Abstract: The welding is one of the most important and prevalent processes for shipbuilding. So welding distortion is inevitable and rectification becomes important task. Aiming at the situation of manual operation in flame rectification, prototype is developed to realize automation of flame rectifica- tion process, three kinds of typical welding components are chosen as objects to establish automatically standard flame rectification process by logic reasoning. In addition, appropr- iate process parameters are obtained based on neural network. An example is implemented to attain approving rectification result. It indicates that the system can improve the efficiency of flame rectification and stability of rectification quality. As a result, the fundament for automation of flame rectification process is established.

Key words: Flame rectification　Shipbuilding Logic reasoning　Neural network