Analysis and Prediction of Melt Pool Geometry in Rectangular Spot Laser Cladding Based on Ant Colony Optimization-Support Vector Regression.

Abstract

The rectangular spot laser cladding system, due to its large spot size and high efficiency, has been widely applied in laser cladding equipment, significantly improving cladding's efficiency. However, while enhancing cladding efficiency, the rectangular spot laser cladding system may also affect the stability of the melt pool, thereby impacting the cladding's quality. To accurately predict the melt pool morphology and size during wide beam laser cladding, this study developed a melt pool monitoring system. Through real-time monitoring of the melt pool morphology, image processing techniques were employed to extract features such as the melt pool width and area. The study used laser power, scanning speed, and the powder feed rate as input variables, and established a prediction model for the melt pool width and area based on Support Vector Regression (SVR). Additionally, an Ant Colony Optimization (ACO) algorithm was applied to optimize the SVR model, resulting in an ACO-SVR-based prediction model for the melt pool. The results show that the relative error in predicting the melt pool width using the ACO-SVR model is less than 2.2%, and the relative error in predicting the melt pool area is less than 9.13%, achieving accurate predictions of the melt pool width and area during rectangular spot laser cladding.