{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,23]],"date-time":"2026-10-23T21:02:01Z","timestamp":1792789321214,"version":"4.5.0"},"reference-count":24,"publisher":"MDPI AG","issue":"9","license":[{"start":{"date-parts":[[2026,9,19]],"date-time":"2026-09-19T00:00:00Z","timestamp":1789776000000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"name":"Woosong University","id":[{"id":"https:\/\/ror.org\/02srty072","id-type":"ROR","asserted-by":"crossref"}]},{"id":[{"id":"https:\/\/ror.org\/02srty072","id-type":"ROR","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["Computation"],"abstract":"<jats:p>Large language model (LLM) agents can perform prognostics and health management (PHM) tasks such as bearing fault diagnosis, but most published systems rely on large, paid, cloud-hosted models that a small or medium enterprise (SME) cannot self-host. The diagnostic accuracy that can be achieved on free, offline, commodity hardware is a practical question. Using an execution-based evaluation (Pass@1, Pass all 3, macro F1), this paper benchmarks eight small, publicly accessible, locally runnable LLMs (1B\u20139B parameters, via Ollama) on vibration-derived features for three-class bearing fault diagnosis (Healthy, Outer race, and Inner race). The proposed work is evaluated on three independent datasets, Paderborn, CWRU, and HUST, across a 0\u20135-shot ablation and at two decoding temperatures to separate accuracy from reliability. The findings replicate across all three datasets, namely, a model-capability gate that only models near 7B parameters and above clears the majority-class floor; few-shot prompting is non-monotonic; accuracy and reliability are distinct axes; and a single free model, gemma2:9b (5.4 GB), is the most accurate and among the most reliable, with no task-specific training. We further propose ensemble agreement gating, a training-free reliability rule that withholds predictions when several free local models disagree, raising accuracy on the answered subset (e.g., 0.77 \u2192 0.87 on Paderborn). In a head-to-head on identical prompts and data, the free local models match a current frontier model in the settings tested at zero cost and fully offline, and we characterize where such a training-free local approach is and is not appropriate for resource-constrained operators of rotating machinery. On the same features, however, a simple supervised baseline such as logistic regression, and even an untrained physics rule, outperform all eight LLMs, so we present this as a cautionary benchmark: the contribution of the free local approach is training-free deployment and a reliability gating rule, not classification accuracy.<\/jats:p>","DOI":"10.3390\/computation14090220","type":"journal-article","created":{"date-parts":[[2026,9,21]],"date-time":"2026-09-21T11:05:53Z","timestamp":1789988753000},"page":"220","update-policy":"https:\/\/doi.org\/10.3390\/mdpi_crossmark_policy","source":"Crossref","is-referenced-by-count":0,"title":["A Study on Locally Runnable Large Language Models for Bearing Fault Diagnosis"],"prefix":"10.3390","volume":"14","author":[{"ORCID":"https:\/\/orcid.org\/0009-0002-4697-660X","authenticated-orcid":false,"given":"Mehadi Hasan","family":"Shawon","sequence":"first","affiliation":[{"name":"Department of AI and Big Data, Woosong University, Daejeon 34606, Republic of Korea"}],"role":[{"vocabulary":"crossref","role":"author"},{"role":"conceptualization","vocabulary":"credit"},{"role":"methodology","vocabulary":"credit"},{"role":"software","vocabulary":"credit"},{"role":"validation","vocabulary":"credit"},{"role":"writing-original-draft","vocabulary":"credit"},{"role":"data-curation","vocabulary":"credit"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-6353-6302","authenticated-orcid":false,"given":"Prashant","family":"Kumar","sequence":"additional","affiliation":[{"name":"Department