Make-Sense-of-your-Data
Search
Close this search box.
Menu-Header

AI Consulting - Making Sense of Your Manufacturing Data

Industrial-Revolutions

Throughout our recent history, we have experienced numerous industrial revolutions. The first one brought us mechanisation through steam and waterpower, whilst the second one in the late nineteenth century brought us electrification and mass production.

In more recent times the third revolution, in the 1970s brought us automation with increasingly powerful automation controllers and networking infrastructures.

We have now, arguably, entered the fourth industrial revolution. This revolution isn’t just about improving productivity, it also provides benefits from:

  • Flexibility of manufacture
  • Reduced time to bring new products to market
  • Increased quality
 

Industry 4.0 Analytics Cyber Security

Unlike the previous revolutions that were based on one or two underlying technologies, this revolution is based on numerous supporting technologies.

Perhaps the question to ask is what technologies are manufacturers mostly investing in?

Primarily there are two. Firstly cybersecurity. This is hardly surprising since almost every week there is news of another cyber or ransomware attack. Secondly, it is data analytics to improve real-time decision making.

Although there is much talk about big data, which is often accompanied by a significant price tag, many manufacturing problems, due to advances in processing power, can be solved by relatively modest hardware and software running on-premise.

The adage that knowledge is power, should now more accurately be replaced with the phrase data is power. Business decisions are no longer made on gut feelings or experience but on empirical evidence. This empirical evidence is provided by data science which includes both artificial intelligence and machine learning. For a high level description see our blog post Artificial Intelligence vs Machine Learning.

Automation Maturity

Many of us will have a connected automation architecture consisting of PLCs with SCADA systems providing the HMI. It is also very likely that these SCADA systems will be logging data to a process historian or SQL database. Thus, should a failure occur, we would be able to retrospectively review the process prior to the failure.

The question is whether we use that data for anything else, or does it just go to the historian to die? Is this an acceptable approach? Today, advances in analytics allow us to use this data together with data science to provide insight into our processes to further optimise them.

Data science isn’t particularly new, but what it has lacked is data to consume. This dearth of data is no longer true. Network speeds have increased and the cost of acquiring and storing data has plummeted. In many cases, it would be true to say that we are now data-rich but information poor. This lack of insight can be addressed by data science.

The-Role-of-Data-Science

What Types of Problems Can Data Science Solve?

Problem Solving

Data science is applicable to solving a wide variety of problems in manufacturing. Examples include:

  • Identifying process or equipment abnormalities by establishing their normal operating envelopes. Should they drift outside their envelopes then an alert is generated.
  • Identify hidden casual relationships between manufacturing data. Thus providing additional insight. This in turn can be used to improve SOPs, create soft or virtual sensors that can be used to ‘measure’ a particular characteristic for which no inline sensor exists. Alternatively, such insight can be used to develop advanced control strategies that automatically adapt to changes in upstream conditions.
  • If we can start to predict outcomes in a robust manner then we can start to optimise our processes. Frequently our processes behave in a non linear manner with competing goals. For example we wish to maximise our production rate whilst ensuring that we meet our quality specification. In this example we could use our ability to predict the outcome as an input to an optimiser with the objective of maximising our production rate whilst ensure that the constraint, our quality specification is achieved.
  • Today we are all familiar with using a web based search engine based on crude key word searches. However all our intellectual property, is typically scattered across multiple document types (Word, PDF, email, presentations etc.). One of the challenges is making this information available to our workforce in a way that allows them to query it using  natural language to provide a coherent and contextual response.  Artificial intelligence allows us to address this requirement by using Retrieval Augmented Generation (RAG). For more information on how RAG may benefit your business see our blog article From Rags to Riches.

Why AI Consulting?

Data-AnalystAI Consulting is a small, UK registered consulting company, that has a substantial background in automation, data analysis and manufacturing.  It provides professional data science consulting services for manufacturing and only manufacturing. We can respond rapidly to new challenges, and you will always be speaking directly to a professional rather than a non-technical sales representative.

We cover all aspects of any data science project from discovery through to site deployment. As we speak your language, you can be sure that the solutions that are presented will be practical and easy to understand, and that we will be able to deliver real value for your company.