Predictive quality as a project

95% scrap reduction and 500,000€ savings? Increasing product quality through automated quality control

It's actually a very simple formula: The poor quality of a product leads to lasting damage to the relationship with the customer. If the quality of a product is high, a sustainable customer relationship is usually established. But keeping the quality of a product permanently improved depends on many variables. Ideally, it is based on automated quality control.

Product quality depends primarily on excellent process quality. Process quality means reducing waste. Predictive quality, i.e. the data-based optimization of process and product quality, is therefore an indispensable tool for increasing product quality. The American company Hill's Pet Nutrition, for example, has long recognized this.

AI-supported predictive quality system using pet food products as an example

Hill's Pet Nutrition manufactures pet food products. The business strategy requires frequent new starts and product changes. With every introduction and changeover, the product development team is faced with the challenge of producing the new product in accordance with strict quality standards for moisture, fat, protein content and density. In the past, every changeover required manual testing, which took 40 minutes each time. Only after this analysis could adjustments be made to the system settings.

that AI-supported predictive quality system gives a prediction of the result of the manual test. Based on this, suggestions are provided to the production team in real time as to how the system settings should be adjusted to ensure the required product quality even before final laboratory data is available. Once the suggested settings have been made, all that is required is final and less time-consuming testing. The use of predictive quality technologies and automated quality control led to a significant reduction in waste and increased the CPK value (the most important parameter for describing the ability of processes) by half. This means that the rejects were reduced by around 95%.

What is predictive quality?

Predictive quality, also predictive quality assurance, comes as part of Data Science and predictive analytics are used. The aim is continuous process optimization based on automated forecasts. Various parameters are used to determine the quality of a product. From the specified tolerance limit of single-part production to environmental conditions such as relative humidity, individual parameters can lead to product problems.

Predictive quality enables automated quality control and thus monitoring of all production parameters. A prediction of any quality defects in the product is made on the basis of the interactions. It is therefore possible to make corrections and change settings before a product is completed. This results in massive time and cost savings.

Key goals of predictive quality

The general goal of predictive quality, or predictive quality assurance, is always to optimize quality by using predictions as a basis for making decisions on further measures. But two main goals in particular can be identified: automated quality control and process parameter optimization.

Automated quality control enables an assessment in 50 to 600 milliseconds, depending on the respective requirements, and of course absolutely outperforms human judgment. It is optimized for costs and benefits, and the AI model also includes general quality requirements.

In process parameter optimization, a model learns the optimal model parameters and applies them directly. Depending on the material or component, for example, optimum individual parameters are used. Process parameter optimization enables automatic parameter adjustment for minimized waste.

Deep learning AI model using technical plastic film production as an example

At the company Toray Plastics Another impressive example of the benefits of predictive quality can be identified. Toray Plastics manufactures a wide range of technical plastic film products for industrial applications. Every application, every customer has its own specifications and requires the corresponding ideal operating conditions for the system. Each film break costs tens of thousands of dollars.

With an IoT Approach 400 key figures were recorded in real time. The relevant key figures per plant could be identified and evaluated using a deep learning AI model. As a result, production teams receive continuous status reports on possible sources of error and a risk analysis for film breakages. Adjustments can be made accordingly quickly. Automated quality control has improved plant efficiency from around 85% to over 90%, and over 500,000 euros are saved annually in material and operating costs.

Standardised approach for AI-powered predictive quality solutions

Production managers, site managers and predictive quality experts agree that the medium-term goal must be better product quality through improved process quality. Every company does set highly individual parameters for optimising its products. However, a basic requirement for increasing product quality is a standardized approach to collecting and further processing data. Such a standard enables the rapid and cost-effective implementation of a proof of concept, which can be used to check the effect of the planned measures in advance.

Data Management However, the tasks of production and site managers are not originally part of the tasks of production and site managers. The collection, processing, analysis and use of data is an independent area of responsibility that can also be easily taken over by external specialists if there is a lack of internal resources. A trained data team, consisting of Data Consultants and Data Engineers, collects data and tests AI solutions In ongoing operation and achieves quick and in-depth results even during this test phase.

Predictive quality processes require cloud-based background architecture

Another important component for establishing predictive quality processes is an individualized background architecture. Systems are constantly changing, for example by adding new machines, introducing new production lines or adding suppliers to the portfolio. So if companies only build for today, they will already be left behind tomorrow. Self-learning systems, on the other hand, are constantly adapting.

The background architecture should be cloud-based. This allows it to be organized flexibly and costs associated with data processing are only incurred for really necessary actions. The costs for the entire system are assigned to individual production lines according to actual expenditure and therefore remain visible. Maximum transparency is therefore ensured.

Challenges for initial projects with predictive quality

Data collection, data preparation and data combination are fundamental topics that must be organized in advance of any predictive quality project. This is because data is the basis for every learning algorithm. The article describes how to overcome the associated challenges”Predictive maintenance as a project” in detail.

Further components for successful predictive quality projects

Predictive quality is based on company-specific requirements and therefore requires a detailed look at what is feasible and meaningful. The following components of a corresponding project depend on numerous factors and must be re-evaluated again and again.

Automated quality control

Automated quality control is not always relevant and applicable. Their use is particularly interesting when quality control is visual and has not yet been automated.

Machine learning

Machine learning is suitable for identifying the relationship between machine parameters, settings and external parameters. In this way, complex patterns can be discovered. An Excel analysis, for example, does not reveal any laws. AI can do that very well. This results in a prediction of product quality.

reversion

One new AI can learn appropriate settings for industrial plants in order to provide required key figures. Both quality indicators and general process indicators can be set as goals. In this way, a plant can not only deliver the required quality, but at the same time keep maintenance costs low.

And now? Proof of concept as an ideal project start

If the decision for predictive quality is clear, although there is still uncertainty about the existing or required framework conditions, a test balloon is worthwhile. As already mentioned, a proof of concept is prepared and implemented relatively quickly, integrates the topic of data organization and reopens it as needed, quickly shows initial results within an average of four months and is within a calculable time and budget.

‍