Elevator Health Monitoring Predictive Maintenance using Anomaly Detection

Elevator Health Monitoring Predictive Maintenance using Anomaly Detection Learning Algorithms

Mission statement: Prevention is better than Cure

The problem 450 units are deployed, 70, 000 maintenance visits… Can we optimize? What if we could predict breakdowns before they occur? Maintenance Visits Deployment 70, 000 450

The solution Machine Learning + IOT Isolation Forests & Support Vector Machines training on real-time Elevator data to flag anomalies Cross platform dashboard to manage maintenance strategies through notifications and collaborative chat

How it works Step 3 Step 1 Collect IOT sensor data Step 2 Detect anomalous behavior Flag Anomalies Send Notifications Collaborative Chat

Architecture Research Notebook Train Model Machine Learning API SMS Elevator Simulator Node Red Raspberry Pi Sensor Data IBM IOT Dash. Board Realtime Analysis Cross Platform

Elevator Simulator

Machine Learning API

Cross Platform Dashboard

Real-time Data Analysis

Track Anomalies with Collaborative Chat

Real-time SMS Notifications

Anomaly Detection Insights

Production Architecture IBM Data Science Jupyter Notebooks Bluemix Python Flask Machine Learning & Data Science KONE Elevators Node. RED On Ras. Pi Bluemix CF Apps IBM Bluemix Io. T PHP AInstein MQTT Dashboard Bluemix Node SDK IBM Cloudant * IBM and KONE are registered trademarks. Web App Node-RED Realtime Front-end Dash Mobile App

The AInstein Dashboard SMS Secure Auth Role-based Access Mobile Anomaly Notification Web Convenient Access Realtime Issue Tracking with Chat Technician Service Scheduling Graphs CSV access Anomaly Dataset

Team AInstein Pranav Chendur TK Aman Hussain Subhojeet Pramanik Entrepreneur, Web Tinkerer, Hobbyist Photographer Machine Learning & Artificial Intelligence Enthusiast Artificial Intelligence Programmer, Founder of Born. Free Labs amandavinci. me github. com/subho 406 www. tkpc. in