rotorvibe 0.1.0a1__tar.gz

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (173) hide show
  1. rotorvibe-0.1.0a1/CHANGELOG.md +24 -0
  2. rotorvibe-0.1.0a1/CITATION.cff +25 -0
  3. rotorvibe-0.1.0a1/LICENSE +21 -0
  4. rotorvibe-0.1.0a1/MANIFEST.in +15 -0
  5. rotorvibe-0.1.0a1/NOTICE +12 -0
  6. rotorvibe-0.1.0a1/PKG-INFO +234 -0
  7. rotorvibe-0.1.0a1/README.ko.md +70 -0
  8. rotorvibe-0.1.0a1/README.md +206 -0
  9. rotorvibe-0.1.0a1/docs/ALGORITHMS.md +142 -0
  10. rotorvibe-0.1.0a1/docs/CLASSIFIER_IMPROVEMENT.rst +146 -0
  11. rotorvibe-0.1.0a1/docs/CONDITIONED_IMPROVEMENT.rst +200 -0
  12. rotorvibe-0.1.0a1/docs/CONTINUATION.md +152 -0
  13. rotorvibe-0.1.0a1/docs/DEVELOPMENT_HISTORY.md +235 -0
  14. rotorvibe-0.1.0a1/docs/EXTERNAL_BLIND_VALIDATION.rst +173 -0
  15. rotorvibe-0.1.0a1/docs/FAULT_DISCRIMINATION_IMPROVEMENT.rst +197 -0
  16. rotorvibe-0.1.0a1/docs/FINAL_CROSS_SPEED_VALIDATION.rst +150 -0
  17. rotorvibe-0.1.0a1/docs/ISO_API.txt +14 -0
  18. rotorvibe-0.1.0a1/docs/MODULE_CATALOG.md +241 -0
  19. rotorvibe-0.1.0a1/docs/Python_API.txt +167 -0
  20. rotorvibe-0.1.0a1/docs/RELEASE_NOTES.md +18 -0
  21. rotorvibe-0.1.0a1/docs/SOURCES_AND_RIGHTS.md +92 -0
  22. rotorvibe-0.1.0a1/docs/TWO_STAGE_AND_DATA.rst +205 -0
  23. rotorvibe-0.1.0a1/docs/USER_GUIDE.md +214 -0
  24. rotorvibe-0.1.0a1/docs/VALIDATION.rst +96 -0
  25. rotorvibe-0.1.0a1/examples/fault_discrimination_composition.py +17 -0
  26. rotorvibe-0.1.0a1/examples/iso_composition.py +10 -0
  27. rotorvibe-0.1.0a1/examples/python_composition.py +33 -0
  28. rotorvibe-0.1.0a1/examples/two_stage_composition.py +43 -0
  29. rotorvibe-0.1.0a1/pyproject.toml +36 -0
  30. rotorvibe-0.1.0a1/requirements.txt +3 -0
  31. rotorvibe-0.1.0a1/research/README.md +33 -0
  32. rotorvibe-0.1.0a1/research/classifier_improvement/development.json +405 -0
  33. rotorvibe-0.1.0a1/research/classifier_improvement/diagnosis.json +235 -0
  34. rotorvibe-0.1.0a1/research/classifier_improvement/final_test.json +70 -0
  35. rotorvibe-0.1.0a1/research/classifier_improvement/integration.json +26 -0
  36. rotorvibe-0.1.0a1/research/collection_manifests/condition_improvement_matched_normal_plan.json +113 -0
  37. rotorvibe-0.1.0a1/research/collection_manifests/condition_improvement_record_plan.json +869 -0
  38. rotorvibe-0.1.0a1/research/collection_manifests/external_blind_plan.json +706 -0
  39. rotorvibe-0.1.0a1/research/collection_manifests/fault_discrimination_plan.json +425 -0
  40. rotorvibe-0.1.0a1/research/condition_improvement/api_integration.json +17 -0
  41. rotorvibe-0.1.0a1/research/condition_improvement/matched_development.json +566 -0
  42. rotorvibe-0.1.0a1/research/condition_improvement/new_final_audit.json +25 -0
  43. rotorvibe-0.1.0a1/research/condition_improvement/new_final_execution.json +10 -0
  44. rotorvibe-0.1.0a1/research/condition_improvement/new_final_predictions.jsonl +320 -0
  45. rotorvibe-0.1.0a1/research/condition_improvement/new_final_protocol.json +123 -0
  46. rotorvibe-0.1.0a1/research/condition_improvement/new_final_results.json +169 -0
  47. rotorvibe-0.1.0a1/research/condition_improvement/qualification_decision.json +7 -0
  48. rotorvibe-0.1.0a1/research/data_sources.json +39 -0
  49. rotorvibe-0.1.0a1/research/evidence_copy_receipts.json +254 -0
  50. rotorvibe-0.1.0a1/research/exposure_registry.json +181 -0
  51. rotorvibe-0.1.0a1/research/external_blind/access.jsonl +43 -0
  52. rotorvibe-0.1.0a1/research/external_blind/audit.json +12 -0
  53. rotorvibe-0.1.0a1/research/external_blind/evaluation_freeze.json +111 -0
  54. rotorvibe-0.1.0a1/research/external_blind/execution.json +8 -0
  55. rotorvibe-0.1.0a1/research/external_blind/predictions.jsonl +580 -0
  56. rotorvibe-0.1.0a1/research/external_blind/protocol.json +66 -0
  57. rotorvibe-0.1.0a1/research/external_blind/qualification_decision.json +24 -0
  58. rotorvibe-0.1.0a1/research/external_blind/results.json +2487 -0
  59. rotorvibe-0.1.0a1/research/fault_discrimination/api_integration.json +26 -0
  60. rotorvibe-0.1.0a1/research/fault_discrimination/audit.json +12 -0
  61. rotorvibe-0.1.0a1/research/fault_discrimination/development.json +790 -0
  62. rotorvibe-0.1.0a1/research/fault_discrimination/execution.json +8 -0
  63. rotorvibe-0.1.0a1/research/fault_discrimination/final_access.jsonl +7 -0
  64. rotorvibe-0.1.0a1/research/fault_discrimination/final_predictions.jsonl +560 -0
  65. rotorvibe-0.1.0a1/research/fault_discrimination/final_protocol.json +161 -0
  66. rotorvibe-0.1.0a1/research/fault_discrimination/final_results.json +388 -0
  67. rotorvibe-0.1.0a1/research/fault_discrimination/previous_failure_diagnosis.json +75 -0
  68. rotorvibe-0.1.0a1/research/fault_discrimination/qualification_decision.json +9 -0
  69. rotorvibe-0.1.0a1/research/final_cross_speed/audit_and_diagnosis.json +73 -0
  70. rotorvibe-0.1.0a1/research/final_cross_speed/execution.json +11 -0
  71. rotorvibe-0.1.0a1/research/final_cross_speed/final_test.json +981 -0
  72. rotorvibe-0.1.0a1/research/final_cross_speed/protocol.json +57 -0
  73. rotorvibe-0.1.0a1/research/final_cross_speed/qualification_decision.json +11 -0
  74. rotorvibe-0.1.0a1/research/final_cross_speed/target_predictions.jsonl +1178 -0
