Every factory is obliged to perform regular quality inspections. On one hand that’s because of legislation and ISO-norms. On the other, it’s because you simply need to in order to stay operational -especially when working with delicate production like food or plastics. For every factory, it’s a nightmare scenario when items have to be withdrawn from shops or even from consumers, because of poor production. Quality inspection is key in avoiding this, and with AI combined with machine learning you can make this inspection more robust.
There are many ways how to perform quality inspections. For instance, one of the basic ways is sample inspection. With this method, you select at random e a few pieces of chicken, fries, metal plates or car toys and see if they satisfy the requirements.
Costly way of inspection
This method has it’s merits. If there is a fault in the production there is a fair chance you will find this by doing a sample test. However, once you’ve discovered it, you probably have to discard the whole production from the last test on. This is not only costly because a lot of production can’t be sold, but also storing and / or demolishing the production is a pricey affair.
How can automated quality inspection make a difference?
With real-time quality inspection based on AI and machine learning, this can be hugely improved. It works as followed:
- According to the type of production, it will be inspected with camera’s, infrared light and/or other sensors.
- Trained on your historical data an algorithm will be able to spot defects on your production line in real-time
- Once this is discovered, the whole production line Will be automatically shut down, so your production line will not deliver defected or damaged goods anymore.
Fourth industrial revolution
Industrialists are calling the use of AI and machine learning the fourth industrial revolution. It radically changes the way in which factories are producing goods. It increases the need for digital know-how and ICT and reduces the need for low skilled labour.
Not easy to implement
Because of this ‘revolution’, it can be difficult to find the right people to do this job. Next to this it is a relatively young field which is still in rapid development. And finally, it’s not always easy to implement real-time quality inspection in such a way that it really works and adds value.
Setting the data right
An algorithm needs data of both good and poor production to be able to decide what’s right and wrong. However, over 99 percent of everything you gather is probably good quality. Data of poor production is scarce and needs a specific approach to make it useable. To make it worse, it can also easily be polluted, leading the algorithm to give a distorted outcome.
Building IT together
Implementing real-time quality inspection the right way is like crafting a Swiss watch. This means you will need a ‘Swiss craftsman’ to design, plan and implement it to make real-time quality inspection work for your situation at your company. At LINKIT we follow a pragmatic approach in which we try to cooperate with your business staff and IT-department as much as possible. Combined with our own ‘eco-system’ of over 1000 highly experienced IT-experts and our agile way of working, this will result in straight forward solutions with maximum flexibility.
Do you want to know more about implementing automated quality inspection in your factory? And how it is done? You can read it in our free to download whitepaper. Or call one of our experts for more information.