Introduction to Machine Data Acquisition in the Plant

[ayshakhatun3113]

Finding patterns in data is a task that keeps data scientists and data experts alike busy. For organizations, data is a critical business asset that adds value. Plant connectivity plays a key role in this.

This blog post uses a (fictional) example to explain the role of connectivity in the plant.
Finding patterns in data is a task that keeps data scientists and process professionals busy. Industrial companies are interested in this because pattern discovery can lead to process optimization (for example, identifying the best recipe or set of process parameters), identifying the causes of anomalies, forecasting materials for supply management, and predicting energy consumption. Data is therefore a critical business asset that can add value. A first step is the collection of machine data .

Plant connectivity plays a key role in this, as it enables access to machine data . Data can then be stored, processed, analysed and modelled. Ultimately, the results of data analysis and modelling can be used to optimise the process and create a competitive advantage for the company.

An example
The plant manager has returned from the part time data latest industry fair. He reports to his team: “I saw a nice demonstration of data collection with machines at one of the stands.” Demonstration at a trade fair? Typically, it’s a (very) simple use case with a programmable logic controller (PLC), some sensors and a nice dashboard to display variables, trends and some indicators. The plant manager gathers his engineering team. The engineering team does a great job to successfully execute projects in the plant. Typical tasks for the engineering team? Upgrading electrical infrastructure, installing new machine lines, replacing automation components. The manager asks for a pilot project on one of the machines for the next quarter. The engineering manager scratches her head – it’s a nice change from the daily grind and a good opportunity to learn something new, right? The project is defined.

The engineering director calls her team together and they immediately start brainstorming. Three or four months is a short time, right? The team approves the project. Okay, the wheels are in motion. They come up with a plan and set key milestones:

Use case proposal (e.g., definition of objectives, success measures, budget, global scope and/or milestones).
Evaluating machine connectivity (e.g. communication cards in PLCs and industrial PC (IPC)).
An initial architectural design for data collection (such as an ad hoc data collection backend and dashboard).
Execution plan (e.g. in the form of a Gantt or Agile chart).
Stakeholders agree. This time we will focus on assessing machine connectivity. It's a big cake (this digitalisation thing), one slice at a time please.

Below we analyze the topic in detail.

Workshop connectivity
The data area belongs to two different worlds: On the one hand, we have the plant with all the machines and devices. That is where the data is created. On the other hand, we have the realm of data science, where the data adds value to the business. Alright! That’s not all. We have other parts, such as the quality control system, enterprise resource planning (ERP), manufacturing execution system (MES) / distributed control system (DCS) / supervisory control and data acquisition (SCADA). As described above, data scientists look for patterns in the data. The most important step in this process is the acquisition of the raw data. Sometimes it is the CSV/XML file provided by the machine or the query of a data point directly from the machine via a communication protocol. Although the path to the data is long, the first milestone is to ensure the connectivity of the machine or the plant. Some of the most common challenges encountered in this activity are: