Telemetry Anomaly Detection Toolkit

telemetry-anomdet is an open-source anomaly detection toolkit for spacecraft telemetry. It ingests raw telemetry (CCSDS, CSV), preprocesses it, and runs a stacking ensemble of classical and deep learning detectors with per-channel SHAP attribution and LLM-generated diagnostic reports — designed to produce actionable diagnostics within the ground station inter-pass window.

Validated on SMAP (NASA) and OPS-SAT (ESA).

Current features:

  • CCSDS and CSV ingestion into long-form TelemetryDataset

  • Preprocessing pipeline: clean, dedupe, resample, interpolate gaps, normalize

  • Windowed feature extraction: statistical features and raw 3D tensors for sequence models

  • BaseDetector interface: unified fit / decision_function / predict / is_anomaly API shared by all detectors

  • PCAAnomaly and KMeansAnomaly classical detectors (3D input, flatten internally)

  • AnomalyEnsemble: stacking combinator with configurable normalization and combine strategy

  • Per-model score decomposition via score_components() (SHAP hook)

Coming in the next few months:

  • IsolationForestAnomaly

  • GDN — graph deviation network for inter-sensor relational anomalies

  • TranAD — transformer-based sequence reconstruction

  • SHAPExplainer — per-channel attribution over score_components()

  • SMAP ingestion and benchmark pipeline

Coming in by the end of 2026:

  • LLM reasoning layer (Llama 3.1 8B on Jetson Orin via llama.cpp)

  • OPS-SAT cross-dataset generalization evaluation

Contents

User Guide

API Reference