Auto Sizing of CANDU Nuclear Reactor Fuel Channel Flaws from UT Scans.

The inspection of nuclear power plants is an essential process that occurs during plant outages. During this process, various systems are inspected, including the reactor's fuel channels to ensure that they are safe and reliable for the plant's operation. The inspection of Canada Deuterium Uranium (CANDU®) reactor pressure tubes, which are the core component of the fuel channels and house the reactor fuel bundles, is performed using Ultrasonic Testing (UT). Based on the current process that is followed by Canadian nuclear operators, the UT scans are manually examined by analysts to locate, measure, and characterize pressure tube flaws. This paper proposes solutions for the auto-detection and sizing of pressure tube flaws using two deterministic algorithms, the first uses segmented linear regression, while the second uses the average time of flight (ToF) within ±σ of µ. When compared against a manual analysis stream, the linear regression algorithm and the average ToF achieved an average depth difference of 0.0180 mm and 0.0206 mm, respectively. These results are very close to the depth difference of 0.0156 mm when comparing two manual streams. Therefore, the proposed algorithms can be adopted in production, which can lead to significant cost savings in terms of time and labor.

Using Deep Learning to Automate the Detection of Flaws in Nuclear Fuel Channel UT Scans.

Nuclear reactor inspections are critical to ensure the safety and reliability of a nuclear facility's operation. In Canada, ultrasonic testing (UT) is used to inspect the health of pressure tubes that are part of Canada's Deuterium Uranium (CANDU) reactor's fuel channels. Currently, analysis of UT scans is performed by manual visualization and measurement to locate, characterize, and disposition flaws. Therefore, there is motivation to develop an automated method that is fast and accurate. In this article, a proof of concept (PoC) that automates the detection of flaws in nuclear fuel channel UT scans using a convolutional neural network (CNN) is presented. The CNN model was trained after constructing a dataset using historical UT scans and the corresponding inspection results. The requirement for this prototype was to identify the location of at least a portion of each flaw in UT scans while minimizing false positives (FPs). The proposed CNN model achieves this target by automatically identifying at least a portion of each flaw where further manual analysis is performed to identify the width, the length, and the type of the flaw.

Practical Considerations for Accuracy Evaluation in Sensor-Based Machine Learning and Deep Learning.

Accuracy evaluation in machine learning is based on the split of data into a training set and a test set. This critical step is applied to develop machine learning models including models based on sensor data. For sensor-based problems, comparing the accuracy of machine learning models using the train/test split provides only a baseline comparison in ideal situations. Such comparisons won't consider practical production problems that can impact the inference accuracy such as the sensors' thermal noise, performance with lower inference quantization, and tolerance to sensor failure. Therefore, this paper proposes a set of practical tests that can be applied when comparing the accuracy of machine learning models for sensor-based problems. First, the impact of the sensors' thermal noise on the models' inference accuracy was simulated. Machine learning algorithms have different levels of error resilience to thermal noise, as will be presented. Second, the models' accuracy using lower inference quantization was compared. Lowering inference quantization leads to lowering the analog-to-digital converter (ADC) resolution which is cost-effective in embedded designs. Moreover, in custom designs, analog-to-digital converters' (ADCs) effective number of bits (ENOB) is usually lower than the ideal number of bits due to various design factors. Therefore, it is practical to compare models' accuracy using lower inference quantization. Third, the models' accuracy tolerance to sensor failure was evaluated and compared. For this study, University of California Irvine (UCI) 'Daily and Sports Activities' dataset was used to present these practical tests and their impact on model selection.