Griffin Open Systems – Glossary of Commonly Used Terms
Published: March 21, 2023
We’ve compiled this list of terms in an effort to make commonly used language understandable by individuals at all levels of an organization.
Adivarent Control – The Purdue Model of ISA-95 has been widely accepted across process industries as the gold standard for the layout and architecture of process control & enterprise integration since the mid-1990s. However, this does not easily accommodate the proliferation of edge computing devices and resources operating in the space between control systems and the operator. Adivarent Control is a term coined to represent this layer which is used to assist both the control system with complex tasks and operators with knowledge-based responses that require constant attention.
Artificial Intelligence (AI) – Broadly speaking, this encompasses several areas where computers control devices (e.g. actuators, robots) to do tasks or adjustments normally done by humans. These types of AI tasks require an ability to discern patterns and correlations, not all of which have been seen before, and react to changes in processes or data sources with the ability to predict proper responses. AI can be used as advisory-type systems or as actionable real-time control systems.
Data Scientists – A new branch of technical experts whose expertise is using a variety of analytical tools to explore complex or large data. This was formerly the domain of statisticians, engineers, and mathematicians. Many data scientists have these backgrounds, but the complexity and size of data (big data) have made it necessary to learn new tools and techniques to gain additional insights into data correlations, trends, diagnostics, and where to focus for process improvements.
Data Analytics – A set of mathematical techniques and algorithms for examining data sets in order to gain insights into the process being examined. There are a number of specialized tools which may be applied, depending on the desired goals for the analysis. These tools are broadly categorized in the areas of trends, diagnostics, predictions, prescriptive and cognitive (AI). As data analytics covers particular expertise, this type of data analysis is often performed by data scientists, though advanced usages may have automated some aspects, especially in the sub-domain of manufacturing data analytics.
Digital Factory – A generic term with a wide range of meanings and a range of implementations. In general, it will refer to a centralized organization of digital data which is then used to create dashboards for users to track production processes, data analytics tools for investigating the production process, and digital models for continuous evaluation and optimization of the production process.
Edge Computing – This term has become popular with the advent of the IIoT (Industrial Internet of Things). With cloud computing, all computation happens on the cloud which can have some latency and bandwidth issues. To address this, Edge computing is used to set up computing resources closer to the data source to provide real-time analysis. The Edge computer can still connect to the cloud for other services. In the world of IIOT, the edge device will not only collect data but process the data to make it edge computing. This makes data more readily usable by dashboards, data scientists, and implementors of real-time digital factory systems.
On-Premise Computing – This term refers to computing resources set up at the customer’s premise. An on-premise solution is used when data collected is considered sensitive to be trusted by a third-party cloud provider or when the internet cannot be due to the risk of cyber-attacks.
IIoT– Industrial Internet of Things refers to the use of smart sensors and actuators networked together in an industrial application. Control systems, in concert with edge computing devices, allow more sophisticated analysis and processing of data for improving the operation of manufacturing facilities.
Manufacturing Data Analytics – This represents the specific application of data analytics for use in operations for process and product manufacturing industries. They can be applied to improve (ensure) quality, increase throughput, improve efficiency, and optimize the production process. The analytics may be done offline by data scientists or may be incorporated into actionable algorithms modifying the production process to achieve the best circumstances for the given conditions.
Machine Learning (ML) – A sub-field of artificial intelligence (AI), with the goal of providing the capability to machines to learn from data and, in many cases, imitate intelligent human behavior. While AI is useful in learning complex tasks performed by humans, it is the ability to learn these tasks without (or minimally embedded) first principal models that make them attractive. Furthermore, the ability to quickly see multi-dimensional relations and trends offers the potential for high ROI projects when looking to improve operations.
Particle swarm optimizer (PSO) – A biology-inspired search algorithm used to explore many solutions with a defined solution space. The algorithms require an objective function that defines the goals or defines a ‘most fit’ solution. The PSO creates a number of possible solutions (particles) that are evaluated in an iterative process. The closeness (distance) to the objective is shared among the particles to help guide them toward the more successful result(s) and eventually to a near-optimal solution (ideally the optimal solution). The inclusion of random particles helps avoid local optimum solutions.