Imagine such a scenario: you have a production line that has thousands of milling machines, and you notice that you are losing money every time a machine breaks down and lower the production rate. What can you do to solve the problem?
You asked your engineers to regularly check every machine and either repair or replace machines that may have problems to prevent future loss. In the beginning, this method worked well because there were indeed fewer machines breaking down. However, you found that you needed to spend extra money on checking, repairing and replacing those machines, which in the end did not help you save and earn more money.
In the first scenario, you apply reactive maintenance, which means fixing machines when problems occur. The method you tried in the second scenario is called preventive maintenance, which means regularly repairing machines to prevent future failure. A third method you can try is predictive maintenance.
What is predictive maintenance?
Predictive maintenance is a condition-based technique that aims to forecast the performance of equipment or systems and find the unusual case so that people can fix the problem with minimal cost.
Predictive maintenance is nothing mysterious as fortune tellers and their tarot cards or crystal balls, but based on collected data, such as working temperature, rotational speed and tool wear of machines. For example, if a worn milling machine is at a higher than usual temperature and speed, which historically causes machines to break down, then most likely this machine may soon fail as well, and it would be better to repair it to avoid future loss.
In reality, predictive maintenance will of course not be as simple as the above example. Building a reliable predictive maintenance model often requires large amounts of data collected with many different dimensions over a long time span, complex algorithms and proper matrix to evaluate the performance of the model.
What are the benefits?
Then why is it still important to apply predictive maintenance? The major benefit is reducing costs, in different ways. Returning to the scenario we mentioned at the beginning, predictive maintenance can help maintain the production rate and reduce potential loss during downtime, compared to reactive maintenance. It also helps to reduce the maintenance cost, including human resources, of preventive maintenance. Besides, by avoiding catastrophic failures, predictive maintenance can increase the safety and service time of machine, which will in return save extra cost as well.
Decision Labs can help you get the best benefits of predictive maintenance. Based on your collected data, we will build transparent and private machine learning model that not only has high accuracy but can maximize business value.
