Predict internal failures using thousands of measurements and tests made for each component along the assembly line. This would enable Bosch to bring quality products at lower costs to the end user. Using collaborative data from multiple sensors to improve the quality of the predictive maintenance algorithm.

What it does

Scalable, economical, automated and Robust Cloud Architecture capable of handling 10 million parallel requests Machine Learning and Artificial Intelligence (AI) Application with 96% predictive maintenance accuracy. Easy Business Integration, Reporting Functionality and notifications/alerting. One Click cloud architecture implementation using ARM templates and modular cloud architecture which can be reused by plugging-in code at different points.

Challenges we ran into

Sensor Fusion as the data was coming from 6000 different sensors Real-time Prediction for more than 6000 sensors Deep learning to detect failures with good accuracy and its deployment Scalability performance for warehousing and visualizing data inside PowerBI, so we integrated analysis services

Lessons Learnt

How to built efficient and robust cloud architecture that is able to handle high velocity of data and how to autoscale. Deploying a stable deep learning model on cloud and how to efficiently generate real-time predictions out of it. ARM templates seem like a lot of effort, but in the long run, they will save a lot of time and bugs.

Future work

We will be rolling out this big data architecture to our use-cases at our work This is more like lambda architecture implementation for both real-time and batch processing, but in future, we will test kappa architecture bit more.

Built With

Share this project: