Taiwanese IoT AI specialist aiSensing has developed an AI-enabled Industrial IoT solution that can determine multiple fault modes for predictive maintenance applications. The vibration sensor monitors equipment status and is able to identify multiple different fault modes.

The solution is based on an EOS S3 multi-core sensor processing SoC and SensiML Analytics Toolkit from San Jose based endpoint AI specialist QuickLogic.

aiSensing’s Predictive Maintenance (PdM) solution integrates AI/ML technology to monitor the status of manufacturing equipment locally without the need for an internet-based cloud connection. 

This approach improves robustness, performance, real-time functionality, and security.

The aiSensing solution integrates a vibration sensor, using AI to differentiate between normal and abnormal operation for a particular piece of manufacturing equipment, and sends an alarm message to engineers and managers as soon as an abnormal state is detected. By identifying pending failures before they happen, the intelligent sensor allows operators to shut down equipment for maintenance in an orderly way and thus manage their production lines more efficiently and cost-effectively. It can also help save cost by avoiding unnecessary preventative maintenance.

“With QuickLogic’s multi-core ultra-low power EOS S3 SoC plus Open Source QuickFeather Dev Kit and SensiML’s Analytics Toolkit, aiSensing has developed three generations of our AI Vibration Detector in less than six months to support different customer requirements,” said Dennis Chu, aiSensing’s chief technology officer. “Our resulting endpoint AI-based IoT solution helps us enable predictive maintenance applications with better performance and cost than cloud-based AI solutions, and positions us well for future growth.”