Machine learning is one of the hottest topics in tech-based industries right now. Engineers and scientists are pursuing machine learning for everything from developing driverless cars to making smart appliances. In the composites industry, the promise of machine learning is one of bringing down the cost of manufacturing by making processes more efficient and less reliant on human labor.
The promise of machine learning for the composites industry lies in the ability to create automated manufacturing systems. But is that actually possible? Yes, it is. Though composites manufacturing is not nearly as automated as something like injection molding, the industry is slowly but surely getting to that point. Machine learning will begin playing a more important role in the future. It’s just not there yet.
The Difference between Machine Learning and AI
The promise of machine learning for composites manufacturing is just that; a promise of something in the future. It’s not yet ready for prime time despite what you might read in the media. Here’s why: the terms ‘machine learning’ and ‘artificial intelligence’ are used interchangeably in blog posts and news articles. They are not the same thing. But because they are used interchangeably, a lot of people talk about machine learning when they are really referring to artificial intelligence.
In the simplest possible terms, artificial intelligence (AI) is the ability of a computerized system to use pre-programmed data to make ‘intelligent’ decisions. An artificially intelligent machine tracks its own actions, compares those actions against an existing data set, and then make the best possible decision based on the results of the comparison. The key here is to understand that AI relies heavily on pre-programmed data sets.
Machine learning is an advanced form of AI that starts with a limited data set that ostensibly grows over time. The machine truly capable of learning gathers its own data over time, adds that data to its pre-programmed data, and then uses the combined data set to make decisions. Furthermore, true machine learning involves purposely going out and finding missing information.
The reality is that true machine learning does not yet exist. Technology is moving toward that point, but it still has a way to go. Machine learning will inevitably mean great things when the day of its reality finally arrives.
Machine Learning in Composites
A big hindrance to widespread composite adoption is cost, explains Utah-based Rock West Composites. So much so that the number one goal of the industry right now is to reduce costs through automation. Both AI and machine learning are a big part of the strategy.
On the AI front, we are already seeing some impressive applications. For example, a joint project being conducted in Europe is utilizing predictive analysis to improve composites manufacturing by minimizing variability within the manufacturing process. The project has resulted in an actual machine that can produce composite molds that greatly reduce the need to rework, scrap, or repair the finished product.
Although still in the development stage, this proof of concept machine is an example of the promise of machine learning for composites manufacturing. If using predictive analysis makes it possible to create composite molds at a fraction of the current cost, there would be a subsequent drop in the cost of fabricated parts as well. This is what the industry is after.
We are still a long way off from that day when machine learning dominates composites manufacturing. But that day is approaching, and it may arrive more quickly than anyone imagined. We look forward to what machine learning will do to transform the composites industry.