A new method for repairing point cloud model holes of aero-engine compressor blades is presented in this paper, which can be used to repair point cloud model holes of aero-engine compressor blades accurately. Firstly, by constructing the boundary diffusion mechanism, the generation judgment of hole repair points was defined, and the generation method of hole repair points based on boundary diffusion mechanism was designed, which unifies the diversified hole repair modes; Secondly, the implicit surface based on radial basis function was constructed, the position of repair points was corrected accurately, the repair area with uniform density was generated, and the hole repair was completed. Finally, it was verified by an example and compared with the central diffusion repair method. The results show that: By using the method in this paper, various holes in the point cloud model of aero-engine compressor blades, such as plane hole, curved face, island hole and cross-face, can be repaired evenly. It also avoids noise confusion caused by irregular hole boundary, and realizes better fitting between hole repair point and curved surface, thus providing accurate data support for aero-engine compressor blade damage detection.

In order to establish the framework for aircraft life-cycle data management and ensure the efficient implementation of maintenance information management of in-service aircraft, the characteristics of maintenance information management of civil aircraft in service were analyzed. BOM views were classified, and a BOM multi-view model was constructed. By analyzing the main mapping forms of BOM multi-views, the node types of BOM multi-view were defined, the mapping rules of BOM multi-view model were proposed, and the mapping algorithm of BOM multi-view model was designed and implemented. Finally, method validation was performed by the BOM mapping of 48-section torsion box of a certain aircraft. The results show that, the mapping of EBOM to M-BOM is realized effectively, ensuring the full integration of the information management structure in the in-service maintenance phase and the development phase, and ensuring the consistency of the aircraft life-cycle data management framework.

Chiplet technology enables the integration of an increasing number of transistors on a single accelerator with higher yield in the post-Moore era, addressing the immense computational demands arising from rapid AI advancements. However, it also introduces more expensive packaging costs and costly Die-to-Die (D2D) interfaces, which require more area, consume higher power, and offer lower bandwidth than onchip interconnects. Maximizing the benefits and minimizing the drawbacks of chiplet technology is crucial for developing largescale DNN chiplet accelerators, which poses challenges to both architecture and mapping. Despite its importance in the post-Moore era, methods to address these challenges remain scarce.To bridge the gap, we first propose a layer-centric encoding method to encode Layer-Pipeline (LP) spatial mapping for largescale DNN inference accelerators and depict the optimization space of it. Based on it, we analyze the unexplored optimization opportunities within this space, which play a more crucial role in chiplet scenarios. Based on the encoding method and a highly configurable and universal hardware template, we propose an architecture and mapping co-exploration framework, Gemini, to explore the design and mapping space of large-scale DNN chiplet accelerators while taking monetary cost (MC), performance, and energy efficiency into account. Compared to the state-of-the-art (SOTA) Simba architecture with SOTA Tangram LP Mapping, Gemini's co-optimized architecture and mapping achieve, on average, 1.98× performance improvement and 1.41× energy efficiency improvement simultaneously across various DNNs and batch sizes, with only a 14.3% increase in monetary cost. Moreover, we leverage Gemini to uncover intriguing insights into the methods for utilizing chiplet technology in architecture design and mapping DNN workloads under chiplet scenarios. The Gemini framework is open-sourced at https://github.com/SET-Scheduling-Project/GEMINI-HPCA2024.

No abstract

No abstract