Artificial Intelligence in NDT: Automatic Error Detection and Higher Data Integrity

Introduction

The field of non-destructive testing (NDT) is undergoing a transformation with the emergence of artificial intelligence (AI). The ability of AI to automate tasks and identify patterns is being increasingly harnessed to improve the efficiency and accuracy of the inspection process. This article explores the applications and limitations of AI in NDT, as well as the role of Picture Archiving and Communication System (PACS) in supporting AI-based inspection.

Applications of AI in NDT

One of the main applications of AI in NDT is the automatic recognition of components and the assignment to an inspection instruction. The software can recognize the inspection part based on the images and automatically place a matching template over the image, which shows exactly where measurements must be taken. The generated inspection data is then fed back into the Inspection Data Management System (IDMS). This automated inspection can also be carried out retroactively for inspections that have already been performed, for example, for quality assurance purposes. AI can analyze past inspection data and identify patterns and trends that can help improve future inspections. By leveraging historical data, AI can help identify areas that require more attention and detect potential issues before they become major problems.

Another application of AI in inspection is the automated detection of erosion, corrosion, and deposits on the test images. This means that the AI system can analyze the images captured during the inspection process and identify corresponding signs without requiring the inspector to manually review each image. This not only improves the efficiency of the inspection process but also enhances the accuracy of the detection, as AI systems can often identify subtle defects that may be missed by human inspectors. Additionally, AI can be used to automate the measurement process. AI algorithms can detect the point on a pipeline where the wall thickness is the thinnest and measure it automatically.

Limitations of AI

While AI has the potential to significantly improve NDT, there are still several limitations to its implementation. One of the main challenges is the need for appropriate data to develop accurate AI models. Obtaining such data can be difficult, as it requires a structured digital dataset with a wide range of defect types, sizes, and orientations. Additionally, human expertise is still required to interpret and make decisions based on the AI results, as AI is rather aimed at supporting inspectors in conducting more accurate and efficient inspections, not their replacement.

PACS as a Prerequisite for AI

The prerequisite for the use of AI solutions is a high-quality database. To build one, the inspection workflow should be digitized in such a way that inspection data and reports are available digitally, and there are no media breaks in the process. PACS fulfills these criteria as it connects leading inspection systems, such as enterprise resource planning (ERP) and risk-based inspection (RBI), and streamlines the entire process - from NDT data acquisition to evaluation, management, and archiving. Therefore, the software provides an excellent database for organizations to train AI models and profit from using them in NDT and inspection.

Conclusion

AI has the potential to significantly improve and optimize NDT. The benefits include increased productivity, better accuracy, and faster inspection times. The key to unlocking the potential of AI in NDT is to implement the necessary technologies to digitize inspection workflows and create a structured digital database. Therefore, it is time for organizations to take further steps toward a more advanced and optimized NDT and utilize the possibilities that AI offers. PACS is a prerequisite for AI-based inspection as it provides a high-quality database to train AI models, which can then be used to improve inspection processes.