  75. rotorvibe-0.1.0a1/research/module_catalog.json +266 -0
  76. rotorvibe-0.1.0a1/research/prose_overlap_audit.json +8 -0
  77. rotorvibe-0.1.0a1/research/public_file_delta.json +1010 -0
  78. rotorvibe-0.1.0a1/research/public_release_exclusions.json +27 -0
  79. rotorvibe-0.1.0a1/research/public_verification.json +26 -0
  80. rotorvibe-0.1.0a1/rotorvibe/__init__.py +12 -0
  81. rotorvibe-0.1.0a1/rotorvibe.egg-info/PKG-INFO +234 -0
  82. rotorvibe-0.1.0a1/rotorvibe.egg-info/SOURCES.txt +171 -0
  83. rotorvibe-0.1.0a1/rotorvibe.egg-info/dependency_links.txt +1 -0
  84. rotorvibe-0.1.0a1/rotorvibe.egg-info/entry_points.txt +3 -0
  85. rotorvibe-0.1.0a1/rotorvibe.egg-info/requires.txt +8 -0
  86. rotorvibe-0.1.0a1/rotorvibe.egg-info/top_level.txt +2 -0
  87. rotorvibe-0.1.0a1/run.py +6 -0
  88. rotorvibe-0.1.0a1/scripts/audit_conditioned_final.py +40 -0
  89. rotorvibe-0.1.0a1/scripts/audit_final_cross_speed.py +84 -0
  90. rotorvibe-0.1.0a1/scripts/bearing_frequency.py +43 -0
  91. rotorvibe-0.1.0a1/scripts/check_uos_overlap.py +38 -0
  92. rotorvibe-0.1.0a1/scripts/collect_condition_data.py +73 -0
  93. rotorvibe-0.1.0a1/scripts/collect_external_blind.py +85 -0
  94. rotorvibe-0.1.0a1/scripts/collect_fault_discrimination.py +40 -0
  95. rotorvibe-0.1.0a1/scripts/collect_gap_materials.py +112 -0
  96. rotorvibe-0.1.0a1/scripts/collect_matched_normals.py +45 -0
  97. rotorvibe-0.1.0a1/scripts/collect_normal_independent.py +116 -0
  98. rotorvibe-0.1.0a1/scripts/collect_public_data.py +169 -0
  99. rotorvibe-0.1.0a1/scripts/develop_conditioned.py +108 -0
  100. rotorvibe-0.1.0a1/scripts/develop_fault_discrimination.py +77 -0
  101. rotorvibe-0.1.0a1/scripts/develop_matched_conditioned.py +72 -0
  102. rotorvibe-0.1.0a1/scripts/diagnose_classifier.py +45 -0
  103. rotorvibe-0.1.0a1/scripts/evaluate_physical_records.py +50 -0
  104. rotorvibe-0.1.0a1/scripts/external_blind_metrics.py +51 -0
  105. rotorvibe-0.1.0a1/scripts/finalize_gap_pack.py +86 -0
  106. rotorvibe-0.1.0a1/scripts/finalize_validation_pack.py +59 -0
  107. rotorvibe-0.1.0a1/scripts/improve_classifier.py +120 -0
  108. rotorvibe-0.1.0a1/scripts/prepare_validation_data.py +229 -0
  109. rotorvibe-0.1.0a1/scripts/send_measured_npz.py +37 -0
  110. rotorvibe-0.1.0a1/scripts/summarize_two_stage.py +81 -0
  111. rotorvibe-0.1.0a1/scripts/validate_gap_data.py +258 -0
  112. rotorvibe-0.1.0a1/scripts/validate_normal_independent.py +102 -0
  113. rotorvibe-0.1.0a1/scripts/verify_classifier_integration.py +63 -0
  114. rotorvibe-0.1.0a1/scripts/verify_conditioned_api.py +36 -0
  115. rotorvibe-0.1.0a1/scripts/verify_conditioned_final.py +123 -0
  116. rotorvibe-0.1.0a1/scripts/verify_external_blind.py +135 -0
  117. rotorvibe-0.1.0a1/scripts/verify_external_source.py +75 -0
  118. rotorvibe-0.1.0a1/scripts/verify_fault_discrimination.py +131 -0
  119. rotorvibe-0.1.0a1/scripts/verify_fault_discrimination_api.py +27 -0
  120. rotorvibe-0.1.0a1/scripts/verify_final_cross_speed.py +162 -0
  121. rotorvibe-0.1.0a1/scripts/verify_implementation.py +336 -0
  122. rotorvibe-0.1.0a1/scripts/verify_models.py +72 -0
  123. rotorvibe-0.1.0a1/scripts/verify_physics.py +277 -0
  124. rotorvibe-0.1.0a1/scripts/verify_stream_performance.py +42 -0
  125. rotorvibe-0.1.0a1/scripts/verify_two_stage.py +97 -0
  126. rotorvibe-0.1.0a1/scripts/verify_two_stage_service.py +54 -0
  127. rotorvibe-0.1.0a1/setup.cfg +4 -0
  128. rotorvibe-0.1.0a1/start.cmd +11 -0
  129. rotorvibe-0.1.0a1/tests/test_classifier.py +77 -0
  130. rotorvibe-0.1.0a1/tests/test_condition_data.py +30 -0
  131. rotorvibe-0.1.0a1/tests/test_conditioned.py +93 -0
  132. rotorvibe-0.1.0a1/tests/test_external_blind_metrics.py +30 -0
  133. rotorvibe-0.1.0a1/tests/test_fault_discrimination.py +54 -0
  134. rotorvibe-0.1.0a1/tests/test_independent_data.py +66 -0
  135. rotorvibe-0.1.0a1/tests/test_public_rules.py +26 -0
  136. rotorvibe-0.1.0a1/tests/test_release_contracts.py +131 -0
  137. rotorvibe-0.1.0a1/tests/test_service_contracts.py +150 -0
  138. rotorvibe-0.1.0a1/tests/test_two_stage.py +132 -0
  139. rotorvibe-0.1.0a1/tools/audit_release.py +49 -0
  140. rotorvibe-0.1.0a1/vibration/__init__.py +21 -0
  141. rotorvibe-0.1.0a1/vibration/adapters.py +48 -0
  142. rotorvibe-0.1.0a1/vibration/alarm.py +39 -0
  143. rotorvibe-0.1.0a1/vibration/baseline.py +88 -0
  144. rotorvibe-0.1.0a1/vibration/bearings.py +41 -0
  145. rotorvibe-0.1.0a1/vibration/bootstrap.py +14 -0
  146. rotorvibe-0.1.0a1/vibration/classifier.py +85 -0
  147. rotorvibe-0.1.0a1/vibration/client.py +58 -0
  148. rotorvibe-0.1.0a1/vibration/condition_data.py +74 -0
  149. rotorvibe-0.1.0a1/vibration/conditioned.py +191 -0
  150. rotorvibe-0.1.0a1/vibration/config.py +141 -0
  151. rotorvibe-0.1.0a1/vibration/contracts.py +101 -0
  152. rotorvibe-0.1.0a1/vibration/core.py +127 -0
  153. rotorvibe-0.1.0a1/vibration/datasets.py +204 -0
  154. rotorvibe-0.1.0a1/vibration/dsp.py +173 -0
  155. rotorvibe-0.1.0a1/vibration/engine.py +359 -0
  156. rotorvibe-0.1.0a1/vibration/fault_discrimination.py +118 -0
  157. rotorvibe-0.1.0a1/vibration/geometry.py +39 -0
  158. rotorvibe-0.1.0a1/vibration/independent_data.py +115 -0
  159. rotorvibe-0.1.0a1/vibration/iso/__init__.py +12 -0
