Rough Fuzzy Soft Computing And Its Application In Welding Dynamic Process Modeling

The trend of welding technology is welding intellectualization.The complexity of welding technology determines that it is difficult to obtain accurate knowledge model of welding process,which limits the application of traditional control methods in welding field.Intelligent knowledge modeling has attracted more and more attention because of its adaptability to highly complex welding process.At present,the main methods of welding process modeling are rough set modeling,fuzzy logic modeling and neural network modeling.Rough set theory is a mathematical tool,which can analyze a large number of incomplete data and automatically obtain rules from a large number of data.Compared with Rough Set,fuzzy logic model is a mathematical model that does not need to know the specific controlled object.It has good adaptability and robustness.Aiming at the weakness of Rough Set in analyzing continuous and fuzzy data in welding system,combining Rough Set Knowledge Modeling with Fuzzy Logic Modeling,giving full play to their respective advantages in rule knowledge acquisition and fuzzy quantitative analysis,a knowledge model based on rough-fuzzy soft computing is proposed.Its main process is divided into four parts: raw data acquisition,data preprocessing.Fuzzy rule acquisition and approximate reasoning and verification.In this paper,the knowledge model is applied to tungsten inert gas shielded welding,and the weld width on the back is predicted.The predicted qualified rate is 93.75%,which can meet the requirements of welding process.To solve the problem of time-varying,fuzzy and random uncertainties of system parameters in actual welding process,a reasoning method of similarity correction relation is proposed.By constructing extended and reduced modified relation functions based on similarity measure and selecting appropriate approximate reasoning operators,the approximate reasoning of welding process can make every change,magnitude and trend of system parameters.Potential to make real-time accurate response.In the process of carbon dioxide gas shielded arc welding,a reasoning model of similarity correction relationship is established to predict the weld formation and compare with the synthetic reasoning method.The experimental results show that the precision of the similarity modified relational reasoning method is slightly higher than that of the traditional synthetic reasoning method.