Bibliothek / Monitoring, Betrieb & Störungen
Monitoring, Betrieb & Störungen
FRA-Streckenmängel mit OpenAI, Slack und Google Drive klassifizieren und Abschaltungen freigeben
Läuft täglich, um Streckenmängel nach FRA-Schweregrad zu klassifizieren, Geschwindigkeitsbeschränkungen zu berechnen und schwerwiegende Abschnitte vor der Abschaltung zur Genehmigung durch den Vorgesetzten zurückzuhalten.
Was sie bewirkt
📘 WORKFLOW DETAILS Short-Line Railroad FRA Track Defect Severity Router & Tag-Out Approval Gate Value: turns a daily track-inspection pull into an FRA-Part-213-style severity classification, a real speed-restriction/remedial-timeframe computation, and — for the segments that matter most — a supervisor approval gate before any segment is actually tagged out of service. Who it's for: short-line and regional railroad track-maintenance departments who run daily or periodic FRA Class-of-Track inspections and need defect severity triaged consistently, with a human always in the loop before the most severe (Class 5) calls take effect. How it works: 1. Schedule Trigger polls daily for segments due inspection. 2. Fetch Segments Due & Inspector Notes pulls each due segment's inspector notes as its own item. 3. Classify Defect Severity (Text Classifier, own Chat Model) reads the inspector's free-text notes and routes each segment into one of 4 categories: No Defect, Class 1-2 Minor, Class 3-4 Serious, Class 5 Immediate Halt. 4. Each category's own Set node stamps a defect_category field (Text Classifier itself never writes one — see the standing rule), then all 4 converge into Compute FRA Speed Restriction & Remedial Timeframe (Code), which computes a real speed restriction (mph) and remedial deadline off the track class + category. 5. Requires Immediate Tag-Out? branches: anything short of Class 5 is shaped straight to filing. A Class 5 call instead triggers Request Supervisor Tag-Out Approval — a genuine Slack send-and-wait gate. 6. Route Approval Response (3-way Switch) reads the supervisor's reply and finalizes as approved-as-recommended, approved-with-edit (supervisor's own text becomes the final action), or rejected (sent back for a field re-check). 7. Every path — non-urgent and all three approval outcomes — converges on Archive Defect Tag-Out Record, then Notify Track Supervisor & Dispatch. Node summary: Schedule Trigger → Set (normalize) → HTTP Request (segments due) → Text Classifier+Chat Model (severity) → Set×4 (category tag) → Code (speed/remedial calc) → If (tag-out gate) → Slack send-and-wait (approval) → Switch (3-way response routing) → Set×3 (finalize outcome) / Set (non-urgent shape) → Google Drive (archive) → Slack (notify). Input: the daily segments-due API response, one item per segment with inspector_defect_notes and track_class. Output: a Drive-
Die Schritte
- Poll Segments Due for Inspection
- Normalize Poll Window
- Fetch Segments Due & Inspector Notes
- Classifier Model
- Classify Defect Severity
- Tag No Defect
- Tag Minor
- Tag Serious
- Tag Immediate Halt
- Compute FRA Speed Restriction & Remedial Timeframe
- Requires Immediate Tag-Out?
- Shape Non-Urgent Record
- Request Supervisor Tag-Out Approval
- Route Approval Response
Insgesamt 19 Schritte.
Ähnliche Automatisierungen
Monitoring, Betrieb & StörungenTriage hospital FHIR observations with Google Gemini, Google Sheets, and Slack15 SchritteMonitoring, Betrieb & StörungenRSS-Themen mit OpenAI, Jina AI und Slack überwachen18 SchritteMonitoring, Betrieb & StörungenRSS-Themen mit OpenAI und Slack-Alarmen überwachen18 SchritteMonitoring, Betrieb & StörungenRSS-Themen mit OpenAI überwachen und Slack benachrichtigen35 Schritte