Introduction

Data plays a critical role in ensuring smooth and optimal operations for a company. By having quality manufacturing, operations, supply chain and products data, companies can build good AI/ML models to streamline manufacturing operations and make the best decisions possible. Unfortunately, many companies just don’t have good data to feed the models. Hence, they suffer from “garbage in, garbage out” syndrome. The key to success is to first have reliable data and then build the best models to ensure good decisions based on sound data.

Data quality is a problem across industries. We at Cognistx see the same at many manufacturing companies. In some cases, they do not capture information correctly, the units are different or there are inconsistent practices across plants. All these lead to poor data quality, which prevents companies from implementing the best AI/ML solutions for their needs.

Once you have clean, actionable data, you can utilize it to reduce manufacturing defects, reduce the number of runs needed to meet specifications, ignore erroneous alerts from sensors and improve staff/labor utilization.

Our Solution: Improved Data and Beyond!

One of the main products we’ve been focusing on during the last three years is our Data Quality Engine (DQE). DQE is based on several years of research and operational experience. DQE helps companies clean their data with a combination of business rules and AI/ML models. DQE systematically helps companies climb the data-quality steps to achieve AI-enabled data, as outlined in the diagram below.

‍

‍

DQE and similar tools help companies identify millions of data issues and allow them to automatically fix them with little human intervention.

• Step 1 identifies the data-quality issues.
• Step 2 fixes these issues.
• Step 3 fixes the underlying systems in an automated manner.
• Step 4 keeps the data clean and AI ready.

By fixing the root cause of the data issues, the company ensures the data remains clean and identifies operational issues before they become big problems. DQE with clean data and AI/ML models puts companies on a prescriptive path, allowing them to stay ahead of constantly changing business conditions and environments to make the best decisions possible.

Clean Data Leads to Improved Manufacturing Operations

The key to success is to demonstrate manufacturing excellence and ROI. Improved data quality is a great starting point that allows the following five (5) optimizations.

• Improved predictive maintenance. This is done by analyzing the numbers of runs and error logs companies can predict before a given line or machine will fail and schedule preventative maintenance. It helps limit costly line downtime and ensures smooth operations. ‍
• Reduction in manufacturing errors. By analyzing the number of defects from a line, the company can use AI/ML to perform a RCA (root cause analysis) of issues that result in higher defects. Addressing these root causes leads to significant improvements in quality, profits and customer satisfaction. ‍

• Ignore erroneous alerts. Time sensors can generate alerts that don’t need to be acted upon. AI/ML models can help triage such alerts and allow the operations team to ignore false alerts. This reduces operational costs and improves manufacturing efficiency.‍
• Reduction in manufacturing batches/costs. AI/ML models can help reduce the number of production runs or batches needed to meet product specifications or goals. By predicting the changes needed to meet the specifications, the company can reduce extra batch runs and the associated time and costs. ‍
• Improved staff productivity. By having data issues automatically identified and most of them fixed, staff can focus on higher-order tasks and not be frustrated with bad-data issues. Once the business rules and AI/ML are reliable, automated processing ensures data errors are identified and fixed automatically. Your team can view what was fixed and focus on tasks that make your company more efficient.

It is important to build such ROI measures into your data-quality and manufacturing processes from the beginning. It is equally critical to be able to report on them and learn from them for ongoing improvement and continued executive sponsorship.

Conclusion

If your organization has grown over time or you have inherited these issues, our Data Quality Engine can help clean your data and then generate the best predictions.

Our AI/ML models can build upon the clean data to deliver significant manufacturing and operational efficiencies. If you’re interested, we can review your data and processes and deliver significant shareholder value.