Artificial Intelligence Aproaches as Tools for Auditing and Improving Data Analysis of Advanced Ultrasound Techniques in Non-Destructive Testing

Abstract

Advanced Non Destructive Testing (NDT) techniques rely on consistent acquisition of data and its reliable analysis. As technology advances, new challenges emerge. For instance, the amount of data produced is exceeding the qualified personnel available for data analysis; moreover, the files produced can be easily handle unethically. Our project proposes the implementation of Artificial Intelligence (AI) techniques to develop algorithms to better harness the data available, in order to enhance the quality of data analysis and better monitor ethical practices. Method: For any NDT technique, the consistency of the results and the reliability of the findings depends heavily on the personnel performing the inspections and evaluating the results. As more advanced NDT techniques are available in the areas of ultrasound, digital data becomes available opening the door for intelligent tools for analysing and handling ultrasound data. This project proposes an innovative analysis software based on Computational Intelligence (CI) techniques. The implementation of this software reduces the time needed for a single analysis and the consistency of results, more importantly; it can be use as an audit tool to ensure data files have not been mishandle. For the first stage of the proposed CI cloud-based software, the data collected by the inspector is send via internet connection to web-based servers. During the second stage, the data will be verify for quality; if any data is missing or poor-quality is detected, the software will inform the inspector that data must be resubmitted. Once the verification of data is completed, the data analysis is then perform by CI-based algorithms, which can recognise and classify features within the data such as geometry indications, anomalies and defects. Following the CI-based analysis, the software generates a complete report and the results are be available to the user via a 3D visualization interface. This report includes details of sizing, characterisation, location of defects, geometry indications and other relevant information related to the analysed data. The information contained in the interface and generated by the software will be available to the final user for validation and approval. The CI-based analysis tools have already been use to monitor several data files from NDT inspections using Phased Array Ultrasound (PAUT) and the results have successfully verify the authenticity of the ultrasound raw data, which is a critical verification step and evidence of any mal practices or mistakes of data handling. There is a lack of reliable tools available to NDT experts that can facilitate the task of analysing data; this innovative software takes advantage of the CI techniques and expert knowledge to address this issue. Furthermore, our approach offers a high degree of consistency for identifying defects; improves the quality of the analysis and reduces the time invested in such critical NDT duty. Auditing raw data from ultrasound inspection is a challenging task; by taking advantage of CI-based analysis tools this task can be easily implement it as reliable auditing practice.