Symptom based diagnostic system for nuclear power plant operations using artificial neural networks

Nuclear power plant experiences a number of transients during its operations. These transients may be due to equipment failure, malfunctioning of process systems and unavailability of safety systems. In such a situation, the plant may result into an abnormal state which is undesired. In case of an undesired plant condition generally known as an initiating event (IE), the operator has to carry out diagnostic and corrective actions. The operator's response may be too late to mitigate or minimize the negative consequences in such scenarios. The objective of this work is to develop an operator support system based on artificial neural networks that will assist the operator to identify the IEs at the earliest stages of their developments. These abnormal plant conditions must be diagnosed and identified through the process instrument readings. A symptom based diagnostic system has been developed to investigate the IEs. The event identification is carried out by using resilient back propagation neural network algorithm. Whenever an event is detected, the system will display the necessary operator actions in addition to the type of IE. The system will also show the graphical trend of relevant parameters. The developed system is able to identify the eight IEs of Narora Atomic Power Station. This paper describes the features of the diagnostic system taking one of the IEs as a case study.     

Diagnostic system for identification of accident scenarios in nuclear power plants using artificial neural networks

This paper presents the work carried out towards developing a diagnostic system for the identification of accident scenarios in 220 MWe Indian PHWRs. The objective of this study is to develop a methodology based on artificial neural networks (ANNs), which assists in identifying a transient quickly and suggests the operator to initiate the corrective actions during abnormal operations of the reactor. An operator support system, known as symptom-based diagnostic system (SBDS), has been developed using ANN that diagnoses the transients based on reactor process parameters, and continuously displays the status of the reactor. As a pilot study, the large break loss of coolant accident (LOCA) with and without the emergency core cooling system (ECCS) in reactor headers has been considered. Several break scenarios of large break LOCA have been analyzed. The time-dependent transient data have been generated using the RELAP5 thermal hydraulic code assuming an equilibrium core, which conforms to a realistic estimation. The diagnostic results obtained from the ANN study are satisfactory. These results have been incorporated in the SBDS software for operator assistance. A few important outputs of the SBDS have been discussed in this paper.     

感性工学在产品配色设计中的应用研究

在感性工学的研究领域中,类神经网络被认为是颇具潜力的分析工具.为了降低产品配色的意象偏差,基于配色理论设计问卷,配合类神经遗传模拟消费者的意象评价,提供符合产品形态的配色建议,并在此基础上提出了计算机辅助配色设计系统的框架和实现方法.     

Artificial Neural Network for Creep Behaviour Predictions of a Parallel‐lay Aramid Rope Under Varying Stresses

Abstract: An improvement of the creep behaviour prediction of parallel‐lay aramid ropes under varying stresses is the scope of the following study in which application of artificial neural networks (ANNs) for the prediction of creep under varying stresses is presented. This qualitatively different approach assumes that the ANN can be trained to simulate time‐dependent response of the rope in the given load (stress) programme and time interval. The classic rheological constitutive equations are not needed in this case, because ANN acts as a constitutive operator trained by stresses and the corresponding creep strains from experimental data. Carried numerical experiments were divided into the following three parts: (i) searching the best ANN for a creep behaviour approximation under varying stresses, (ii) investigating the best topology of the selected neural network and (iii) investigating the best results for the creep function identification. Comparison between the experimentally observed creep strains of the parallel‐lay aramid rope under varying stresses, predicted creep strains when the linear creep constitutive equation is applied and predicted creep strains when the obtained Jordan neural network with the 3‐10‐1 topology is used confirmed that the Jordan neural network developed achieved less than half mean square error beside the existing creep constitutive analytical approach.     

Aging process optimization for a copper alloy considering hardness and electrical conductivity

A multi-objective optimization methodology for the aging process parameters is proposed which simultaneously considers the mechanical performance and the electrical conductivity. An optimal model of the aging processes for Cu–Cr–Zr–Mg is constructed using artificial neural networks and genetic algorithms. A supervised artificial neural network (ANN) to model the non-linear relationship between parameters of aging treatment and hardness and conductivity properties is considered for a Cu–Cr–Zr–Mg lead frame alloy. Based on the successfully trained ANN model, a genetic algorithm is adopted as the optimization scheme to optimize the input parameters. The result indicates that an artificial neural network combined with a genetic algorithm is effective for the multi-objective optimization of the aging process parameters.     