  160. rotorvibe-0.1.0a1/vibration/iso/contracts.py +180 -0
  161. rotorvibe-0.1.0a1/vibration/iso/evaluation.py +51 -0
  162. rotorvibe-0.1.0a1/vibration/iso/measurements.py +80 -0
  163. rotorvibe-0.1.0a1/vibration/iso/stage.py +25 -0
  164. rotorvibe-0.1.0a1/vibration/models.py +167 -0
  165. rotorvibe-0.1.0a1/vibration/ports.py +20 -0
  166. rotorvibe-0.1.0a1/vibration/rotor_features.py +79 -0
  167. rotorvibe-0.1.0a1/vibration/server.py +272 -0
  168. rotorvibe-0.1.0a1/vibration/stages.py +30 -0
  169. rotorvibe-0.1.0a1/vibration/storage.py +88 -0
  170. rotorvibe-0.1.0a1/vibration/two_stage.py +159 -0
  171. rotorvibe-0.1.0a1/vibration/ui/app.js +72 -0
  172. rotorvibe-0.1.0a1/vibration/ui/index.html +35 -0
  173. rotorvibe-0.1.0a1/vibration/ui/style.css +1 -0
@@ -0,0 +1,24 @@
1
+ # Changelog
2
+
3
+ ## 0.1.0a1 — 2026-11-04
4
+
5
+ First public alpha of RotorVibe, released from an existing local research workspace.
6
+
7
+ - Publish the synchronous NumPy analysis core and stream contracts, bearing geometry,
8
+ replaceable providers, stages/sinks and optional local monitoring application.
9
+ - Retain Gaussian, normal-first, RPM-conditioned and morphology diagnostic algorithms
10
+ as experimental APIs with explicit applicability and abstention semantics.
11
+ - Add the `rotorvibe` facade while preserving the `vibration` compatibility namespace.
12
+ - Document the complete experimental sequence, including failed cross-speed,
13
+ cross-bearing and independent-rig validation. No universal accuracy claim is made.
14
+ - Publish authored metric/protocol/prediction/access evidence without raw measurements
15
+ or trained rig-specific profiles.
16
+ - Exclude private standard tables/selectors/fixtures and replace the public rule route
17
+ with explicit caller-supplied limits and synthetic verification examples.
18
+ - Supply an original README, operating guide, algorithm notes, module catalog,
19
+ development record, continuation handoff, rights inventory and citation metadata.
20
+ - MIT applies to original code and authored documentation; external source terms remain.
21
+
22
+ The private local studies preceded public Git history. See the development record
23
+ for that reconstructed chronology rather than interpreting this one release commit
24
+ as a complete history of earlier experimentation.
@@ -0,0 +1,25 @@
1
+ cff-version: 1.2.0
2
+ message: "Please cite the software version and the relevant experiment protocol separately."
3
+ type: software
4
+ title: "RotorVibe: Composable Rotating-Machine Vibration Analysis"
5
+ version: 0.1.0a1
6
+ date-released: 2026-11-04
7
+ authors:
8
+ - family-names: Lee
9
+ given-names: Ugwiyeon
10
+ repository-code: "https://github.com/lyullee/rotorvibe"
11
+ url: "https://github.com/lyullee/rotorvibe"
12
+ doi: 10.5281/zenodo.23937890
13
+ license: MIT
14
+ abstract: >-
15
+ Python vibration analysis and composable diagnostic research APIs with explicit
16
+ signal contracts, normal-reference scope, original-file evaluation evidence and
17
+ documented transfer failures. The public release omits standards lookup tables,
18
+ external raw data and trained rig-specific profiles. It is not field-certified
19
+ fault diagnosis or an automatic protection system.
20
+ keywords:
21
+ - vibration analysis
22
+ - rotating machinery
23
+ - bearing kinematics
24
+ - condition monitoring
25
+ - reproducible research
@@ -0,0 +1,21 @@
1
+ MIT License
2
+
3
+ Copyright (c) 2026 Ugwiyeon Lee
4
+
5
+ Permission is hereby granted, free of charge, to any person obtaining a copy
6
+ of this software and associated documentation files (the "Software"), to deal
7
+ in the Software without restriction, including without limitation the rights
8
+ to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
9
+ copies of the Software, and to permit persons to whom the Software is
10
+ furnished to do so, subject to the following conditions:
11
+
12
+ The above copyright notice and this permission notice shall be included in all
13
+ copies or substantial portions of the Software.
14
+
15
+ THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
16
+ IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
17
+ FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
18
+ AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
19
+ LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
20
+ OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
21
+ SOFTWARE.
@@ -0,0 +1,15 @@
1
+ include README.md README.ko.md LICENSE NOTICE CITATION.cff CHANGELOG.md requirements.txt run.py start.cmd
2
+ recursive-include vibration *.py *.html *.css *.js
3
+ recursive-include rotorvibe *.py
4
+ recursive-include tests *.py
5
+ recursive-include examples *.py
6
+ recursive-include scripts *.py
7
+ recursive-include tools *.py
8
+ recursive-include docs *.md *.rst *.txt
9
+ recursive-include research *.json *.jsonl *.md
10
+ prune docs/integration/upstream
11
+ prune data
12
+ prune output
13
+ prune tmp
14
+ prune runtime
15
+ global-exclude *.pdf *.PDF *.npy *.npz *.mat *.csv *.lvm *.db *.sqlite* *.pyc
@@ -0,0 +1,12 @@
1
+ RotorVibe original software and original documentation: Copyright 2026 Ugwiyeon Lee, MIT.
