Zhang Aihua_IEEE dataset

Evaluating the collaborative efficiency of industrial academic research (IAR) systems in industrial design remains a key challenge under computationally intensive, real-time conditions. Fragmented GPU resource allocation and asynchronous design\u2013simulation workflows often hinder innovation. To address this, we propose a slack-based measurement framework for data-driven optimization of design collaboration that integrates real-time GPU scheduling metrics into a three-stage network slack-based measure-network data envelopment analysis (NSBM-NDEA) model. The system incorporates GPU-hour reallocation rates, user validation cycles, and sustainability outputs such as carbon efficiency and material utilization metrics to diagnose inefficiencies across 35 IAR projects in Shandong, China. By fusing GPU scheduling logs with innovation outcome indicators, the model identifies slack variables and stage-specific bottlenecks across resource mobilization, human\u2013AI knowledge co-creation, and outcome transformation. Designed with modular stage definitions and abstracted data interfaces, the framework supports replicability across platforms, and scalability across industries. The empirical results show that elevated GPU slack significantly reduces co-creation efficiency (\u03c1 = \u20130.72, p < 0.01), reinforcing the importance of dynamic orchestration. This study proposed a transferable framework for performance evaluation and real-time resource coordination in industrial design ecosystems. The model yields direct practical benefits, including enhanced GPU scheduling strategies for intelligent manufacturing systems and accelerated Computer-Aided Engineering (CAE) prototyping workflows within digital twin environments.