How Industrial Data Was Born (Part II)
One of the big draw backs of the first automated control panels was the difficulty in changing the logic to modify the control over industrial automation applications. This problem became solvable when the next disruptive technology hit the scene, computers. These computers were nothing like the PCs we have today.
The early days of computers were like the wild wild west. They lacked standards, applications, and dependability. So what spawned out of there for the industrial world was industrial computers designed for purpose specific industrial applications. These machines evolved into several disciplines like programmable controls which became programmable logic controls (PLCs as PCs began to denote personal computers instead), distributed control systems, servo control systems, motor drive systems, and single loop controllers.
I’ll stay with PLCs as one of the dominant technologies used to replay relay control. In fact, in the early days and still in use today is a specific language developed for PLCs was called “ladder logic.” Ladder Logic was designed to mimic relay diagrams to help transition hard wired relay controls into a software driven logic engine. Below is a sample picture of a stop/start circuit in this “ladder logic” programming language.
Now it gets interesting, people started to realize that there was data in that industrial controller as denoted by the memory map table for a GE 90-70 PLC:
Many applications spawned to harness this data and provide different levels of value to it. Many of the starting applications simply read it in real time and graphically represented it to give operators much better information on what was going on with their operation.
In addition to that, other programs were developed to take snapshots of that data and associate it with the time of the snap-shot, hence time-series data. Industrial data was born.
For me, it is interesting to remember that industrial data is basically sensor data hooked up to some form of automation technology. It consists of snapshots of sensor values taken periodically and stored. From this simple concept, we can extract tremendous value, value that continues to grow as our modeling and analytics get better to analyze it. But that is a story for another day.