2
+
3
+ MIT does not relicense any external data, standards, papers, catalogs or code.
4
+ Third-party raw measurements, trained dataset-specific profiles, source PDFs,
5
+ standards-derived lookup tables and external repository snapshots are not shipped.
6
+ Dataset authors, identifiers, separate terms and acquisition requirements are
7
+ listed in docs/SOURCES_AND_RIGHTS.md and research/data_sources.json.
8
+
9
+ The vocabulary of signal processing, standard identifiers and citation metadata
10
+ is retained to identify methods accurately. License text and mandatory notices
11
+ are preserved verbatim; substantive external explanatory prose is not reproduced.
12
+ No ISO, KS, instrument-vendor or dataset-owner endorsement is asserted.
@@ -0,0 +1,234 @@
1
+ Metadata-Version: 2.4
2
+ Name: rotorvibe
3
+ Version: 0.1.0a1
4
+ Summary: Composable rotating-machine vibration analysis with explicit evidence and applicability limits
5
+ Author: Ugwiyeon Lee
6
+ License-Expression: MIT
7
+ Project-URL: Homepage, https://github.com/lyullee/rotorvibe
8
+ Project-URL: Repository, https://github.com/lyullee/rotorvibe
9
+ Project-URL: Documentation, https://github.com/lyullee/rotorvibe/blob/main/docs/USER_GUIDE.md
10
+ Project-URL: Issues, https://github.com/lyullee/rotorvibe/issues
11
+ Project-URL: Archive, https://doi.org/10.5281/zenodo.23937890
12
+ Keywords: vibration,rotating machinery,signal processing,bearing,condition monitoring
13
+ Classifier: Development Status :: 3 - Alpha
14
+ Classifier: Programming Language :: Python :: 3
15
+ Classifier: Topic :: Scientific/Engineering
16
+ Classifier: Operating System :: OS Independent
17
+ Requires-Python: >=3.11
18
+ Description-Content-Type: text/markdown
19
+ License-File: LICENSE
20
+ Requires-Dist: numpy<3,>=2
21
+ Requires-Dist: scipy<2,>=1.13
22
+ Provides-Extra: verification
23
+ Requires-Dist: build<2,>=1.2; extra == "verification"
24
+ Requires-Dist: setuptools>=77; extra == "verification"
25
+ Requires-Dist: wheel; extra == "verification"
26
+ Requires-Dist: twine<7,>=6; extra == "verification"
27
+ Dynamic: license-file
28
+
29
+ # RotorVibe
30
+
31
+ **Composable rotating-machine vibration analysis for Python.**
32
+
33
+ RotorVibe turns calibrated waveform blocks into spectra, bearing-frequency markers,
34
+ condition-relative evidence and inspectable diagnostic candidates. Use the NumPy API
35
+ inside your own application; the local web dashboard is optional.
36
+
37
+ **Alpha research release, version 0.1.0a1.** Signal calculations and bounded software
38
+ checks are documented. Fault classifiers have **not** qualified for unattended field
39
+ diagnosis: new-bearing and independent-rig transfer tests failed. No remaining-life
40
+ prediction, automatic shutdown authority or standards certification is supplied.
41
+
42
+ [한국어 설명](README.ko.md) · [User guide](docs/USER_GUIDE.md) ·
43
+ [Algorithm notes](docs/ALGORITHMS.md) · [Development record](docs/DEVELOPMENT_HISTORY.md) ·
44
+ [Continue this research](docs/CONTINUATION.md) · [Source/rights policy](docs/SOURCES_AND_RIGHTS.md)
45
+
46
+ ## Install
47
+
48
+ Python 3.11+ with NumPy 2.x and SciPy 1.13+ is required.
49
+
50
+ ```console
51
+ python -m pip install rotorvibe==0.1.0a1
52
+ ```
53
+
54
+ For an editable checkout and verification tools:
55
+
56
+ ```console
57
+ git clone https://github.com/lyullee/rotorvibe.git
58
+ cd rotorvibe
59
+ python -m pip install -e ".[verification]"
60
+ python -m unittest discover -s tests -v
61
+ ```
62
+
63
+ The public facade is `rotorvibe`; the earlier `vibration` namespace remains available
64
+ so existing Python integrations can keep their imports. Raw experimental data and
65
+ trained rig-specific profiles are acquired separately, not installed by pip.
66
+
67
+ ## A complete first calculation
68
+
69
+ This generated waveform verifies the API; it is not a measured fault example.
70
+
71
+ ```python
72
+ import numpy as np
73
+ from rotorvibe import AnalysisConfig, SignalBlock, VibrationAnalyzer
74
+
75
+ fs = 12000
76
+ t = np.arange(fs) / fs
77
+ samples = 0.20 * np.sin(2 * np.pi * 20 * t)
78
+
79
+ analyzer = VibrationAnalyzer(AnalysisConfig(
80
+ sample_rate_hz=fs, channels=1, unit="m/s^2",
81
+ window_seconds=1, hop_seconds=0.25, rpm=1200,
82
+ ))
83
+ result = analyzer.analyze(SignalBlock(
84
+ samples, fs, unit="m/s^2", rpm=1200,
85
+ channels=("drive_radial",), clock_id="daq_elapsed", episode_id="run-001",
86
+ metadata={"asset_id": "motor-01", "condition_id": "rated_load"},
87
+ ))
88
+ print(result.valid, result.metrics)
89
+ serializable = result.to_dict(include_features=False)
90
+ ```
91
+
92
+ `samples` may also be an `N × C` array. A one-dimensional input represents one
93
+ channel. Supply the physical unit, channel ordering, time basis and actual asset
94
+ identity; a numeric array alone cannot establish sensor calibration or installation.
95
+
96
+ ## Connect modules directly
97
+
98
+ The host controls acquisition, scheduling, storage and decisions. An analyzer instance
99
+ owns one stream; application callbacks receive the results synchronously.
100
+
101
+ ```python
102
+ from rotorvibe import AnalysisConfig, VibrationAnalyzer
103
+
104
+ def receive_result(result):
105
+ host_event_bus.publish("vibration", result.to_dict())
106
+
107
+ class MyCalculation:
108
+ def __call__(self, result):
109
+ return {"quality_accepted": result.valid}
110
+
111
+ analyzer = VibrationAnalyzer(
112
+ AnalysisConfig(sample_rate_hz=12000, unit="m/s^2"),
113
+ stages={"host_calculation": MyCalculation()},
114
+ sinks=(receive_result,),
115
+ )
116
+ # host_event_bus is supplied by your application. No broker is imported here.
117
+ ```
118
+
119
+ For streaming input, call `analyzer.push(block)` and consume the returned list of
120
+ completed windows. Gaps, sequence changes, operating identity and clock continuity
121
+ matter. `analyze(block)` handles one complete block directly. Raw-window stages can
122
+ implement `analyze_window(result, readonly_values)` to use calibrated samples instead
123
+ of a display-decimated waveform. See [composition examples](examples/python_composition.py)
124
+ and the [module catalog](docs/MODULE_CATALOG.md).
125
+
126
+ ## What is implemented
127
+
128
+ | Area | Available behavior | Required evidence or boundary |
129
+ |---|---|---|
130
+ | Waveform and spectra | AC RMS, peak, crest, Pearson kurtosis, FFT amplitude, PSD, spectrogram and envelope views | Physical units, sample rate and usable sensor bandwidth |
131
+ | Bearing markers | BPFO, BPFI, FTF, BSF and 2BSF from explicit geometry or injected providers | RPM and validated bearing dimensions; ideal kinematics, no slip model |
132
+ | Normal reference | Robust feature center/scale, scope checks, relative scores and time-based alarm state | Confirmed normal records separated from threshold calibration and testing |
133
+ | Fault candidates | Gaussian baseline, normal-first diagnostics, RPM-conditioned reference bank and morphology head | Experimental; asset, channel, band, duration and speed contracts apply |
134
+ | Rule evaluation | Quantity/unit/frame contracts and caller-provided limits | **No ISO/KS limit tables are distributed**; obtain and assess rules lawfully |
135
+ | Optional service | Local HTTP, UDP ingest, SSE, SQLite history and offline UI | Start explicitly; configure authentication before non-local exposure |
136
+
137
+ ### Bearing-frequency example
138
+
139
+ ```python
140
+ from rotorvibe import BearingGeometry
141
+
142
+ # Illustrative geometry in a common diameter unit, not a vendor catalog entry.
143
+ bearing = BearingGeometry(rolling_elements=8, rolling_diameter=7.0,
144
+ pitch_diameter=35.0, contact_angle_deg=0)
145
+ print(bearing.frequencies(1200))
146
+ ```
147
+
148
+ ### Diagnostic states are different outcomes
149
+
150
+ `normal` means the applicable normal gate accepted the observation. `hold` means
151
+ there is no supported decision, not that the machine is healthy. `anomaly_unknown`
152
+ means deviation is detected but a trained fault type is not supported.
153
+ `fault_candidate` is a hypothesis within the model's recorded scope, not confirmation
154
+ of a physical root cause. Distances and similarity weights are not calibrated fault
155
+ probabilities. Keep `operational_use=False` for the research classifiers.
156
+
157
+ The morphology head must follow an anomaly gate. A standalone call to
158
+ `FaultDiscriminator` assumes that anomaly screening has already happened; it cannot
159
+ identify a healthy class. [A two-stage composition](examples/fault_discrimination_composition.py)
160
+ shows the ordering. No downloaded profile is silently enabled.
161
+
162
+ ## Evidence: retain the failures
163
+
164
+ The following experiments have different datasets and protocols. Their percentages
165
+ must not be merged into a single product accuracy score.
166
+
167
+ | Historical experiment | Recording-level separation | Observed result | Interpretation |
168
+ |---|---|---|---|
169
+ | Brito Gaussian comparison | Fixed four-recording final split; subsequently exposed | 63.69% overall; healthy recording 0/420 | Historical reproduction, not field qualification |
170
+ | CrossSpeed supported-speed check | 24 held acquisitions after rig-specific fitting | 93.74% of 479 windows | Matched apparatus and own normal reference |
171
+ | CrossSpeed 1200 RPM transfer | 59 acquisitions / 1178 windows | 11.04%; all 399 healthy windows anomalous | Cross-speed transfer failed |
172
+ | UOS 6204 conditioned model | Eight new source files / 320 windows | 51.88% overall; fault type 48.21% | Better normal reference handling; weak type coverage |
173
+ | UOS morphology improvement | Seven new 8 kHz fault files / 560 windows | 559/560 = 99.82%; old comparator 49.11% on the same new files | Same apparatus, bearing and severities; no new healthy test |
174
+ | New UOS 30204 bearing domain | Six healthy + seven fault files; all test-only | Strict API: all hold. Offline scope-bypassed stress: healthy 240/240 anomalous, types 124/280 | Generalization failed even after scope bypass |
175
+ | Independent UFRJ ABVT rig | Fifteen healthy + fifteen fault files; all test-only | Strict API: all hold. Two predeclared amplitude assumptions: types 0/30 | Unit unresolved; no physical FPR qualification |
176
+
177
+ In the latest unadapted evaluation, 580 windows all held in the real API. Zero alerts
178
+ with zero accepted healthy observations does **not** demonstrate a zero false-alarm
179
+ rate. The offline stress route is explicitly separate from the product API and must
180
+ not be used to hide an asset mismatch. UFRJ native CSV units are not explicit in its
181
+ official page; both fixed amplitude assumptions are reported without choosing one.
182
+
183
+ The private research workspace passed 547 checks, including its licensed standards
184
+ route. That is a historical count, **not the public release's test count**. Public
185
+ release checks are recorded separately in `research/public_verification.json`; the
186
+ normative table tests are excluded and replaced with synthetic caller-rule checks.
187
+ Read [the latest external assessment](docs/EXTERNAL_BLIND_VALIDATION.rst) and
188
+ [the evidence index](research/README.md) before quoting any performance figure.
189
+
190
+ ## Standards and copyright boundary
191
+
192
+ The private development project investigated ten supplied KS/ISO editions. This
193
+ public release deliberately excludes the PDFs, transcriptions, threshold lookup
194
+ tables, normative selectors and corresponding table fixtures. `vibration.iso` now
195
+ accepts an explicit `custom_rule`; it does not select an ISO machine category for you.
196
+ The example rules use invented demonstration numbers, not published standard limits.
197
+
198
+ Independent algorithm implementations, our explanatory comments and our experiment
199
+ results are released under MIT. The original external measurements, papers, vendor
200
+ catalogs and standards retain their own terms. Citation identifiers and legal license
201
+ text are intentionally preserved; third-party substantive prose and figures are not
202
+ repackaged. See [the rights inventory](docs/SOURCES_AND_RIGHTS.md).
203
+
204
+ ## Optional dashboard
205
+
206
+ ```console
207
+ rotorvibe-monitor --host 127.0.0.1 --port 8765
208
+ ```
209
+
210
+ The service does not simulate measured validation data when no dataset is present.
