Characteristic Parameters Prediction of Robotic Welding Based on Belief Rule-Based Inference

Welding is a complex industrial process, and accurately predicting the characteristic process parameters is crucial to ensuring the quality of weld formation. Among the factors affecting weld formation quality, there are uncertainties, including unknown influences and inaccurate variable values. Traditional data-driven methods, including rule-based reasoning and various neural network approaches, generally establish deterministic models when predicting weld formation quality, neglecting the impact of uncertainties. This paper introduces the belief rule-based reasoning method, which can address uncertainties in the welding process. By fully considering various uncertainties, a belief rule-based reasoning rule base for the welding process is established. This method has achieved better results in predicting actual welding process parameters compared to traditional rule-based reasoning methods and neural networks.