Advancement of sensor array processing and MIMO signal processing for non-destructive testing

作为需求不断增长的主要电力生产国，加拿大电力工业努力提高其发电站的安全性、生产力和可靠性至关重要。特别是，核电站的安全和持续运行关键取决于对某些表面的适当维护和检查。无损检测（NDT）是一种分析，用于检测成品的疲劳效应和结构缺陷，同时确保在测试过程中不损坏或破坏。本课题研究新型传感器阵列处理和多输入多输出成像技术在核电站超声无损检测中的应用。传感器阵列处理是信号处理的一个分支，对一组传感器的输出进行采集和处理，以获得周围环境的图像。MIMO信号处理作为一种提高无线连接领域的可靠性和数据吞吐量的手段已经得到了广泛的研究和理解。在这个项目中，我们开发了新的统计和阵列信号处理技术，能够检测核电站锅炉管道内的各种异常。为此，我们提出了能够捕捉折射率、校准误差、信号特性、噪声特性和传感器间耦合变化影响的数据模型。我们对无损检测程序的研究是为了满足两个行业目标：1)通过开发更快的检测方法和处理算法来减少检测过程中产生的经济损失；2)减少培训专业安全检查人员所产生的成本，这些人员由于规定了人类可以安全承受的可接受辐射剂量而有很高的流失率。主要的经济效益将是同时检测多种缺陷类型，如裂纹、变薄和孔隙。这将减少未来在停机期间的检查活动时间，从而带来净经济优势。

As a major electricity producer with an ever-increasing demand, it is vital that the Canadian power industry strive to improve the safety, productivity and reliability of its generating stations. In particular, nuclear power plants' safe and continuous operation critically depends on the proper maintenance and inspection of certain surfaces. Non-destructive testing (NDT) is a type of analysis performed to detect fatigue effects and structural imperfections on a finished product while ensuring that it is not damaged or destroyed during testing. This project is concerned with the application of novel sensor array processing and multi-input multi-output (MIMO) imaging techniques to ultrasonic NDT in nuclear power plants. Sensor array processing is a branch of signal processing where the outputs of an array of sensors are collected and processed in order to obtain the image of the surrounding environment. MIMO signal processing has been widely investigated and is well understood as a means to increase reliability and data throughput in the area of wireless connectivity. In this project, we develop novel statistical and array signal processing techniques that enable the detection of various types of abnormalities within the pipes that feed the boiler of a nuclear power plant. To do so, we propose data models which capture the effects of variations in the refraction indices, the calibration errors, signal characteristics, noise properties, and inter-sensor coupling. Our research in non-destructive testing procedures is directed at meeting two industry targets: i) decreasing financial losses incurred during testing procedures by developing faster testing methods and processing algorithms, and ii) reducing costs incurred during the training of specialized safety inspection personnel who have a high-turnover rate due to regulations stipulating the acceptable radiation doses humans can safely sustain. The major economic benefit will be the simultaneous detection of multiple flaw types such as cracks, thinning, and porosity. This will reduce future inspection campaign times during outages, thereby leading to a net economical advantage.