EdgePredict Concept

EdgePredict is a modular platform for edge AI that detects anomalous vibration patterns (wear, misalignment). Our system performs signal acquisition, preprocessing (FFT/RMS), inference with OpenVINO, and local dashboard visualization. It supports integration with microcontrollers, industrial sensors, and MES/ERP systems.

Modular architecture
Support for multiple sensors and protocols
Inference optimized for Intel CPUs

Data Flow

Pipeline: Sensor (IP68) → MCU (Arduino/ESP32) → Edge node (FFT/RMS, normalization) → Inference (OpenVINO) → Dashboard / API (REST/WebSocket)

Sensor: vibration/audio/acceleration acquisition (1–10 kHz).
MCU: light filtering and packet forwarding (UART/I²C/SPI).
Edge node: preprocessing (FFT, RMS, windowing) and standardization.
Inference: Intel OpenVINO-optimized models on CPU.
Visualization/Integration: local dashboard and REST/WebSocket API.