Tata Consultancy Services Develops Distributed Cloud-Based Vehicle Predictive Maintenance Solution
The Solution Improves Predictive Maintenance Efficiency and Cost-Effectiveness
MATLAB seamlessly integrates the entire machine learning pipeline: from onboard computers to the cloud.
Key Outcomes
- Reduced cost of cloud-based machine learning by deploying AI models to onboard computers
- Using MATLAB tools expedited the process of developing, training, and deploying models, resulting in significant time savings
- Low code tools and Diagnostic Feature Designer app enabled feature extraction and algorithm development
Software-defined vehicles generate massive amounts of data that can be used for critical predictive maintenance tasks. However, sending all this data to the cloud for processing can be inefficient and costly. Tata Consultancy Services (TCS), an India-based IT company, used MATLAB® to create a distributed machine learning solution for predictive vehicle maintenance.
TCS developed an architecture that processes most sensor data locally, with machine learning models running on the vehicle’s onboard computers and sending computed features to the cloud for more computationally intensive analytics. When a fault is detected, the on-device model alerts the cloud, which can then run more complex failure prediction models.
The TCS team used MATLAB tools to visualize the data and discover useful patterns. Parallel Computing Toolbox™ helped them accelerate this process by dividing the data and simultaneously processing it in chunks. With Statistics and Machine Learning Toolbox™, Classification Learner app, Regression Learner app, and Diagnostic Feature Designer app, the team explored sensor and time-series data to discover predictive features, train different machine learning models, and compare their performance. To visualize the insights, they developed an app and deployed it on Microsoft® Azure® with MATLAB Web App Server™.
