What is intelligence, anyway? It is the ability to acquire knowledge and skills and then apply the knowledge and skills to solve complex problems, think abstractly, and comprehend multifaceted ideas in a logical fashion. Natural intelligence is found throughout the animal kingdom. One example of intelligence is the act of a mother lion teaching her cubs how to hunt. The desire to hunt is innate but the skill set to perform well needs to be taught. Within the natural world this type of education is called tribal knowledge. It is knowledge that was acquired by one generation and then taught to the next generation. If there is a mass extinction within a local population, then an entire history of tribal knowledge can be lost.
Artificial intelligence is the ability for a computer to mimic the thought processes of the human mind. When considering the impact artificial intelligence can have on gearing systems, we should first look back at the history of how gears are monitored and controlled.
Before steam power and electric power, geared systems had to be operated by human or animal power. These systems were monitored by the human operator and the only measuring tool for confirming the proper operation was the human eye. With the introduction of steam power, gear systems could be operated independently and physical constraints could be incorporated into the design in order to prevent malfunction. One example of this would be to include a gear within the system that would fail if the torque generated exceeded a certain value. This sacrificial piece would prevent the gear system from causing greater damage to the overall machine.
With the introduction of electricity, gear systems were able to be better controlled. With the introduction of PLCs (programmable logic controllers), optical encoders, and lubrication monitors, machines and the gears that drive their operation are able to be monitored in ways that can predict failure, measure absolute position, and predict when maintenance is needed. These tools are a form of artificial intelligence. However, they can only measure activity and report back the results. True artificial intelligence would go one set further and take action in the results.
Another form of artificial intelligence in gearing is the use of simulation and finite element analysis software. These tools allow a gear designer to create a digital version of a gear system, and apply motion and forces to the system in order to gain knowledge about the probable outcome of the fit, form, and failure of the proposed design. The limitation of these tools is that they do not offer a solution if there is a fit, form, or failure discovered. Again, true artificial intelligence would make changes to the system in order to allow for the desired outcome within the prescribed constraints.
In order for artificial intelligence to work, millions of data points and outcomes need to be available for a computer to learn what the proper response should be. Currently, such a data set does not exist. When such a data set is available, gear designers should be able to ask the system to propose a gear system that will operate within the desired parameters and constraints without the need for tedious calculations and simulations. Such a data set may also learn that the current involute profile is not the best tooth profile for certain applications and a new tooth profile will emerge as the preferred standard.
Currently we have 3D printers that can replicate a gear from an existing product or from a CAD model. With artificial intelligence, it could be possible for 3D printers to not only replicate a broken gear but to analyze the failure and construct a replacement that would not have the defect that caused the sample gear to fail.
The concept of using artificial intelligence in gear design and gear applications is something that will improve gear systems in the future. Unfortunately, much like the magic material, unobtanium, artificial intelligence in gear is many years away from actual real-world use.