211
+ Use your own live waveform input or explicitly acquired research files. Python hosts
212
+ that construct `Application` must call `close()` to stop workers and sockets.
213
+ `VIBRATION_RUNTIME_HOME` selects a writable state directory;
214
+ `VIBRATION_PROJECT_HOME` locates separately acquired datasets. Do not point these
215
+ variables at confidential data when hosting an externally reachable service.
216
+
217
+ ## Reproduce, continue, cite
218
+
219
+ 1. Start with synthetic core checks; they need no PDFs or raw data.
220
+ 2. Use the source inventory to acquire external files under their own terms.
221
+ 3. Distinguish historical replay from a new blind test: every previously opened
222
+ recording is listed as exposed and cannot become a fresh holdout again.
223
+ 4. Freeze model, preprocessing, thresholds, metadata mapping and record membership
224
+ before opening the next holdout. Audit by original acquisition, not random windows.
225
+ 5. Keep independent normal reference, calibration and final-test records. Record
226
+ unsupported coverage and false negatives as well as correct classifications.
227
+
228
+ [Continuation instructions](docs/CONTINUATION.md) identify exact commands, preserved
229
+ artifacts, unavailable private components, model hashes and next validation needs.
230
+ The archive DOI is [10.5281/zenodo.23937890](https://doi.org/10.5281/zenodo.23937890);
231
+ use `CITATION.cff` for software citation. A DOI identifies an archived release; it
232
+ does not certify a diagnostic model.
233
+
234
+ Copyright © 2026 Ugwiyeon Lee. Original software: [MIT](LICENSE).
@@ -0,0 +1,70 @@
1
+ # RotorVibe — 회전기기 진동분석
2
+
3
+ Python 배열을 직접 연결해 파형·FFT·베어링 주파수·정상 기준·고장 후보를 계산하는
4
+ 연구용 라이브러리입니다. 웹 서버를 띄우지 않고 다른 모듈의 함수·객체와 조립할 수 있습니다.
5
+ 기존 `vibration` import를 유지하며 새 `rotorvibe` 진입점도 제공합니다.
6
+
7
+ **0.1.0a1은 알파 공개본입니다. 다른 장비와 베어링으로 옮기는 무학습 진단은 검증에
8
+ 실패했습니다. 고장 자동 확정·자동 정지·잔여수명·ISO 적합성 인증을 제공하지 않습니다.**
9
+
10
+ ## 설치와 시작
11
+
12
+ ```console
13
+ python -m pip install rotorvibe==0.1.0a1
14
+ ```
15
+
16
+ Python 3.11 이상, NumPy 2.x, SciPy 1.13 이상이 필요합니다. 합성 파형으로 실행 가능한
17
+ 전체 코드는 [영문 README](README.md)의 첫 계산 예제에 있습니다. 실제 사용 시에는
18
+ 센서 원시 배열, 샘플링률, 물리 단위, RPM, 채널 방향, 장비·운전 조건 식별정보를 전달합니다.
19
+
20
+ `VibrationAnalyzer.analyze(block)`은 완성된 블록 하나를 계산합니다.
21
+ `push(block)`은 연속 입력을 분석 창으로 나누어 완료된 결과 목록을 반환합니다.
22
+ `stages`로 사용자 계산을 추가하고 `sinks`로 결과를 전달합니다. 스케줄링과 저장은
23
+ 호스트가 맡습니다. 한 분석기 인스턴스를 여러 설비 스트림이 동시에 공유하지 않습니다.
24
+
25
+ ## 계산 결과를 해석하는 방법
26
+
27
+ - 파형 품질·단위·시간축이 유효해야 숫자의 의미를 해석할 수 있습니다.
28
+ - 정상 기준이 없거나 장비/설정이 다르면 `hold`입니다. 정상 판정이 아닙니다.
29
+ - `anomaly_unknown`은 이상은 감지했지만 고장 종류의 근거가 부족한 상태입니다.
30
+ - `fault_candidate`는 후보이며 물리 원인 확정이 아닙니다. 점수는 고장 확률이 아닙니다.
31
+ - 베어링 주파수는 명시한 내부 치수와 이상 운동학을 바탕으로 계산합니다. 실제 미끄럼과
32
+ 전달 경로 때문에 측정 피크가 정확히 일치한다고 보장하지 않습니다.
33
+
34
+ ## 가장 중요한 검증 결과
35
+
36
+ 동일 시험대·6204 베어링·같은 고장 강도의 별도 원본에서는 고장 종류 559/560개를
37
+ 맞혔습니다. 하지만 새 30204 베어링과 다른 기관 시험대의 자료를 추가하자 정식 API는
38
+ 580개 창을 모두 보류했습니다. 정상 수락률과 종류 판정률 모두 0%였습니다.
39
+
40
+ 적용 범위를 제외한 오프라인 수치 시험에서도 30204 정상 240개를 전부 이상으로
41
+ 표시하고, 고장 종류는 124/280개만 맞혔습니다. 다른 기관의 UFRJ 자료는 종류 정답이
42
+ 0/30이었고 원시 값 단위도 명확하지 않았습니다. 따라서 기존 99.82%를 새 설비의
43
+ 성능으로 인용할 수 없습니다. 기능 시험 통과와 예측 성능 합격은 다릅니다.
44
+
45
+ 각 단계의 데이터, 원본 분할, 보류, 실패, 모델 해시를 문서와 연구 기록에 남겼습니다.
46
+ 이미 개봉한 기록을 다시 나누어 새로운 독립 시험이라고 주장하면 안 됩니다.
47
+
48
+ ## 공개본에서 제외한 자료
49
+
50
+ 원래 연구 폴더의 KS/ISO 원문과 기준 수치표, 외부 원시 측정 자료, 외부 저장소 코드
51
+ 사본, 훈련된 시험대별 프로파일, 실행 DB와 개인 경로는 공개하지 않습니다.
52
+ 표준을 참고해 작성한 수치표도 코드에 넣어 재배포하지 않도록 분리했습니다.
53
+ 공개 `vibration.iso`는 사용자가 적법하게 준비한 `custom_rule`만 평가합니다.
54
+ 표준 번호를 입력했다고 규정 기준이 자동으로 제공되는 구조가 아닙니다.
55
+
56
+ 이 변경은 원본 연구 폴더에 덮어쓰지 않았습니다. 로컬 원본의 동결 모델과 감사
57
+ 기록은 그대로 보존했습니다. 공개본의 달라진 파일과 과거 결과의 범위는 별도 목록에 있습니다.
58
+
59
+ ## 자세한 문서
60
+
61
+ - [사용 설명서](docs/USER_GUIDE.md): 설치, 신호 계약, 스트리밍, API 연결, 고장 처리.
