Data–driven Based Predictive Control Design For Temperature Of Molten Pool During Laser Additive Manufacturing

Laser metal additive manufacturing is an innovative manufacturing technology derived from laser cladding technology and rapid prototyping technology. It is capable of rapidly forming and manufacturing products with complex structure. In laser additive manufacturing process, melting pool’s temperature directly affects product’s forming precision, metallurgical defect, solidifying microstructure and mechanical property. Therefore, a closed-loop control of melting pool’s temperature can effectively improve the stability of the molten pool and the quality of the forming parts. However, limitations exist for the melting pool’s theoretical temperature model and the PID control algorithm. For example, melting pool’s theoretical temperature model often needs a massive calculation, and PID control algorithm is intractable in setting up multiple constraints in control system. Therefore, a suitable algorithm of closed-loop control for molten pool temperature is not common for laser additive manufacturing process.The questions mentioned above are systematically researched in this paper. This research combines subspace system identification with the model-based predictive control algorithm and developed a data–driven–based close-loop control system of melting pool’s temperature. Finally, an ideal result was proved through the experimental verification. The main achievements are as below:Firstly, a stable and high-efficient laser additive manufacturing monitoring system is developed. The system is capable of intelligently detecting and control the powder and gas, working simultaneously with robot, and fault detection alarming.Secondly, a data–driven–based predictive control algorithm is propo sed and a melting pool’s temperature control system is built based on it. This algorithm combines subspace system identification with the model-based predictive control algorithm, and directly uses the I/O data to design a control system. Therefore, this a lgorithm can be used in designing a control system which is hard to build an accurate mathematical model for a complicated industrial production process. Furthermore, this control system uses two-color pyrometer to detect melting temperature, and directly adjusts laser power to control melting pool’s temperature. Besides by utilizing random inputs work on the system, the system’s I/O data is obtained for designing the controller. And through system identification, the subspace predictor presenting the dynam ic relationship of laser power and melting pool’s temperature is obtained. Finally, the data–driven–based controller is designed. Offline simulation is used to select proper parameters for the controller. Then this controller is tested in a real laser addi tive manufacturing system to follow the preset temperature curve. Results showed that the melting pool’s temperature could be effectively controlled when this temperature control system is applied.Finally, thin–wall depositing experiment is conducted on t he convex shaped substrate to verify the effectiveness of using melting pool’s temperature control system and the effect of using melting pool’s temperature control on deposition’s morphology, microstructure and hardness. Results reveal that this system ca n effectively improve the stabilization of melting pool’s temperature, and a more stable temperature helps improve the thin–wall product’s accuracy, homogenize its microstructure and increase its hardness.