of AI and Big Data, Woosong University, Daejeon 34606, Republic of Korea"}],"role":[{"vocabulary":"crossref","role":"author"},{"role":"supervision","vocabulary":"credit"},{"role":"project-administration","vocabulary":"credit"},{"role":"funding-acquisition","vocabulary":"credit"},{"role":"writing-review-editing","vocabulary":"credit"}]}],"member":"1968","published-online":{"date-parts":[[2026,9,19]]},"reference":[{"key":"ref_1","doi-asserted-by":"crossref","first-page":"1929","DOI":"10.1007\/s11831-020-09446-w","article-title":"Review on Machine Learning Algorithm Based Fault Detection in Induction Motors","volume":"28","author":"Kumar","year":"2021","journal-title":"Arch. Comput. Methods Eng."},{"key":"ref_2","doi-asserted-by":"crossref","first-page":"106899","DOI":"10.1016\/j.rineng.2025.106899","article-title":"Intelligent Fault Diagnosis in Rolling Element Bearings: Combining Envelope Spectrum and Spectral Kurtosis for Enhanced Detection","volume":"27","author":"Rohan","year":"2025","journal-title":"Results Eng."},{"key":"ref_3","doi-asserted-by":"crossref","first-page":"93155","DOI":"10.1109\/ACCESS.2020.2990528","article-title":"Bearing Fault Detection and Diagnosis Using Case Western Reserve University Dataset With Deep Learning Approaches: A Review","volume":"8","author":"Neupane","year":"2020","journal-title":"IEEE Access"},{"key":"ref_4","unstructured":"Das, A., and Patel, D. (2026). PHMForge: A Scenario-Driven Agentic Benchmark for Industrial Asset Lifecycle Maintenance. arXiv."},{"key":"ref_5","doi-asserted-by":"crossref","unstructured":"Huang, K., and Chen, Q. (2025). Motor Fault Diagnosis and Predictive Maintenance Based on a Fine-Tuned Qwen2.5-7B Model. Processes, 13.","DOI":"10.3390\/pr13072051"},{"key":"ref_6","doi-asserted-by":"crossref","first-page":"109850","DOI":"10.1016\/j.ress.2023.109850","article-title":"ChatGPT-like Large-Scale Foundation Models for Prognostics and Health Management: A Survey and Roadmaps","volume":"243","author":"Li","year":"2024","journal-title":"Reliab. Eng. Syst. Saf."},{"key":"ref_7","doi-asserted-by":"crossref","unstructured":"Bao, P., Yi, W., Zhu, Y., Shen, Y., and Peng, H. (2025). A Spectral Interpretable Bearing Fault Diagnosis Framework Powered by Large Language Models. Sensors, 25.","DOI":"10.3390\/s25123822"},{"key":"ref_8","doi-asserted-by":"crossref","first-page":"112544","DOI":"10.1016\/j.engappai.2025.112544","article-title":"A Unified Rotating Machinery Health Management Framework Leveraging Large Language Models for Diverse Components, Conditions, and Tasks","volume":"162","author":"Peng","year":"2025","journal-title":"Eng. Appl. Artif. Intell."},{"key":"ref_9","doi-asserted-by":"crossref","unstructured":"Di Maggio, L.G. (2025). Toward Autonomous LLM-Based AI Agents for Predictive Maintenance: State of the Art, Challenges, and Future Perspectives. Appl. Sci., 15.","DOI":"10.3390\/app152111515"},{"key":"ref_10","doi-asserted-by":"crossref","unstructured":"Lessmeier, C., Kimotho, J.K., Zimmer, D., and Sextro, W. (2016, January 5\u20138). Condition Monitoring of Bearing Damage in Electromechanical Drive Systems by Using Motor Current Signals of Electric Motors: A Benchmark Data Set for Data-Driven Classification. Proceedings of the PHM Society European Conference, Bilbao, Spain.","DOI":"10.36001\/phme.2016.v3i1.1577"},{"key":"ref_11","doi-asserted-by":"crossref","first-page":"100","DOI":"10.1016\/j.ymssp.2015.04.021","article-title":"Rolling Element Bearing Diagnostics Using the Case Western Reserve University Data: A Benchmark Study","volume":"64\u201365","author":"Smith","year":"2015","journal-title":"Mech. Syst. Signal Process."