Just a few decades back, the defence industry was quick to exploit the potential of intelligent machines and systems that could use in-the-field data in order to ‘think for themselves’; primarily to make simple, fact-based decisions that removed the workload from human operatives.
This concept of data-based machine learning and Artificial Intelligence (AI) is now seeing much wider deployment in civilian applications too – with many of us not actually realising that we are the information sources creating the data upon which these systems base their decisions. Leading tech firms, such as Google and Netflix, regularly leverage huge amounts of on-tap user-specific data to feed algorithms that ultimately help them streamline and enhance their customer offerings.
Replace the term ‘user-specific data’ with ‘machine-specific data’ and it quickly becomes clear that Big Data and machine learning have the potential to revolutionise the manufacturing, process and, indeed, any other networked or machine-based industries. In fact AI-driven procedures are already making a difference across a number of industries, with companies such as Presenso and Predikto already providing AI-based analytical solutions for maintenance operations.
Predictive maintenance is a perfect illustration for AI-based systems, as it can leverage a relatively small number of simple but incredibly informative data sets and then use them to predict and provide remedial actions. With the advent of the Industrial Internet of Things (IIoT) and, as a result, Big Data, coupled to an almost exponential expansion in the number of data-providing end points or nodes, this type of capability will become more prevalent, powerful, informative and effective.
However, even a single simple manufacturing cell can generate in excess of 50 points of data per second, and over the period of a shift this can quickly turn into a tidal wave of ones and zeros. This is where AI and machine learning will step in.
As part of Big Data analysis, AI will distinguish relevant data from noise, define logical connections and correlations between these data sets, remove any non-connectable data and then provide pertinent information upon which decisions can be made – either automatically or via human interaction.
The good news is that future AI and machine learning can be catered for now, by any company running any solution. Adding intelligence and data gathering capabilities to even the ‘dumbest’ manufacturing operation is relatively straightforward and can be achieved with a very palatable financial outlay.
Even if the data generated is not leveraged immediately, the solution will be ready for when it can be; and the historical data can be put to very good use.
The chances are that these AI systems and analysis solutions will take the form of cloud-based subscription services, which will also leverage data from other similar entities and applications in order to present the best possible solution.
And, with a younger and more tech savvy, multi-skilled workforce stepping into the shoes of older engineers, data, especially its collection, collation, translation and deployment, is creating the next industrial paradigm. Luckily Big Data does not need deep pockets.
There’s no better time to start collecting than now.
At RS, we already have many systems and initiatives in place to both cater for the demands and exploit the benefits of the Big Data-driven manufacturing economy, and we are feeding these products and support solutions to our customer base. Even the most basic manufacturing data – especially when leveraged intelligently – can make a huge difference and the path to adoption is already wide open and incredibly well supported for applications of any size and budget.