Many of our customers were already using vibration measurement or vibration analysis at the time of the first discussions with Senzoro. However, since our technology has many advantages due to the combination of ultrasound and artificial intelligence (AI), a rethinking took place.
Significant cost advantages through ultrasound with Artificial Intelligence (AI)
In essence, significant cost advantages can be achieved with ultrasound and AI, since no manual analysis by humans are necessary. The solution is also Industry 4.0 "par excellence", since data is the basis for objective results, on the basis of which the system itself learns (e.g. estimating remaining lifetime more and more accurately)
(1) Subjectivity of the analysis / dependence on experts with years of experience
Vibration data is still evaluated by people who have years of knowledge in the evaluation and interpretation of these data. On the one hand, this is very time-consuming and on the other hand, there is a high dependency on these experts should they leave the company or go into well-deserved retirement. 1:0 for ultrasound with AI, the analysis is fully automated.
(2) Accuracy estimation of remaining life time
The estimation of the Remaining Useful Lifetime (RUL) is not part of the analysis results of vibration data, only the identification of problems and abnormalities. For ultrasound with AI, the derivation of a Remaining Useful Lifetime is a fixed component of the analysis in the sense of Predictive Maintenance. The accuracy about the remaining useful lifetime is currently "good" and will become more and more accurate in the future due to additional training data. 2:0
(3) Simple training of new employees
New employees have to be trained in vibration measurement both for the very specific measurement on site and for the even more complex evaluation of the data. For ultrasound with AI, a specific training is not necessary. The score is 3:0
(4) Awareness of the underlying standards
In this dimension the vibration measurement can score a goal. It is 3:1. The standards have been established for many years (e.g. DIN 10816) while the standards for ultrasonic data (e.g. VDI 3832) are not so well known. However, this will change rapidly with the spread of the technology.
(5) Degree of digitalization/degree of automation of the evaluation process
Since ultrasound data is evaluated by means of artificial intelligence, the degree of automation is significantly higher, so the solution has a higher degree of digitization overall. We are free to say, that the scoreboard now shows 4:1.
(6) Insight of the analysis
If sufficient information is available (e.g. shaft speed, bearing data, gearbox data), vibration measurement can provide an impressive depth of analysis and, given the right conditions, identify the damaged component in a complex system. However, data, data and even more data will lead to the fact that the depth of knowledge of ultrasound with AI is increasing every day. Currently still a goal for vibration measurement, but the gatekeeper already touched the ball. It’s 4:2
(7) Number of different Use Cases
Vibrations can only be detected when something is rotating at a certain speed. The focus is therefore on systems that rotate quickly or are otherwise strongly excited. The applications of ultrasound with AI go far beyond this. For example, slow rotating systems can be analyzed, or systems where parts move up and down, or left to right(e.g. hydraulic cylinders). Let’s take a look at the scoreboard: 5:2
(8) Systematische Verbesserung des Systems durch jede Messung (“Re-enforced learning”)
When analyzing vibration data again and again, the operator may gain experience, but the overall system is not improved automatically. The evaluation system itself does not automatically learn more with every piece of analysis. This is different with ultrasound and artificial intelligence. With every measurement the artificial intelligence learns, with every case of damage, but also with every asset that has been running for another month without problems and this way, the prediction of the remaining useful lifetime becomes more and more accurate. A simple goal, it is 6:2
(9) Cost / effort per measurement
With ultrasound and artificial intelligence you don't need fixed sensors, the approach is a "zero investment" approach. Furthermore, the analysis of the data is fully automated, the data transfer from customer to AI is completely digital. The AI provides you with results directly on the tablet. The artificial intelligence likes to work 7 days a week and 24 hours a day. The cost advantages are enormous, which leads to the final score of 7:2 for ultrasound with artificial intelligence.
In conclusion, it remains to be said that ultrasound data in particular is predestined for evaluation by artificial intelligence in the sense of predictive maintenance.
