Pipelines

End-to-end workflows that turn raw Azure timeseries into actionable production insights. Each pipeline starts with just three inputs:

1. Azure connection config (connection string, SAS URL, or AAD credentials)
2. UUID list (the signal identifiers for your use case)
3. Time range (start and end timestamps)

---

Common Pattern

Every pipeline follows the same flow:

graph LR
    A[Azure Blob Storage] --> B[Load by UUIDs + Time Range]
    B --> C[Enrich with Metadata]
    C --> D[Validate Data Quality]
    D --> E[Harmonize & Transform]
    E --> F[Event Detection / Analytics]
    F --> G[Results & KPIs]

---

Available Pipelines

• OEE Dashboard

  ---

  Machine state, part counters, and reject signals into daily OEE breakdown by shift with availability, performance, and quality components.

  Signals: 4 UUIDs

• Cycle Time Analysis

  ---

  Cycle triggers and part numbers into cycle time statistics, slow cycle detection, trend analysis, and golden cycle comparison.

  Signals: 3 UUIDs

• Downtime Pareto

  ---

  Machine state and reason codes into Pareto analysis, shift-level downtime comparison, and availability trends.

  Signals: 2 UUIDs

• Quality & SPC

  ---

  Measurement signals with tolerances into outlier detection, SPC rule checks, control charts, and Cp/Cpk capability trending.

  Signals: 1+ measurement UUIDs

• Process Engineering

  ---

  Setpoint, actual value, and process state signals into setpoint adherence, startup detection, control loop health, and stability scores.

  Signals: 3 UUIDs

---

Prerequisites

All pipelines require:

pip install ts-shape
pip install azure-storage-blob   # for Azure loaders

For detailed module documentation, see the API Reference or the Guides.