Automated Production Monitoring (APM) is a system within MachineMetrics that allows users to configure Operations in a way that Production Runs will automatically be started and stopped when certain program information is reported by the machine.
Depending on the make, model, age, and connectivity of a machine, we are able to capture more or less information from the machine — either by querying the control or through the use of sensors. Automated Production Monitoring only works for certain machines because we need to be able to retrieve program information to make it work — this immediately excludes machines integrated exclusively with sensors (what we sometimes refer to as Digital IO-connected machines).
Note: To use Automated Production Monitoring, your machine must be capable of reporting programs that can be correlated to Operations.
As the machine is running, we are constantly asking it what program information it is running and sending it through our data ingestion pipeline. For machines that have Automated Production Monitoring algorithm enabled, we examine the program number, name, header, and/or file name (depending on the data available) and automatically detect part numbers and operation/sequence numbers that are present within them. When an Operation within the Operations table that has the Part and Operation associated with the machine data found, and that are allowed to run on that machine will be linked automatically and started. When no Operation match is made, a new Operation will be created automatically and named with the Part and Operation# found in the machine data. The previously running Production Run will then be stopped, and a new one created with the new Operation attached.
TOPICS COVERED IN THIS ARTICLE
*Keep in mind: if the machine doesn't have support for one of these fields, it will not be present in the list of options. Below we outline the parameters for using name, numbers, and headers when configuring your Automated Production Monitoring.
Program Data and Headers
The MachineMetrics APM2.0 Algorithm was designed to recognize and determine part numbers from program data. Many companies utilize program headers, also known as comments, to call out specific information about the part and operation being made by the program. Our algorithm can detect part numbers with a high amount of accuracy without any of the below specifics, but for best results, the following examples show common header formatting that best suit the algorithm's behaviors.
(TAPE # OT076.A)
(PART # 1077280_REV_6)
(OP # 110)
Algorithm determined Part/Op: 1077280 OP#110
By labeling what the actual part number and operation number is, our algorithm can make the correct decision with how to proceed in matching Operations that are currently in the database correctly, or by creating a new Operation for you if none such exist.
Unlabeled part information can still be captured and matched on by the algorithm, but the possibility of improper creation of new Operations is more likely to happen than with labeled data. The below example shows unlabeled program information and what the algorithm would determine is its best guess at the part number.
(RP/JT 2-08-23 B0)
(MAT 17-4R .250 DIA)
(CYC TME 3 MINS 15SECS)
(TOOLING IN BROWN BAG)
Algorithm determined Part/Op: B12055
The algorithm made its best guess based on many factors what the part number was, even though there wasn't something to label it as such. With no Op# call out, the algorithm can't determine one, though. So, no Operation number was affixed to the part.
*Note: If the part and operation information in the machine data is labeled, your Operations table information should be labeled as well. Labeled information on both ends is necessary for the matching algorithm to comprehensively know what Operation belongs to what machine data.
Enabling the APM2.0 Algorithm
We built APM2.0 to be as simple to use as possible. To enable the algorithm on your machine, it's as simple as selecting an option from the dropdown in your machine's Operation and Production Settings page.
Common Automated Production Monitoring Questions:
- What if a machine runs multiple programs to produce a part?
So long as the program information contains the same part and operation information within each program, the algorithm will automatically link the programs to a single Operation within the table.
- What if a few machines are capable of running an Operation, but some have different program numbers since they're slightly different pieces of equipment?
You can add machine-specific overrides to an Operation, allowing you to indicate which program numbers are expected on each machine.
- What if the main program is generic across all parts and sub programs or macro variables are called to do the part-specific work?
This use case is yet to be fulfilled by the new algorithm, as it takes multiple layers of information to determine what part is running when. Future iterations of the algorithm will help solve this problem, including families of parts that use a single parent program and macro variable determined child.
- What about machines with multiple paths?
As long as the machine isn't making different parts between the paths, the algorithm will be able to capture the main path of the machine that contains the part information. Machines that produce different parts at the same time are not within the scope of MachineMetric's Production Monitoring service currently.