Engineering change (EC) is a complex, manual, and expert-driven topic. It has
significant downstream effects on logistic operations and cost structures. The
impact of logistics cost is a critical consideration for any profiting company
or original equipment manufacturer (OEM). Evaluating the logistics cost impact
of an EC is a time-consuming and tedious labor-intensive task. There are
multiple steps taken by engineers before making an evaluation of the logistics
cost of an EC, these include examining multiple sources, computer-aided design
(CAD) drawings, PDF documents, PowerPoint files, and descriptions of the
modification. Automation is further complicated by the wide variation of ECs
across vehicle model, module group, and product type. To address this, we
introduce the logistics impact screening application (LISA), an AI-based system
designed to predict logistics cost impacts automatically. LISA pulls together
both structured and unstructured data and uses a mix of techniques: machine
learning methods (like logistic regression, support vector machine (SVM), random
forest, gradient boosting, AdaBoost, XGBoost), deep learning approaches (such as
long–short-term memory (LSTM), bi-LSTM, multilayer perceptron (MLP), gated
recurrent unit (GRU), and even large language models (OpenAI’s GPT-4o)). Each
method was evaluated for accuracy, interpretability, and computational
efficiency. Results demonstrate that certain configurations achieved 83%–85%
accuracy and reduced assessment time by up to 70% compared with manual
evaluation. This indicates that AI-powered prediction is not merely feasible,
but also efficient and cost-effective. By enabling engineers to rapidly
comprehend the ECs’ logistics effect, LISA is a scalable method for enhancing
decisions. Increased application within carmaker OEM processes can enable
proactive cost management and logistics planning optimization.