Intelligent system for generating quality reports based on the synergy of machine learning and large language models: overlapping the gap between diagnostics and the quality management system

The article examines the problem of the gap between the detailed results of machine learning in the tasks of quality diagnostics in mechanical engineering and the needs of quality management systems in complex, interpretable and standardized reporting. An intelligent system based on synergistic integration of distributed machine learning modules for local diagnostics and large language models for automatic generation of context-dependent reports of the quality management system is proposed. The methodology, system architecture, algorithms of aggregation of diagnostic information, techniques of engineering of proxies for large language model and mechanisms of verification of generated reports are described, including factchecking for minimization of «hallucinations».

The results of the pilot testing on typical quality control tasks confirm system performance, high accuracy of machine learning modules components and ability to generate factually accurate, relevant and compliant quality management system reports. The proposed approach allows to effectively bridge the gap between technical diagnosis and management needs of quality management system, automating the process of forming intellectual reporting on quality.

1. Beketova O. N., Arifullin M. V., Frolov A. L. Innovatsionnyye tekhnologii upravleniya kachestvom produktsii na predpriyatiyakh mashinostroyeniya [Innovative technologies of product quality management at mechanical engineering enterprises]. Vestnik Altayskoy akademii ekonomiki i prava. Bulletin of the Altai Academy of Economics and Law. 2023. No. 4-1. P. 18–22. DOI: 10.17513/vaael.2756. EDN: FWPCGA. (In Russ.).

2. Wang S., Zhao J., Ta N. [et al.]. A real-time deep learning forest fire monitoring algorithm based on an improved Pruned + KD model. Journal of Real-Time Image Processing. 2021. Vol. 18. P. 2319–2329. DOI: 10.1007/s11554-021-01124-9.

3. Blagorodov A. A., Prokhorov V. T., Korablina S. Y. [et al.]. Ob osobennostyakh obespecheniya kachestva sistemy menedzhmenta dlya proizvodstva vostrebovannoy i prioritetnoy produktsii [On the features of quality assurance of the management system for the production of in-demand and priority products]. Fundamental’nyye i Prikladnyye Nauchnyye Issledovaniya v Oblasti Inklyuzivnogo Dizayna i Tekhnologiy: Opyt, Praktika i Perspektivy. Vol. 3. Moscow, 2025. P. 191–198. EDN: EULQLN. (In Russ.).

4. Borgardt E. A., Bobel’ D. N. Tekhnologii iskusstvennogo intellekta v sisteme upravleniya kachestvom [Using artificial intelligence technologies in the quality management system]. Mezhdunarodnyy zhurnal gumanitarnykh i estestvennykh nauk. International Journal of Humanities and Natural Sciences. 2021. No. 8-1 (59). P. 178–180. DOI: 10.24412/2500-1000-2021-8-1-178180. EDN: JHAFSD. (In Russ.).

5. Zubkov V. I. Innovatika protsessov upravleniya kachestvom v slozhnykh tekhnicheskikh sistemakh [Innovation of quality management processes in complex technical systems]. Vestnik Ufimskogo gosudarstvennogo aviatsionnogo tekhnicheskogo universiteta. Vestnik USATU. 2012. Vol. 16, no. 6 (51). P. 242–251. (In Russ.).

6. Kuznetsov M. S., Kalmykova S. V. Iskusstvennyy intellekt v upravlenii kachestvom [Artificial intelligence in quality management]. Dialog. Dialogue. 2024. No. 3. P. 37–39. (In Russ.).

7. Malkina I. V. Avtomatizatsiya kontrolya kachestva geometricheskikh parametrov detaley v mashinostroyenii [Automation of quality control of geometric parameters of parts in mechanical engineering]. Izvestiya Samarskogo nauchnogo tsentra Rossiyskoy akademii nauk. Izvestia of Samara Scientific Center of the Russian Academy of Sciences. 2022. Vol. 24, no. 3. P. 34–41. DOI: 10.37313/1990-5378-2022-24-3-34-41. EDN: DDURMC. (In Russ.).

8. Popenko I. I. Avtomaticheskiye sistemy kontrolya v mashinostroyenii [Automatic control systems in mechanical engineering]. Rossiya Molodaya. Kemerovo, 2016. P. 234. EDN: WKJZRV. (In Russ.).

9. Suntsov A. P. Sovremennyye sredstva kontrolya izdeliy v mashinostroyenii [State-of-the-art tools for product control in mechanical engineering]. Studencheskiy Nauchnyy Forum. Ekaterinburg, 2018. P. 1–6. (In Russ.).

10. Xie T., Cheng X., Wang X. [et al.]. Cut-Thumbnail: A Novel Data Augmentation for Convolutional Neural Network. Proc. of the 29th ACM Intern. Conf. on Multimedia, Virtual Event. 2021. P. 1627–1635. DOI: 10.1145/3474085.3475302.

11. Terven J., Cordova-Esparza D.-M., Romero-Gonzбlez J.A. A comprehensive review of YOLO architectures in computer vision: From YOLOv1 to YOLOv8 and YOLO-NAS. Machine Learning and Knowledge Extraction. 2023. No. 5. P. 1680–1716. DOI: 10.3390/make5040083.

12. Machine defects computer vision model. URL: https://universe.roboflow.com/j-component-ml/machine-defects (accessed: 27.10.2025).

13. Defects computer vision model. URL: https://universe.roboflow.com/mai-ldivt/defects-lavkc (accessed: 27.10.2025).

14. Welding defect detection computer vision model. URL: https://universe.roboflow.com/final-year-project-kswbt/weldingdefect-detection (accessed: 27.10.2025).

15. CaliperTop_backup computer vision dataset. URL: https://universe.roboflow.com/imagineai/calipertop_backup (accessed: 27.10.2025).

16. Jocher G. Ultralytics YOLOv8. URL: https://github.com/ultralytics/ultralytics (accessed: 27.10.2025).

17. Kirillov A., Mintun E., Ravi N. [et al.]. Segment Anything. 2023. DOI: 10.48550/arXiv.2304.02643.

18. Lewis P., Perez E., Piktus A. [et al.]. Retrieval-augmented generation for knowledge-intensive NLP tasks. Advances in Neural Information Processing Systems. 2020. Vol. 33. P. 9459–9474. DOI: 10.48550/arXiv.2005.11401.

19. GOST R ISO 5817–2009. Svarka. Svarnyye soyedineniya iz stali, nikelya, titana i ikh splavov, poluchennyye svarkoy plavleniyem. Urovni kachestva v zavisimosti ot defektov [Welding. Fusion-welded joints in steel, nickel, titanium and their alloys (beam welding excluded). Quality levels for imperfections]. Moscow, 2010. 31 p. (In Russ.).