},{"key":"ref_12","doi-asserted-by":"crossref","unstructured":"Thuan, N.D., and Hong, H.S. (2023). HUST Bearing: A Practical Dataset for Ball Bearing Fault Diagnosis. BMC Res. Notes, 16.","DOI":"10.1186\/s13104-023-06400-4"},{"key":"ref_13","unstructured":"Chen, M., Tworek, J., and Jun, H. (2021). Evaluating Large Language Models Trained on Code. arXiv."},{"key":"ref_14","doi-asserted-by":"crossref","first-page":"109528","DOI":"10.1016\/j.compchemeng.2025.109528","article-title":"Explainable Neural Network Meets Graph Neural Network: Recent Advances in Process Fault Detection and Diagnosis","volume":"206","author":"Liu","year":"2026","journal-title":"Comput. Chem. Eng."},{"key":"ref_15","first-page":"1","article-title":"AKGNN: When Adaptive Graph Neural Network Meets Kolmogorov\u2013Arnold Network for Industrial Soft Sensors","volume":"74","author":"Yang","year":"2025","journal-title":"IEEE Trans. Instrum. Meas."},{"key":"ref_16","doi-asserted-by":"crossref","first-page":"107187","DOI":"10.1016\/j.conengprac.2026.107187","article-title":"CLARI: Constrained Reflective LLM Inference for Few-Shot Soft Sensing under Operating-Condition Drift","volume":"177","author":"Yang","year":"2026","journal-title":"Control Eng. Pract."},{"key":"ref_17","doi-asserted-by":"crossref","first-page":"117698","DOI":"10.1016\/j.eswa.2022.117698","article-title":"Fault Diagnosis in Wind Turbines Based on ANFIS and Takagi\u2013Sugeno Interval Observers","volume":"206","author":"Puig","year":"2022","journal-title":"Expert Syst. Appl."},{"key":"ref_18","doi-asserted-by":"crossref","first-page":"106940","DOI":"10.1016\/j.conengprac.2026.106940","article-title":"Wind Turbine Fault Diagnosis Using Structural Analysis and Optimized-Rectangular GPR Interval Estimation","volume":"172","author":"Puig","year":"2026","journal-title":"Control Eng. Pract."},{"key":"ref_19","unstructured":"Meta Llama Team (2024). The Llama 3 Herd of Models. arXiv."},{"key":"ref_20","unstructured":"Gemma Team (2024). Gemma 2: Improving Open Language Models at a Practical Size. arXiv."},{"key":"ref_21","unstructured":"Qwen Team (2024). Qwen2.5 Technical Report. arXiv."},{"key":"ref_22","unstructured":"Abdin, M., Aneja, J., Awadalla, H., Awadallah, A., Awan, A.A., Bach, N., Bahree, A., Bakhtiari, A., Bao, J., and Behl, H. (2024). Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone. arXiv."},{"key":"ref_23","unstructured":"Jiang, A.Q., Sablayrolles, A., Mensch, A., Bamford, C., Chaplot, D.S., de las Casas, D., Bressand, F., Lengyel, G., Lample, G., and Saulnier, L. (2023). Mistral 7B. arXiv."},{"key":"ref_24","unstructured":"(2026, August 10). Ollama: Get Up and Running with Large Language Models Locally. Available online: https:\/\/ollama.com."}],"container-title":["Computation"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.mdpi.com\/2079-3197\/14\/9\/220\/pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,9,22]],"date-time":"2026-09-22T04:24:27Z","timestamp":1790051067000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.mdpi.com\/2079-3197\/14\/9\/220"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,9,19]]},"references-count":24,"journal-issue":{"issue":"9","published-online":{"date-parts":[[2026,9]]}},"alternative-id":["computation14090220"],"URL":"https:\/\/doi.org\/10.3390\/computation14090220","relation":{},"ISSN":["2079-3197"],"issn-type":[{"value":"2079-3197","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,9,19]]}}}