An AI-based predictive maintenance analytics solution for a multinational automaker

• Analytics & Modeling - Predictive Analytics
• Processors & Edge Intelligence - Embedded & Edge Computers

• Automotive

• Discrete Manufacturing
• Maintenance

• Predictive Maintenance
• Real-Time Location System (RTLS)

• Data Science Services
• System Integration

A Fortune 500 American multinational automaker was seeking a solution to predict faults in their auto parts to proactively ensure fault-free production, thereby saving maintenance time and improving the customer experience. The company faced several challenges. Data was being generated from multiple discrete systems, all of which had to be processed simultaneously to get a complete picture. The data was in different formats like JSON, CSV, and other proprietary formats. The cutting tools had to be replaced before they reached end-of-life, affecting the production quality. Therefore, the automaker was looking for a solution that would predict in real-time, giving them enough time to replace the waned cutting tools. The data collected from multiple systems had several quality issues and missing records. This data had to be formatted, cleansed, and prepared before it could be fed into the predictive analytics models. The manufacturing unit had thousands of machines generating millions of events every minute. The automaker needed to process this massive amount of data in real-time using a single solution and shared infrastructure. Real-time alerts to floor operators and the downstream application was a crucial component. Any failure or delay in these alerts had a direct impact on the quality of parts produced.

The customer is a Fortune 500 American multinational automaker. They are one of the world's largest automotive manufacturers, with thousands of machines in their manufacturing unit generating millions of events every minute. The company was looking for a solution that could predict faults in their auto parts to proactively ensure fault-free production. This would not only save maintenance time but also improve the customer experience. The company needed a solution that could handle data from multiple discrete systems, process it in real-time, and provide alerts to floor operators and the downstream application.

The automaker deployed Gathr in their auto-parts manufacturing division to collect data from various sources, combine them, and predict the life of the cutting tools used to create auto parts. Gathr enabled the client to implement an end-to-end predictive maintenance solution leveraging out-of-the-box drag-and-drop operators. It helped them effortlessly design a complete solution with the following ready-to-use capabilities: Reading data from various sources, Out-of-the-box data-wrangling transformations, Data quality management, Rule-based, Scalable scoring of trained models, Data aggregation, Data profiling, Monitoring and reporting. To build a predictive maintenance solution for tool replacement, Gathr implemented a five-stage approach: Real-time data ingestion, Data cleansing and integration, Data transformation and online scoring, Real-time alerts, Monitoring and dashboard.

• Connected multiple discrete sources to join and process data
• End-to-end data quality and preparation
• Model lifecycle management

• Improved productivity through real-time fault prediction
• Reduced maintenance time through proactive tool replacement
• Improved customer experience through fault-free production