Over the last few months, we’ve been quietly monitoring and discovering new use cases for our predictive machining product. Thanks to a streamlined operator interface, which also conveniently doubles as a “behind the scenes” labeled-data collection platform, we’ve seen hundreds of tool failures, bearing failures, and other machine failure scenarios thus far.
In concert with our customers, we’ve come up with a large catalog of what typical machine failures look like from a motor-data perspective, which we’re excited to share with you today. We’ll review six scenarios we’ve seen repeated across multiple customer sites, and what they look like from…
At MachineMetrics, we chose to make our company logo green as a kind of winking signal of productivity. Green means “go,” and within the MachineMetrics system, it also means “on target” for production goals.
But what about Green in the environmental sense? What are the concrete ways that we are improving shop efficiency and reducing waste? And how is that coming back to benefit our customers in terms of that other kind of green: money?
A dramatic illustration has come from a recent release of the MachineMetrics software, which allowed us to deploy a solution that reduces scrap parts…
In our last blog post, we likened machine PLC (programmable logic control) data to being able to hear the complete symphony of notes from the machines. We recently discovered that our metaphor wasn’t far off…we can quite literally hear the sounds machines are making through the raw PLC data we’re collecting, without any microphones. Using methods from digital signal processing, these are some examples of pure machining sounds we extracted. Feel free to poke around here.
This is “noteworthy” 🥁, as one of the most common refrains we hear from machinists is that before…
Why High-Frequency Data?
Imagine you’re trying to learn a new tune on the piano, but the sheet music only has one note out of every ten. Wouldn’t that be pretty hard?
That’s what it’s like learning what your machine is doing with data that only plays a few notes from the entire piece. Just like Chopin, a machine writes complex, nuanced melodies that get shortchanged when an incomplete picture is represented in the notes it plays: the data.
And sometime in the course of the last few years, we realized that the sampling frequency of our normal adapters was great…
In the past few decades, data for macroeconomic indicators have largely been collected through surveys or inferred through secondary means. Specific to manufacturing, key metrics for industry health, such as industrial production and capacity utilization, have been collected via polls of manufacturers and estimated from import/export volumes.
This clearly has its drawbacks — chief among them an “information delay” due to the time it takes to survey, collect, and report on this data. The consequence is that market movements coincide with the release of this data; the stock market jumps up and down, often at the end of the month…
A major challenge in manufacturing is predicting how long cutting tools in your machine are going to last. They get worn out eventually after steadfastly grinding through really hard metal, creating part ($$) after part ($$) for you.
Tools are “rated” for a certain life by the manufacturer, but this varies highly based on the type of material being cut and ambient conditions in the machine shop. In response, operators have developed some rules of thumb to determine when to change them out.
In the fourth installment of this series, we’ll take a look at some examples of anomalies we caught in live machining environments. We’ll then discuss how we put this into production.
If you haven’t read the rest of the series, you can start here:
We identified five distinct categories of anomalies.
Example 1: Preceding Tool Failure
The original purpose of taking this approach is because we hypothesized that machines experience…
In parts 1 and 2, we discussed the business problem and preprocessing involved with detecting anomalous behavior on machines. In this post, we’ll cover some creative data wrangling and clustering methods. This piece will be more technical in nature than the others.
Isolating Part Signatures and Creating Transformations
Once we have our clean signal, we need to split that signal up into its individual components, the part machining signature. Each part machining signature represents one part being made and the corresponding positions, feeds, speeds and loads which are attached to it.
We take each signature and line them up next…
In the last post of this series, we went over why it was important to try and detect anomalous behavior on machines. In this post, we’ll dive right into how we preprocessed and cleaned the data.
We tried this method on many machines, but to illustrate our points, we’ll single out just one example. We’ll begin by looking at what the machining process looks like from our standpoint. Below we plot streaming data for feeds, speeds, and load for one particular machine from 10 PM to 9 AM. This machine is making the same part the whole time.
In the world of Computer Numerically Controlled (CNC) machining, many operators struggle to keep their machines on schedule and running efficiently. Unanticipated events can plague machine performance, and a significant portion of these surprises result in costly repairs with a high opportunity cost. Machine tools can cost upwards of a million dollars, with raw materials and labor adding significant variable cost per hour as well. Downtime is disruptive, expensive, and downright frustrating.
We realized this was a huge problem for our customers, and that just monitoring their machine performance wasn’t enough. It became critical to try and leverage our database…
Pioneering research in manufacturing data science