Robots in production lines work with micrometer precision, unless a component fails. If, for example, the linear actuator used to precisely position a car body in front of an assembly robot is damaged, the robotic arm will no longer be able to position the car door exactly. The result is a misaligned door.
A system was developed comprising intelligent sensors that continuously collect a wide array of measurement data from inside plant machinery, and compare the signal patterns against those for normal operating conditions. If the system detects a difference in the patterns that indicates a potential fault, it immediately notifies the equipment operator about what remedial measures should be taken. This helps engineers plan maintenance more effectively, and protects them from unpleasant surprises and unexpected production losses.
The system subjects robots to what is effectively a continuous medical checkup. The human equivalent would be equipping a person with an activity tracker, a continuous digital ECG, and blood pressure monitor so their state of health could be analyzed at any time. The system enables operators to continuously visualize the current condition of industrial robots, and provide advanced warning of potential damage.
Sensors are installed inside the machines, and interact with each other and with existing process sensors. Industrial equipment will begin to make a different noise, or will vibrate or overheat, long before it actually fails. The trick is that the characteristic manner in which a machine hums or vibrates during normal operation is different than that observed when something has changed within the machine; these differences can be very subtle and undetectable to normal senses.
The sensors can detect these slight changes, and can assign them automatically to specific fault profiles. Signal patterns, such as the frequency of vibrations, alter during common damage or fault states. The research team examined the patterns in thousands of measurement datasets and identified those associated with particular types of damage or mechanical wear. That information is fed to the sensors, transforming them into smart devices that detect the signal differences on their own. This essentially eliminates the need for an external analyzer, as the system is able to perform the analysis itself.
The goal of the research is to develop a set of sensors and modules that will allow companies that operate industrial robots to put together a fitness check specifically tailored to the needs of their plant or equipment. The customized sensors can either be integrated into the machinery when it is being made, or can be retrofitted. Initially, the sensors spend their time collecting baseline data that reflects the normal operating state of the machine. Once that has been done, the system is ready to continuously compare the current operational data with those typical sensor signal patterns associated with incipient equipment failure or damage. The system can also be used for quality control purposes by analyzing whether production machinery was operating properly during a manufacturing process.