Modern Manufacturing

How Artificial Intelligence is Improving Quality Management and Control

John Klinke
Director, Oracle Industry Strategy Group

Guest Author: Aniello Pepe, Director, Industry Solutions Group, Oracle.

We all know the impact of quality in manufacturing and how a defective product can jeopardize the reputation of even the most established brands. Highly visible cases in the automotive industry include the huge vehicle recall tied to Takata airbags and the Toyota vehicle recall due to accelerator pedal issues. These and other product defects cause heavy financial impacts on manufacturers and – in the Takata case – resulted in the company bankruptcy. Even less significant product defects can impact the financials and the reputation of manufacturers. 

Quality management and control are universally applied business practices that rely on well established, theoretically founded methodologies and techniques. But the practical application of concepts has to face the intrinsic difficulty for control, analysis and data interpretation. Here artificial intelligence (AI) and machine learning (ML) techniques can now play an important role, similar to how AI and ML are being adopted across other business areas.

Inspection and Validation

Quality inspections are typically related to checks on dimension, weight, aspect (e.g. level of finishing, color, etc.), capabilities (e.g. elasticity, impermeability, resistance, etc.) and features (e.g. device functionalities). Most of these checks can be automated using sensors or measuring equipment, while others requires some more sophisticated controls that are hard to automate and typically require human intervention that can add to costs and be error prone. 

Consider, for instance, visual checking the quality of final packaging. Or other checks requiring the use of other human senses like hearing (e.g. detecting “strange noises”) or even taste (e.g. organoleptic properties in food & beverage). In these cases, appropriate sensors can collect raw data (e.g. video, audio) but then data interpretation is required, since there are often no clear rules to define what’s good and what’s not. Here ML techniques – based on training from a number of cases to derive a model that is continuously updated from the experience of any new cases – are being effectively applied with a high level of confidence.   

Cause Analysis and Prevention

Complementing human effort to improve, automate and accelerate quality checks is not the only or most important application of ML for quality management in manufacturing. For many years researchers and practitioners have addressed the issue of preventing quality problems in manufacturing, formalized in a number of techniques and procedures – like total quality management (TQM) and others. Many of these either focus on improving the intrinsic quality of production processes or attempt to facilitate the analysis of the root causes of quality problems (e.g. Ishikawa diagrams). However, the real issue behind these efforts has always been the inherent complexity of actual manufacturing environments with the need to take into account a large number of internal parameters that can be summarized in the 5M framework (man, machine, material, method, and measurement).

ML techniques can now provide a valuable tool for root cause analyses (RCA). Indeed, at the core of ML is the capability to find correlations from large sets of information, which is exactly what is needed in RCA. The digitization of business processes – not only manufacturing, but also design, logistics, service, finance – together with the continued adoption of the industrial internet of things (IIoT) is transforming the production world, providing a vast, rich set of data that modern ML techniques can readily leverage.
Oracle is Here to Help
Oracle understands the importance of quality in all stages of manufacturing and is a leading vendor with regards to incorporating AI and ML technologies into our business solutions. Oracle has embedded ready-to-use AI capabilities in key applications including enterprise resource planning, supply chain management, and IoT analytics. For example, our Smart Manufacturing industry solution leverages AI to provide a robust platform for quickly resolving maintenance issues, reducing downtime, and improving product quality. For more information about Oracle’s AI-powered solutions for manufacturers, check out our Adaptive Intelligent Applications for Supply Chain and Manufacturing.