Example - Arc Fault

Introduction

Edge AI on TI microcontrollers enables running machine learning algorithms directly on devices like C28, C29, and MSPM0 series. These MCUs are particularly valuable for arc fault detection in electrical systems, where they analyze current waveforms in real-time to identify dangerous electrical arcs that could cause fires. The low-latency processing is critical for safety applications, as decisions must be made in milliseconds. TI's specialized hardware and DSP capabilities allow these MCUs to efficiently run complex pattern recognition algorithms that distinguish normal electrical operations from hazardous conditions without requiring cloud connectivity.

Configuration

Edge AI Studio and Modelmaker provides the following configuration best suited for Arc Fault applications. The configuration revolves around large frame size, FFT, frequency binning and frame concatenation.

1. FFT1024Input_256Feature_1Frame_Full_Bandwidth
2. FFT1024Input_256Feature_1Frame_Half_Bandwidth
3. FFT1024Input_64Feature_4Frame_Half_Bandwidth
4. FFT1024Input_32Feature_8Frame_Quarter_Bandwidth

API

• AI - Feature Extract APIs
• AI - Neural Network Utility APIs