Review of Data-driven Decision Support Systems and Methodologies for the Diagnosis of Casting Defects

The concept of 'Industry 4.0' has introduced great dynamism into production environments, making them more integrated, connected and capable of generating large volumes of data. The digital transformation of traditional companies into innovative smart factories is made possible by the potential of Artificial Intelligence (AI), which is able to perform predictive analytics inspired by the development of Industrial Internet of Things (IoT) technologies or to support highly complex decision-making, in the era of zero-defect manufacturing. The need for innovative techniques and automated decision-making in diagnosing the causes of casting defects is increasing due to the growing complexity and higher level of automation of industrial systems. Particularly important are fully data-driven predictive approaches that enable the discovery of hidden factors influencing defects in castings and the prediction of the specific time of occurrence by analyzing historical or real-time measurement data. In this context, the main objective of this article is to provide a systematic overview of data-driven decision support systems that have been developed to diagnose the causes of casting defects. In addition, different methods for predicting casting defects are presented. Finally, current research trends and expectations for future challenges in the field are highlighted. It is hoped that this review will serve as a reference source for researchers working in the field of innovative casting defect prediction and cause diagnosis.