Predictronics had a fruitful experience at the Motor and Drive Systems Conference 2020.

The Motor and Drive Systems Conference joins together leading engineers, manufacturers, system integrators and machine builders to discuss the latest technological advancements in systems design, development and efficiency, as well as motion control for automation and robotics within manufacturing and industrial applications.

Senior Data Scientist Aaron Shelly presented Predictronics case studies, which demonstrate real-world applications of predictive solutions for motor and drive systems, including industrial robots, CNC machine tools, electric motors and tractor motors.

For motors and drives, issues typically arise in other components and subsystems first. By utilizing the data from motor and drive systems, you can better understand the overall system health in order to predict and prevent failures elsewhere. Predictive maintenance solutions provide a significant value for these types of applications due to their scalability, low cost and high failure detection accuracy. Implementing these solutions helps minimize repair downtime, optimize production scheduling, improve product quality and reduce labor and part costs.

Here are five elements for success when developing a predictive solution for motor and drive systems:

1. Don’t be hesitant when implementing new technologies. Solutions are constantly evolving. If you choose to wait for the “perfect” one, you will fall behind the competition. Instead, start small and fail small. For example, when developing a predictive solution in manufacturing, you should begin with a proof of value project or focus on a single asset or cell in a production line. If value is provided in this initial study, the solution can then be expanded and deployed to a wider set of machines at the facility. If the proof of value is not successful, then you have not lost a significant amount of resources or time and you have the ability to re-focus your attention to other potential solutions or more impactful issues.
2. Select a high value problem, but not the most challenging. Machine learning and AI are not magic; they are limited by data availability. You need to make sure you have the correct volume of data, as well as the right signal measurements and data quality. If the data quality is poor, you won’t achieve an accurate solution and fruitful results.
3. When deciding on an application, it is important to determine what information you need in order to achieve an advantage over competitors and differentiate your product in the industry. Enabling motor and drive systems with predictive capabilities can help distinguish your company’s offerings in a saturated market. Some applications to consider include fixing connectivity issues, determining the quality of component adhesives in magnetic motors, and even feeding data back into the design process by creating a digital twin.
4. Make sure your solution is actually predictive. Many companies believe they are garnering enough information from voltage and current threshold data from control charts. However, true predictive solutions utilize machine learning models, which are more useful and robust.
5. When selecting the most optimal application and ensuring you have the correct data and solution, it is best to invest in a team of experts with a diverse background of domain knowledge, as well as analytics skills and industry experience. This can be done by hiring an internal team or reaching out to skilled external solutions providers. By finding data scientists with the right combination of engineering talent and fluency in the industry, you can deploy accurate solutions with greater ease to improve and optimize business operations, saving your company time, money and resources.