Data-Driven Optimization Method for Aluminum Alloy Casting Process Parameters and Alloy Composition

Melting and casting are important parts of aluminum alloy production, in which process parameters and alloy composition are key factors in determining ingot performance and are essential to aluminum alloy product quality. In this paper, a data-driven method for in-depth optimization of process parameters and alloy composition in an aluminum alloy casting process is presented, which can automatically optimize the process parameters and alloy compositions to improve the ingot performance. The method includes three parts: data pre-processing, relational model construction, and process parameters/alloy composition optimization. At the data pre-processing stage, feature engineering, outlier handling, and missing value handling are used in turn to obtain a high-quality dataset. At the relational model construction stage, XGBoost is used to construct a relational model between the process parameters/alloy composition and ingot performance. At the process parameters or alloy composition optimization stage, the genetic algorithm is used to optimize the process parameters and alloy compositions respectively. To verify the effectiveness of the proposed algorithm, the melting and casting production process dataset of aluminum alloy is constructed. Experiments show that our algorithm can automatically optimize the process parameters or alloy compositions in a short time, and significantly improve the performance of ingots. The proposed method is an important guideline for practical production.