IoT06

[Audio] RUPA JNA NATIONAL LEVEL PROTOTYPE CHALLENGE PROBLEM CODE : IoT06 PROBLEM STATEMENT : AI/ML-BASED PREDICTIVE MAINTAINANCE SYSTEM FOR INDUSTRIAL MACHINES.

[Audio] THE PROBLEM : Catastrophic machine failure causes sudden, unplanned production stops. THE IMPACT : Lost production and repair costs : Rs. 50,000 – Rs.1,00,000 per hour for large industries . CURRENT SOLUTIONS FAIL: Existing threshold-based alerts are too slow or generate too many false alarms , leading to either missed failures or unnecessary maintenance..

[Audio] OUR AI-POWERED SOLUTION THE GOAL: To move maintenance from Reactive/Preventive to Predictive . THE SOLUTION: Our system uses a multi-sensor array to continuously monitor motor health and feeds the data into a Deep Learning Model that accurately predicts the Remaining Useful Life (RUL) THE RESULT : We provide maintenance teams with a 14-day advance notice to schedule repairs during off-production hours ..

[Audio] SYSTEM ARCHITECTURE: DATA FLOW EDGE LAYER : Low-cost ESP32 Microcontroller and a TriAxial Accelerometer . Collects data and performs local FFT CLOUD LAYER:- (OUR BACKEND ) Data is transmitted via MQTT protocol to our Cloud Database . AI ENGINE: The database feeds into our LSTM( i.e., Long Short – Term Memory) Neural Network, which is trained to predict RUL. USER LAYER: The dashboard clearly displays the predicted RUL and machine health status ..

[Audio] THE LSTM DEEP LEARNING ADVANTAGE 1. THE FLAW IN EXISTING SYSTEMS: Simple systems use static thresholds . Degradation is non-linear and complex. 2. OUR INNOVATION : The LSTM model is a type of Recurrent Neural Network designed for Time-Series Data. It learns the complex, subtle history and sequence of degradation , giving us far superior accuracy . 3. OUTPUT : Instead of a simple "ON/OFF" alert , we output a continuous , precise "Remaining Useful Life" (RUL) value..

[Audio] ROI : THE COMMERCIAL CASE TARGET MARKET : Textile , Cement , and FMCG manufacturing sectors. UNIT COST : Est. Rs.75,000 per unit (sensor + microcontroller) PRICING: Monthly Subscription of Rs.800 per machine . ROI JUSTIFICATION: The subscription is negligible compared to the cost of one hour of lost production. Our system guarantees a minimum 10x Return on Investment by eliminating unplanned downtime. U.

[Audio] CURRENT STATUS AND TEAM TEAM NAME :TEAM MEMBERS: ROLES PROTOTYPE FUNCTIONAL:Our V1 prototype is fully functional and successfully calculates the RUL on a controlled test rig. TESTING/RESULTS :We have completed XX hours of data collection and initial model training . Initial validation shows a 95% accuracy in RUL prediction . We will confirm these numbers in the next step. FUTURE ROADMAP:PILOT PROJECT : Secure a 3-month pilot project with a local textile or FMCG factory . MODEL DEPLOYMENT : Transition from test rig to cloud-based data ingestion for real-time industrial deployment . COMMERCIALIZATION: Finalize IP protection and move to commercial deployment ..

[Audio] CONCLUSION PreDict-AI is a robust , scalable solution that solves a multi-core problem for Indian industry . We have the technical expertise and the business model to succeed. * Our Prototype is ready to build. We are seeking the RS.10 Lakh Ignition Grant to deploy a pilot program in our target industry . THANK YOU We are committed to industrializing the future of predictive maintenance , one asset at a time. If you have any further questions : Please contact [email protected].