Search

The conditions of various nuclear power plant facilities are regularly examined through manual inspections. Remote visual inspection is commonly applied and requires engineers to watch lengthy inspection footage and seek anomaly features therein. This is a labor-intensive process as anomaly features of interest usually only appear in very short segments of the original whole video. Therefore, an automated anomaly detection system is preferred to lessen the intensive labor cost in the inspection process. The detection process could also benefit from useful information that could potentially contribute to addressing reasoning traceability. With a well-prepared training data set of the anomaly feature, a convolutional neural network (CNN) can be developed to automatically detect anomaly indications in the inspection video. However, false-positive detections may occur and can be difficult to remove without seeking manual verification. To overcome this problem, we present a new automated video-level anomaly detection framework that utilizes the latency mechanism to effectively lessen false-positive occurrences, and therefore, increase detection accuracy. In this framework, a CNN-based anomaly classifier first performs initial scanning of the anomaly type of interest in every region of the sampled frames. Then our latency mechanism is applied to refine the initial scanning results by flagging up a region as an "anomaly" indication only when "anomaly" is detected by CNN in the current frame and also in a sequence of previous consecutive frames of the same region. We present a case study of crack feature detection in superheater inspection videos to illustrate the performance of the proposed framework. The results show that the latency mechanism can effectively remove the original false-positive detections seen in the initial scanning. In order to provide a primary exploration of suggesting possible formats for addressing reasoning traceability, knowledge graphs of the reasoning process in the video-level detection framework are built to provide a better understanding of why a specific section of the video is flagged as anomaly contents by the video-level detection framework. [ABSTRACT FROM AUTHOR]

Gas circulator (GC) units are an important rotating asset used in the advanced gas-cooled reactor design, facilitating the flow of CO$_2$ gas through the reactor core. The ongoing maintenance and examination of these machines are important for operators in order to maintain safe and economic generation. GCs experience a dynamic duty cycle with periods of nonsteady state behavior at regular refueling intervals, posing a unique analysis problem for reliability engineers. In line with the increased data volumes and sophistication of available technologies, the investigation of predictive and prognostic measurements has become a central interest in rotating asset condition monitoring. However, many of the state-of-the-art approaches finding success deal with the extrapolation of stationary time series feeds, with little to no consideration of more complex but expected events in the data. In this paper, we demonstrate a novel modeling approach for examining refueling behaviors in GCs, with a focus on estimating their health state from vibration data. A machine learning model was constructed using the operational history of a unit experiencing an eventual inspection-based failure. This new approach to examining GC condition is shown to correspond well with explicit remaining useful life measurements of the case study, improving on the existing rudimentary extrapolation methods often employed in rotating machinery health monitoring. [ABSTRACT FROM PUBLISHER]

This paper presents a self-tuning framework for the diagnosis of routine alarms in steam turbine generators utilizing a combination of inductive machine learning and knowledge-based heuristics. The techniques provide a novel basis for initializing and updating time series feature extraction parameters used in the automated decision support of vibration events due to operational transients. The data-driven nature of the algorithms allows for machine-specific characteristics of individual turbines to be learned and reasoned about. The paper provides a case study illustrating the routine alarm paradigm, and the applicability of systems using self-tuning techniques. The approaches discussed throughout are presented to provide useful diagnosis tools for the reliability and maintenance analysis of steam turbine generators. [ABSTRACT FROM PUBLISHER]

Unplanned outages can be especially costly for generation companies operating nuclear facilities. Early detection of deviations from expected performance through condition monitoring can allow a more proactive and managed approach to dealing with ageing plant. This paper proposes an anomaly detection framework incorporating the use of the Hidden Markov Model (HMM) to support the analysis of nuclear reactor core condition monitoring data. Fuel Grab Load Trace (FGLT) data gathered within the UK during routine refueling operations has been seen to provide information relating to the condition of the graphite bricks that comprise the core. Although manual analysis of this data is time consuming and requires considerable expertise, this paper demonstrates how techniques such as the HMM can provide analysis support by providing a benchmark model of expected behavior against which future refueling events may be compared. The presence of anomalous behavior in candidate traces is inferred through the underlying statistical foundation of the HMM which gives an observation likelihood averaged along the length of the input sequence. Using this likelihood measure, the engineer can be alerted to anomalous behaviour, indicating data which might require further detailed examination. It is proposed that this data analysis technique is used in conjunction with other intelligent analysis techniques currently employed to analyse FGLT to provide a greater confidence measure in detecting anomalous behaviour from FGLT data. [ABSTRACT FROM AUTHOR]

A critical component of an advanced-gas cooled reactor (AGR) station is the graphite core. As a station ages, the graphite bricks that comprise the core can distort and may eventually crack. As the core cannot be replaced the core integrity ultimately determines the station life. Monitoring these distortions is usually restricted to the routine outages, which occur every few years, as this is the only time that the reactor core can be accessed by external sensing equipment. However, during weekly refueling activities measurements are taken from the core for protection and control purposes. It is shown in this paper that these measurements may be interpreted for condition monitoring purposes, thus potentially providing information relating to core condition on a more frequent basis. This paper describes the data-mining approach adopted to analyze this data and also describes a software system designed and implemented to support this process. The use of this software to develop a model of expected behavior based on historical data, which may highlight events containing unusual features possibly indicative of brick cracking, is also described. Finally, the implementation of this newly acquired understanding in an automated analysis system is described. [ABSTRACT FROM AUTHOR]

AbstractIn high-dose radiological environments, where precision and safety are of utmost importance, the ability to acquire accurate and clear visual information is of paramount importance for ensuring safety and reliability in critical industrial processes. However, these environments inherently introduce significant challenges due to the adverse effects of radiation on imaging equipment. As a consequence, inspection videos captured within such high-radiation environments often contain a significant amount of noise. This noise substantially complicates the task of identifying and assessing potential defects in vital components. It also diverts attention and resources toward investigating false positives created by noise, leading to inefficiencies, and for industrial processes on the critical path, this can further prolong the outage. Addressing this noise is essential not only for precision, but also for ensuring safety, reliability, and efficiency in critical industrial processes.In this paper, we present a custom-designed filter utilizing a priori information about camera position and trajectory to remove the noise from the inspection videos, making the defects easier to manually identify. As the camera movement is in one direction at a constant speed, the proposed approach uses this temporal and spatial information to accurately remove the noise. This approach applies to a subset of visual inspection problems throughout the nuclear industry, as well as many other industries where there is knowledge available about the camera speed and direction of travel. The proposed approach is compared with three accepted/well-known approaches, median filtering, bilateral filtering, and fast nonlocal means denoising, and an additional state-of-the-art deep learning model is also used for comparison. It was found that the proposed approach produces the most accurate video denoising in terms of visual quality and the retainment of the defect features throughout the videos tested. [ABSTRACT FROM AUTHOR]