system of friction stir welding is constructed.The fuzzy controller is used to optimize the external closed-loop,the welding temperature is controlled by adjusting the rotating speed of the stirring head,and the model predictive controller is used to optimize the internal closed-loop,which mainly solves the problem of welding temperature delay.Based on this idea,using Simulink of MATLAB for simulation research,the simulation results are compared with the output of the original open-loop temperature control system.The results show that the output of the closed-loop temperature control system is not affected by welding displacement,and can effectively track the target temperature value.In order to better analyze the designed system,the model predictive controller in the closed-loop temperature control system is replaced by the PID controller,and a new temperature control system is established.The simulation results show that the temperature control system constructed by the fuzzy controller and the model predictive controller has smaller temperature error and smaller change rate of temperature error,so the temperature traceability is better.Moreover,the speed regulation of the stirring head of the system is smaller,so the execution efficiency is more efficient.On the basis of the original friction stir welding equipment,the temperature closed-loop control of friction stir welding is realized by using the TMS320F28335,touch screen,infrared thermometer and PC terminal.The hardware part of the temperature control system is mainly based on the DSP chip of TMS320F28335,and other related peripheral circuits are designed according to the actual needs of the temperature control system.On the basis of reliable hardware composition,software designs are made for DSP,touch screen and PC respectively.In the aspect of DSP,with the main program module as the core,the timer module,serial communication module,AD acquisition module and FIR filter module are designed respectively,which are used to realize the mutual communication between the devices in the temperature control system and the acquisition and processing of real-time temperature data.In touch screen,embedded configuration software is used to design the interface of DSP mode and manual mode,which is convenient for operator to switch between automatic control and manual control of welding device.The display modules in the interface are also convenient for real-time data monitoring.On the PC side,we use MATLAB to write serial communication program,and Simulink is called to optimize the real-time algorithm for temperature data in the program.