Advancements in robotic welding

The next frontier of the Industry 4.0 movement is the concept of artificial intelligence (AI). Now a welding system can provide valuable data about the welding equipment and process, as well as the robot and various supporting devices

When the first robot went into production, there was most likely a person tasked with its maintenance and performance. Back then the typical scenario was that the equipment continued to do its task until there was a fault or failure. When that occurred, someone in the area notified supervision, who in turn notified the assigned maintenance person about the issue. The maintenance person then investigated the situation, and if he was able to resolve the problem easily, the equipment was up and running again. If the issue could not be resolved easily, the maintenance worker assessed the situation to determine which parts and tools were needed, and then scheduled the downtime and repair—most likely on top of an already busy workload.

Enter the Internet Age

Fast-forward several decades to the internet age, which provided a paradigm shift for factory automation. Robots were designed with the latest internet technology and arrived from the factory with Ethernet ports built into their controllers. Some of the major benefits of Ethernet were its relatively low cost, decent immunity to noise, and good data transfer rates that were very reliable when used properly. These features facilitated having all the robots within a factory included in a local area network, or LAN. Using the original down robot example, factoring in the connectivity benefits provided by the internet, it became much easier to diagnose and repair the robot. The maintenance person received status notifications from a supervisor or from the down robot itself via an email. The email included details about the program that was running and the errors posted on the controller. If the root cause of the issue was not easily identified, the maintenance person remotely logged into the robot controller from the PC in his office or anywhere else within the factory, as long as it was on the same network. Navigating through simple webpages made it easy to examine the program and determine where it stopped. Maintenance also looked at the history or trend of various types of errors to identify a possible pattern.

In addition, maintenance looked at the numerous states of the I/O to see if a particular clamp was open or to confirm that a part proximity sensor was on. After making an intelligent assessment of the situation, he then proceeded to the physical robot on the factory floor with the correct tools and parts required to address the down robot. Alternatively, if he could resolve the issue without tools, the local machine operator was guided over a hand radio or phone on how to resolve the problem while the maintenance personnel watched the operator on the teach pendant in real time from his PC and guided him through the troubleshooting steps.

Another major advantage of having robots on a network is the ability to perform regular system backups easily. Before Ethernet, someone had to physically go to each robot controller and manually back up the robot on a memory storage device like a floppy disk. Floppy disks were small in capacity and communication rates were slow. It often took over an hour to back up a complete robot system onto multiple floppy disks, and the management of multiple disks for multiple robots became a job in itself.

Later, with better data transfer speeds, it was possible to use a PC on the network to back up the robots faster and on a larger, more robust storage device. This allowed backups to be scheduled more frequently during offshifts or on weekends to capture regular programming updates.

As Ethernet technology progressed, so did its speed, security, and reliability. Now the local networks can be part of larger networks and permit access from outside their facility. These improvements ushered in a new era in which peripheral equipment such as welding power supplies, lasers, CNCs, and other machine tools could communicate with the robot over Ethernet. Not only can welding power supplies communicate over Ethernet, now the information from that device is also available to the robot, which in turn makes the information available to the user through the robot connection.

This communication allows error messages or faults from welding equipment to be posted on the robot and gives the operator more information to help identify the root cause of the problem. This also minimizes the number of required Ethernet connections to put all of the equipment on a network.

Artificial Intelligence

Today we are in the midst of the Fourth Industrial Revolution, also known as Industry 4.0. This revolution capitalizes on the idea that all industrial equipment is communicating over Ethernet and providing real-time data to a collection source for documentation and data analysis. For simplicity, let’s focus on the areas of realtime data processing, machine learning, and artificial intelligence (AI).

Real-time data processing allows you to monitor your welding applications for process anomalies, for example. The robot, data collector, or even the welding power supply in some cases can be configured with predefined limits for each welding process. For example, if your weld parameters command 24 volts and a current of 220 amps, you can monitor these values while welding and set predefined limits. These limits can be set in specific units like volts or amps, or a percentage value. When one of the values exceeds the limits for a predetermined time, the data processing system can send a notification, stop the process completely, or identify the part for post-inspection or possible rework, but ultimately prevents the part in question from getting to your customer.

Machine learning also uses real-time data collection to make determinations based on data trends. Using the prior example with data points of 24 volts and 220 amps, imagine that after a short period of time the trend shows about a 2-amp drop at the end of every week. Machine learning identifies the trend and, to help avoid issues, makes recommendations like change out the torch tip, replace the wire line, or inspect wire guides for slipping. This data also may show that the wire drive system is starting to pull more amps for a given process, further supporting the notion a wire delivery issue is pending. Similarly, temperature changes in electric motors can be monitored to indicate when lubrication properties of the oil or grease are starting to break down within the gearbox.

Placing your robot on a network makes it available for remote monitoring and preventive maintenance.

The next frontier of the Industry 4.0 movement is the concept of AI. This is where the real-time data is collected, possibly pre-analyzed by some of the machine learning algorithms, and then further processed using AI.

Now a welding system can provide valuable data about the welding equipment and processes as well as the robot and various supporting devices. The intent is to gather information and use it for predictive maintenance versus preventive maintenance.

With all of the available real-time information, you can assess the health of the system and not just the individual components. If you were to monitor the temperature of an electric motor and the current used to move that motor, you could create a signature of that motor’s health while doing that task. When the signature is not consistent, AI can make predictions about events like when the bearings might fail (if nothing is done to stop it).

This type of prediction allows you to proactively schedule downtime for maintenance and repairs before the equipment suffers from failures that would stop production. The ability to collect and store system information on remote servers or in the cloud makes it possible for the equipment manufacturers to access analytics and monitor performance off-site.

Manufacturers can in turn access that information from anywhere in the world and visualize the health of their individual factories based on machine performance. This information then can be used to assess machine utilization and identify opportunities to insource more work, further justifying that machinery. Having all of this information consolidated and available on a server or in the cloud allows remote data access using a PC, tablet, or smartphone.

Get the Most From Your Robotic Welding System

When purchasing a robotic welding system, one of your main decision factors is the value that the new system will provide your facility. When that new automation system is set up on your shop floor, it needs to operate at peak efficiency with minimal downtime and maximum output.

Placing your robot on a network makes it available for remote monitoring and preventive maintenance. More importantly, it opens up a whole new world in terms of information and benefits that incorporate machine learning and AI, including eliminating unexpected downtime, optimizing maintenance costs, and extending the life of your capital assets—all leading to maximized profits and a competitive advantage.

Terry Tupper was a staff engineer at FANUC America, 3900 W. Hamlin Rd., Rochester Hills, Mich. 48309, 888-326-8287, www.fanucamerica.com.