62
+ - [알고리즘 설명](docs/ALGORITHMS.md): 식, 가정, 특징, 정상 판정 및 후보 거리의 의미.
63
+ - [개발 과정](docs/DEVELOPMENT_HISTORY.md): 초기 구현부터 독립 시험 실패와 공개 준비까지.
64
+ - [연구 인계](docs/CONTINUATION.md): 남은 파일, 재현 명령, 개봉 이력, 다음 검증 조건.
65
+ - [모듈 목록](docs/MODULE_CATALOG.md): 파일별 역할과 함수/클래스 인터페이스.
66
+ - [출처·저작권](docs/SOURCES_AND_RIGHTS.md): 외부 자료 조건과 공개 제외 원칙.
67
+ - [연구 결과 모음](research/README.md): 예측 집계·동결 조건·접근 감사 기록.
68
+
69
+ 직접 작성한 코드·문서의 라이선스는 MIT입니다. 외부 데이터나 표준의 권리를
70
+ MIT로 바꾸는 것은 아닙니다. 소프트웨어 DOI: https://doi.org/10.5281/zenodo.23937890.
@@ -0,0 +1,206 @@
1
+ # RotorVibe
2
+
3
+ **Composable rotating-machine vibration analysis for Python.**
4
+
5
+ RotorVibe turns calibrated waveform blocks into spectra, bearing-frequency markers,
6
+ condition-relative evidence and inspectable diagnostic candidates. Use the NumPy API
7
+ inside your own application; the local web dashboard is optional.
8
+
9
+ **Alpha research release, version 0.1.0a1.** Signal calculations and bounded software
10
+ checks are documented. Fault classifiers have **not** qualified for unattended field
11
+ diagnosis: new-bearing and independent-rig transfer tests failed. No remaining-life
12
+ prediction, automatic shutdown authority or standards certification is supplied.
13
+
14
+ [한국어 설명](README.ko.md) · [User guide](docs/USER_GUIDE.md) ·
15
+ [Algorithm notes](docs/ALGORITHMS.md) · [Development record](docs/DEVELOPMENT_HISTORY.md) ·
16
+ [Continue this research](docs/CONTINUATION.md) · [Source/rights policy](docs/SOURCES_AND_RIGHTS.md)
17
+
18
+ ## Install
19
+
20
+ Python 3.11+ with NumPy 2.x and SciPy 1.13+ is required.
21
+
22
+ ```console
23
+ python -m pip install rotorvibe==0.1.0a1
24
+ ```
25
+
26
+ For an editable checkout and verification tools:
27
+
28
+ ```console
29
+ git clone https://github.com/lyullee/rotorvibe.git
30
+ cd rotorvibe
31
+ python -m pip install -e ".[verification]"
32
+ python -m unittest discover -s tests -v
33
+ ```
34
+
35
+ The public facade is `rotorvibe`; the earlier `vibration` namespace remains available
36
+ so existing Python integrations can keep their imports. Raw experimental data and
37
+ trained rig-specific profiles are acquired separately, not installed by pip.
38
+
39
+ ## A complete first calculation
40
+
41
+ This generated waveform verifies the API; it is not a measured fault example.
42
+
43
+ ```python
44
+ import numpy as np
45
+ from rotorvibe import AnalysisConfig, SignalBlock, VibrationAnalyzer
46
+
47
+ fs = 12000
48
+ t = np.arange(fs) / fs
49
+ samples = 0.20 * np.sin(2 * np.pi * 20 * t)
50
+
51
+ analyzer = VibrationAnalyzer(AnalysisConfig(
52
+ sample_rate_hz=fs, channels=1, unit="m/s^2",
53
+ window_seconds=1, hop_seconds=0.25, rpm=1200,
54
+ ))
55
+ result = analyzer.analyze(SignalBlock(
56
+ samples, fs, unit="m/s^2", rpm=1200,
57
+ channels=("drive_radial",), clock_id="daq_elapsed", episode_id="run-001",
58
+ metadata={"asset_id": "motor-01", "condition_id": "rated_load"},
59
+ ))
60
+ print(result.valid, result.metrics)
61
+ serializable = result.to_dict(include_features=False)
62
+ ```
63
+
64
+ `samples` may also be an `N × C` array. A one-dimensional input represents one
65
+ channel. Supply the physical unit, channel ordering, time basis and actual asset
66
+ identity; a numeric array alone cannot establish sensor calibration or installation.
67
+
68
+ ## Connect modules directly
69
+
70
+ The host controls acquisition, scheduling, storage and decisions. An analyzer instance
71
+ owns one stream; application callbacks receive the results synchronously.
72
+
73
+ ```python
74
+ from rotorvibe import AnalysisConfig, VibrationAnalyzer
75
+
76
+ def receive_result(result):
77
+ host_event_bus.publish("vibration", result.to_dict())
78
+
79
+ class MyCalculation:
80
+ def __call__(self, result):
81
+ return {"quality_accepted": result.valid}
82
+
83
+ analyzer = VibrationAnalyzer(
84
+ AnalysisConfig(sample_rate_hz=12000, unit="m/s^2"),
85
+ stages={"host_calculation": MyCalculation()},
86
+ sinks=(receive_result,),
87
+ )
88
+ # host_event_bus is supplied by your application. No broker is imported here.
89
+ ```
90
+
91
+ For streaming input, call `analyzer.push(block)` and consume the returned list of
92
+ completed windows. Gaps, sequence changes, operating identity and clock continuity
93
+ matter. `analyze(block)` handles one complete block directly. Raw-window stages can
94
+ implement `analyze_window(result, readonly_values)` to use calibrated samples instead
95
+ of a display-decimated waveform. See [composition examples](examples/python_composition.py)
96
+ and the [module catalog](docs/MODULE_CATALOG.md).
97
+
98
+ ## What is implemented
99
+
100
+ | Area | Available behavior | Required evidence or boundary |
101
+ |---|---|---|
102
+ | Waveform and spectra | AC RMS, peak, crest, Pearson kurtosis, FFT amplitude, PSD, spectrogram and envelope views | Physical units, sample rate and usable sensor bandwidth |
103
+ | Bearing markers | BPFO, BPFI, FTF, BSF and 2BSF from explicit geometry or injected providers | RPM and validated bearing dimensions; ideal kinematics, no slip model |
104
+ | Normal reference | Robust feature center/scale, scope checks, relative scores and time-based alarm state | Confirmed normal records separated from threshold calibration and testing |
105
+ | Fault candidates | Gaussian baseline, normal-first diagnostics, RPM-conditioned reference bank and morphology head | Experimental; asset, channel, band, duration and speed contracts apply |
106
+ | Rule evaluation | Quantity/unit/frame contracts and caller-provided limits | **No ISO/KS limit tables are distributed**; obtain and assess rules lawfully |
107
+ | Optional service | Local HTTP, UDP ingest, SSE, SQLite history and offline UI | Start explicitly; configure authentication before non-local exposure |
108
+
109
+ ### Bearing-frequency example
110
+
111
+ ```python
112
+ from rotorvibe import BearingGeometry
113
+
114
+ # Illustrative geometry in a common diameter unit, not a vendor catalog entry.
