Machine Learning / AI based Arc Fault Detection (AFCI) Demonstration Application

Introduction: Why Arc Fault Detection, and What Is This Demo?

Electrical arc faults are a leading cause of fires in residential, commercial, and industrial environments, often resulting from damaged wires or loose connections. Traditional arc fault detection methods rely on fixed thresholds and simple rules, which can struggle to distinguish between true faults and harmless electrical noise—leading to false alarms or missed events.

To address these challenges, Microchip introduces a machine learning (ML)-based arc fault detection reference demo built on the powerful dsPIC33A Digital Signal Controller (DSC) platform. This demo leverages advanced ML algorithms and real-time digital signal processing to deliver faster, more accurate, and more reliable arc fault detection at the edge. Designed as a reference solution, it enables customers to accelerate the development of their own ML-enabled arc fault detection systems for applications such as solar PV, EV charging, smart ignition, and industrial safety.

Features of the reference demo:

• High-Performance Edge ML: Utilizes dsPIC33AK128MC DSC with a 200 MHz CPU, high-speed DSP engine with optimized DSP libraries, and hardware floating-point unit for efficient, real-time ML inferencing.
• Superior Detection Accuracy: ML models trained on real-world data can achieve up to 98% arc fault classification accuracy, significantly reducing false positives compared to traditional methods.
• Wide Application Range: Supports both AC and DC arc fault detection across solar PV, micro inverters, EV chargers, automotive ignition systems, data centers, and more.
• Flexible Voltage and Current Ratings: Designed to accommodate 120–400 V AC, 48–1000 V DC, and up to 100 A current, with scalable hardware for various application needs.
• Comprehensive Reference Package: Includes schematics, PCB files, bill of materials, optimized DSP libraries, demo code, and a detailed user guide to accelerate customer development.
• Modular and Scalable: Provides foundational building blocks for customers to adapt and expand the solution to their specific requirements.