Artificial Intelligence (AI) has made major breakthroughs in the past decade, and multiple industries are now relying on these new technologies to revolutionize various complex tasks and help workers improve their efficiency. Among the various technologies considered as “artificial intelligence”, a subset is referred to as “machine learning”. Machine learning techniques can be applied to data generated from tubing inspections.

By combining the expertise of AI specialists and eddy current experts, Eddyfi Technologies created a machine learning model capable of analyzing the data generated from tubing inspections. In practical terms, this model can reproduce a complex processing chain including a combination of scaling, filters, and channel mixing. It can analyze all frequencies of both absolute and differential channels at the same time and combine them to amplify the signal of defects while minimizing noise and other unwanted signals enabling inspectors to increase efficiency and improve data reliability.

Magnifi 5 + Ectane

Figure 1: Ectane® instrument with Magnifi® software

Conventional detection tools based on user-defined rules have been available for a while and are even part of standardized procedures in the nuclear market. These rule-based systems have limitations that are overcome by AI.

 

HEAT EXCHANGER INSPECTION WITH EDDYFI TECHNOLOGIES ECTANE

Figure 2: Heat exchanger inspection with Eddyfi Technologies Ectane

Good tube regions are essential for the detection of indications and depend on correct detection of landmarks. Conventional landmark detection tools are limited and cannot accurately determine the width of a support plate or tube sheet in all contexts. If the length of a support plate is not well defined, an analysis system will not apply the proper algorithm, possibly based on a mixed channel, to analyze the data near the plate. An AI-based landmark detection solves this problem because it can accurately mark the length regardless of whether the pulling speed is constant or not.

Rule-based detection can be done by placing boxes and thresholds on the Lissajous, to circumscribe the area where a defect can be found. However, this way of doing things is also limited. Detection boxes can be very difficult to adjust, and readjustments may be required to maintain acceptable detection over the entire heat exchanger. Although this tool is useful at times, it is difficult to use effectively and takes a long time to set up. AI detection does not have these limitations and can detect defects even if there is a drift of the signal or if the amplitude of the defects does not quite reach the limits of a threshold.

Interested in seeing how Eddyfi Technologies AI-based eddy current tube inspection technology can play a role in your heat exchanger tube inspections? Learn more in this article, and contact us to be connected with our tubing experts ready to help your operations.