← Back to Architecture Edge · Sensing · Bogie

Pod A
Vibration · Temperature

IoT sensing unit installed on bogies and running gear. It captures triaxial acceleration up to 6,667 kHz and continuous bearing temperature, without modifying the vehicle or accessing the TCMS.

What it is

Pod A is IN-SIGHT's primary sensing unit, designed for bogie dynamics monitoring and bearing condition. It incorporates a 6-degree-of-freedom MEMS IMU —triaxial accelerometer plus gyroscope— and a precision temperature sensor, integrated in a compact sealed enclosure.

The mounting combines a magnetic base and an M6 screw, allowing installation on the bogie frame, axle or axle box in under 4 hours per vehicle, with no TCMS access or special tools. Pod A is completely passive from the perspective of the existing railway system: it does not interfere with any of the vehicle's control or communication circuits.

Non-intrusive installation: No wiring to the vehicle. Autonomous power from the 24 V auxiliary bus via an isolated DC/DC converter or a LiFePO₄ battery with inductive charger. Total commissioning time per unit: under 4 hours.

Role in IN-SIGHT

Pod A acts as the entry point of the monitoring chain. Its raw signals feed the DSP pipeline of the CM4 Edge CPU, which extracts frequency features associated with each known failure mode:

  • Wheel bearings: BPFO, BPFI, BSF and FTF characteristic frequencies computed from the bearing geometry and the rotation speed estimated by FFT of the vibration signal.
  • Wheel wear (flat spots): Detection of periodic impacts in the vertical acceleration signal at the wheel revolution frequency.
  • Track irregularities: Spectral analysis in the 1–50 Hz range correlated with the known track profile to separate roughness from component failure.
  • Bearing temperature: Continuous axle-box monitoring with detection of anomalous thermal drift as a precursor of failure due to insufficient lubrication.

Acquisition pipeline

Pod A transmits raw data via SPI to the CM4 on each sampling cycle. The complete flow from sensor to structured telemetry is as follows:

IMU MEMS  (SPI @ 6.667 kHz)
    │  ax, ay, az [g]  +  Temperature [°C]
    ▼
Double circular buffer  (4096 samples)
    │  Overflow → data-loss event
    ▼
Hann window  +  1024-point FFT
    │  Spectral resolution: ≈ 6.5 Hz/bin
    ▼
Per-subsystem band-pass IIR filters
    │  Bearings   :  500 Hz – 3 kHz
    │  Wheel/rail :    5 Hz – 500 Hz
    │  Low freq.  : 0.5 Hz – 50 Hz
    ▼
Feature extraction
    │  RMS · Kurtosis · Crest Factor · FFT peaks
    ▼
Local EKF  →  comparison vs. Golden Run
    │
    ├─ Normal    → aggregate and send every 30 s
    └─ Anomaly   → immediate event + classification

Detected failure modes

Pod A covers the subsystems with the greatest impact on the availability of short/medium-distance fleets:

  1. Bearing outer-race damage (BPFO): Amplitude increase at the ball-pass frequency of the outer-race defect.
  2. Inner-race damage (BPFI): Amplitude modulation of the carrier at the revolution frequency.
  3. Rolling-element wear (BSF): Change in the high-frequency spectral content.
  4. Wheel flat spot: Single impact per revolution with a characteristic time signature.
  5. Bearing overheating: Temperature persistently above the baseline ±Δ (configurable).