115
+ bearing = BearingGeometry(rolling_elements=8, rolling_diameter=7.0,
116
+ pitch_diameter=35.0, contact_angle_deg=0)
117
+ print(bearing.frequencies(1200))
118
+ ```
119
+
120
+ ### Diagnostic states are different outcomes
121
+
122
+ `normal` means the applicable normal gate accepted the observation. `hold` means
123
+ there is no supported decision, not that the machine is healthy. `anomaly_unknown`
124
+ means deviation is detected but a trained fault type is not supported.
125
+ `fault_candidate` is a hypothesis within the model's recorded scope, not confirmation
126
+ of a physical root cause. Distances and similarity weights are not calibrated fault
127
+ probabilities. Keep `operational_use=False` for the research classifiers.
128
+
129
+ The morphology head must follow an anomaly gate. A standalone call to
130
+ `FaultDiscriminator` assumes that anomaly screening has already happened; it cannot
131
+ identify a healthy class. [A two-stage composition](examples/fault_discrimination_composition.py)
132
+ shows the ordering. No downloaded profile is silently enabled.
133
+
134
+ ## Evidence: retain the failures
135
+
136
+ The following experiments have different datasets and protocols. Their percentages
137
+ must not be merged into a single product accuracy score.
138
+
139
+ | Historical experiment | Recording-level separation | Observed result | Interpretation |
140
+ |---|---|---|---|
141
+ | Brito Gaussian comparison | Fixed four-recording final split; subsequently exposed | 63.69% overall; healthy recording 0/420 | Historical reproduction, not field qualification |
142
+ | CrossSpeed supported-speed check | 24 held acquisitions after rig-specific fitting | 93.74% of 479 windows | Matched apparatus and own normal reference |
143
+ | CrossSpeed 1200 RPM transfer | 59 acquisitions / 1178 windows | 11.04%; all 399 healthy windows anomalous | Cross-speed transfer failed |
144
+ | UOS 6204 conditioned model | Eight new source files / 320 windows | 51.88% overall; fault type 48.21% | Better normal reference handling; weak type coverage |
145
+ | UOS morphology improvement | Seven new 8 kHz fault files / 560 windows | 559/560 = 99.82%; old comparator 49.11% on the same new files | Same apparatus, bearing and severities; no new healthy test |
146
+ | New UOS 30204 bearing domain | Six healthy + seven fault files; all test-only | Strict API: all hold. Offline scope-bypassed stress: healthy 240/240 anomalous, types 124/280 | Generalization failed even after scope bypass |
147
+ | Independent UFRJ ABVT rig | Fifteen healthy + fifteen fault files; all test-only | Strict API: all hold. Two predeclared amplitude assumptions: types 0/30 | Unit unresolved; no physical FPR qualification |
148
+
149
+ In the latest unadapted evaluation, 580 windows all held in the real API. Zero alerts
150
+ with zero accepted healthy observations does **not** demonstrate a zero false-alarm
151
+ rate. The offline stress route is explicitly separate from the product API and must
152
+ not be used to hide an asset mismatch. UFRJ native CSV units are not explicit in its
153
+ official page; both fixed amplitude assumptions are reported without choosing one.
154
+
155
+ The private research workspace passed 547 checks, including its licensed standards
156
+ route. That is a historical count, **not the public release's test count**. Public
157
+ release checks are recorded separately in `research/public_verification.json`; the
158
+ normative table tests are excluded and replaced with synthetic caller-rule checks.
159
+ Read [the latest external assessment](docs/EXTERNAL_BLIND_VALIDATION.rst) and
160
+ [the evidence index](research/README.md) before quoting any performance figure.
161
+
162
+ ## Standards and copyright boundary
163
+
164
+ The private development project investigated ten supplied KS/ISO editions. This
165
+ public release deliberately excludes the PDFs, transcriptions, threshold lookup
166
+ tables, normative selectors and corresponding table fixtures. `vibration.iso` now
167
+ accepts an explicit `custom_rule`; it does not select an ISO machine category for you.
168
+ The example rules use invented demonstration numbers, not published standard limits.
169
+
170
+ Independent algorithm implementations, our explanatory comments and our experiment
171
+ results are released under MIT. The original external measurements, papers, vendor
172
+ catalogs and standards retain their own terms. Citation identifiers and legal license
173
+ text are intentionally preserved; third-party substantive prose and figures are not
174
+ repackaged. See [the rights inventory](docs/SOURCES_AND_RIGHTS.md).
175
+
176
+ ## Optional dashboard
177
+
178
+ ```console
179
+ rotorvibe-monitor --host 127.0.0.1 --port 8765
180
+ ```
181
+
182
+ The service does not simulate measured validation data when no dataset is present.
183
+ Use your own live waveform input or explicitly acquired research files. Python hosts
184
+ that construct `Application` must call `close()` to stop workers and sockets.
185
+ `VIBRATION_RUNTIME_HOME` selects a writable state directory;
186
+ `VIBRATION_PROJECT_HOME` locates separately acquired datasets. Do not point these
187
+ variables at confidential data when hosting an externally reachable service.
188
+
189
+ ## Reproduce, continue, cite
190
+
191
+ 1. Start with synthetic core checks; they need no PDFs or raw data.
192
+ 2. Use the source inventory to acquire external files under their own terms.
193
+ 3. Distinguish historical replay from a new blind test: every previously opened
194
+ recording is listed as exposed and cannot become a fresh holdout again.
195
+ 4. Freeze model, preprocessing, thresholds, metadata mapping and record membership
196
+ before opening the next holdout. Audit by original acquisition, not random windows.
197
+ 5. Keep independent normal reference, calibration and final-test records. Record
198
+ unsupported coverage and false negatives as well as correct classifications.
199
+
200
+ [Continuation instructions](docs/CONTINUATION.md) identify exact commands, preserved
201
+ artifacts, unavailable private components, model hashes and next validation needs.
202
+ The archive DOI is [10.5281/zenodo.23937890](https://doi.org/10.5281/zenodo.23937890);
203
+ use `CITATION.cff` for software citation. A DOI identifies an archived release; it
204
+ does not certify a diagnostic model.
205
+
206
+ Copyright © 2026 Ugwiyeon Lee. Original software: [MIT](LICENSE).