This thesis presents an on-line feedback control method to obtain uniform weld quality in $CO_2$ gas arc welding process. In this study, weld pool size is chosen as an output variable to represent weld quality, and power is used as an input variable to represent weld quality, and power is used as an input variable to control it. To construct the control system encompassing throughout whole welding process, this thesis consist of four parts: scheduling heat input power for initial transient region, searching a stable arc condition to adapt to varying power under control, estimating the weld pool size using multiple point surface temperatures during welding and finally designing a feedback control system.
1) Heat input scheduling :
A mathematical method is presented for scheduling heat input powers to make uniform desired weld pool size during initial transient region. In this method, the powers to give the desired weld pool size are obtained through linear programming technique. An analytical weld thermal model is used in this procedure. To investigate the effectiveness of the method, simulation and experimental studies are performed under various welding conditions. The results show that the effectiveness in experiment is not so good as that in simulation. This is due to model uncertainty and unknown in-process disturbance, thus provides the need of on-line feedback control.
2) Searching a stable are condition:
For varying heat input power in the above and probably occurred during feedback control, stable arc condition(welding voltage/current combination) should be basically maintained to obtain the desired weld pool size. A fuzzy rule-based searching method is presented to obtain such a stable voltage/current combination for a given power. In estimating arc stability, Mita's arc stability index is used. The effectiveness of the method is investigated through a series of experiments. The results show that the presented algorithm provides fast searching ability.
3) Estimating the weld pool based on surface temperatures:
During welding, weld pool size which is the output in the feedback control system cannot be directly measured. However, fortunately surface temperatures reflects the behavior of weld pool. Thus, a weld pool size index is constructed based on multiple surface temperatures. Multiple regression method is used in this procedure. The effectiveness of the index is investigated through a series of simulations and experiments, as compared with the case of one point temperature. The results show that the index provides a direct estimation method of weld pool size and it has more precise estimating ability than one point temperature.
4) Designing the feedback control system:
Based-on the above results, a feedback control system of weld pool size is designed to cope with the unwanted variation in weld pool size. Due to the complexity and nonlinearities of the process, a fuzzy rule-based predictive controller is devised. The effectiveness of the control system is investigated through a series of simulations and experiments, as compared with the cases of no control and PID control. The simulation results show that the proposed control algorithm provides a desirable control performance. In the experimental result, the good performance is also obtained in the case of the disturbance. However, in the initial transient region the deteriorated performance is resulted in. This is due to the estimation error between the actual pool size and the estimated one through the index. To obtain the desirable control performance in this region, improvement in the estimation performance of weld pool size is needed.