The Application of Neural Network Control Algorithm in RF-excited CO2 Laser Power Supply

In order to improve the injection power of the laser and guarantee the normal operation of the RF-excited Fast Axial Flow CO2 Laser, A kind of neural network control technology is proposed in a 2MHz laser RF power supply, which consists of two single-phase half-bridge MOSFET inverter modules. By using the SLPS Interface (SLPS), the RF power supply system with neural network controller is built and simulated. The result shows that the output power from the laser power supply can be up to 2000W, and a little of distortion (THD is 0.013%) is obtained. Moreover, it suggests that the neural network controller possesses better adaptivity than the PI controller to nonlinear, time-varying uncertain impedance characteristics of the load. So, the analysis of this paper shows that this kind of control algorithm scheme is fully applicable to the control design requirements of the laser.