“Intelligence” implies “creativity.” The ability to create something meaningful from nothing is an art that will likely forever be exclusive to the human mind. Machines, on the other hand, excel in an area where human often fail. They remember, correlate, and detect complex patterns in vast amounts of data. When confronted with the results of such data exploration we are often surprised and perceive “artificial intelligence”, where really the machine can only impress with boundless memory.
Artificial Intelligence at an industrial scale
Industrial process data with endless streams of time-series real-time values should, therefore, be the ideal playground for these algorithms. For example, detecting data patterns that resulted in machine errors in the past results in the detection of such patterns of impending failures. Voila – predictive maintenance delivered!
Putting AI into practice
However, there are problems in taking Artificial Intelligence algorithms from the laboratory setting to the shop floor. Clean data is required to detect patterns reliably – that is data with no or little noise or discontinuities. The human eye has no difficulty recognizing an “Apple” in 10 pictures of apples, even though they will vary in color, shape, size.
A machine programme will find this task much harder. In the same way, process data is filled with noise and other perturbations, which upsets algorithms. The results are poorly working algorithms and disappointing performance – missed events and false predictions.
Advance noise filtering techniques
We recognized this problem early on and worked hard to develop adaptive cleaning algorithms to first remove these perturbations from the data before applying pattern detection. The result is artificial intelligence in action – avoiding costly downtimes and repairs, and boosting productivity. To learn more, read our technical paper which is to be published and presented at the IEEE World Forum on Internet of Things in Limerick, Ireland, in April.
Fine our what artificial intelligence can do for you. Read the technical paper.