Methods and practices used in incident analysis in the Finnish nuclear power industry

According to the Finnish Nuclear Energy Act it is licensee's responsibility to ensure safe use of nuclear energy. Radiation and Nuclear Safety Authority (STUK) is the regulatory body responsible for the state supervision of the safe use of nuclear power in Finland. One essential prerequisite for the safe and reliable operation of nuclear power plants is that lessons are learned from the operational experience. It is utility's prime responsibility to assess the operational events and implement appropriate corrective actions. STUK controls licensees' operational experience feedback arrangements and implementation as part of its inspection activities. In addition to this in Finland, the regulatory body performs its own assessment of the operational experience. Review and investigation of operational events is a part of the regulatory oversight of operational safety. Review of operational events is done by STUK basically at three different levels. First step is to perform a general review of all operational events, transients and reactor scram reports, which the licensees submit for information to STUK. The second level activities are related to the clarification of events at site and entering of events' specific data into the event register database of STUK. This is done for events which meet the set criteria for the operator to submit a special report to STUK for approval. Safety significance of operational events is determined using probabilistic safety assessment (PSA) techniques. Risk significance of events and the number of safety significant events are followed by STUK indicators. The final step in operational event assessment performed by STUK is to assign STUK's own investigation team for events deemed to have special importance, especially when the licensee's organisation has not operated as planned. STUK launches its own detail investigation once a year on average. An analysis and evaluation of event investigation methods applied at STUK, and at the two Finnish nuclear power plant operators Teollisuuden Voima Oy (TVO) and Fortum Power and Heat Oy (Fortum) was carried out by the Technical Research Centre (VTT) on request of STUK at the end of 1990s. The study aimed at providing a broad overview and suggestions for improvement of the whole organisational framework to support event investigation practices at the regulatory body and at the utilities. The main objective of the research was to evaluate the adequacy and reliability of event investigation analysis methods and practices in the Finnish nuclear power industry and based on the results to further develop them. The results and suggestions of the research are reviewed in the paper and the corrective actions implemented in event investigation and operating experience procedures both at STUK and at utilities are discussed as well. STUK has developed its own procedure for the risk-informed analysis of nuclear power plant events. The PSA based event analysis method is used to assess the safety significance and importance measures associated with the unavailability of components and systems subject to Technical Specifications. The insights from recently performed PSA based analyses are also briefly discussed in the paper.     

NAROAS: a neural network-based advanced operator support system for the assessment of systems reliability

We have developed and implemented a computerized reliability monitoring system for nuclear power plant applications, based on a neural network. The developed computer program is a new tool related to operator decision support systems, in case of component failures, for the determination of test and maintenance policies during normal operation or to follow an incident sequence in a nuclear power plant. The NAROAS (Neural Network Advanced Reliability Advisory System) computer system has been developed as a modularized integrated system in a C++ Builder environment, using a Hopfield neural network instead of fault trees, to follow and control the different system configurations, for interventions as quickly as possible at the plant. The observed results are comparable and similar to those of other computer system results. As shown, the application of this neural network contributes to the state of the art of risk monitoring systems by turning it easier to perform online reliability calculations in the context of probabilistic safety assessments of nuclear power plants.     

神经网络的进化学习方法研究

提出了两种用于前向神经网络的进化学习算法,实验证明它们能有效地网络权值空间中寻找全局最优解。在比较实验的基础上,得出了在神经网络的进化学习过程中变异是起主导作用的遗传算子的结论,并以此为指导配置算法的各个关键参数。通过对ＸＯＲ问题和ＩＲＩＳ模式分类问题的学习证明,这两种算法均能获得远高于传统的ＢＰ算法的性能。     

PET image reconstruction using ANN

The aim of this study is to improve the positron emission tomography (PET) image quality for medical diagnosis. The statistical reconstructions on the maximum a posteriori (MAP) algorithm often results in a blurring effect, which fails to determine the toughness class in the reconstructed image. The development of new reconstruction algorithms for PET is an active field of research. In this article, artificial neural network (ANN) is proposed for replicating the output image, which is generated from the acquired projection data with the corresponding angles using the PET images. This article proposes the advantage of arranging the neural network to stock up the information of the continuous capacity. This reduces the storage space and recuperates as much sequence of the continuous quantity as possible. The performance of image quality parameters using ANN is better when compared with MAP, FBP‐NN (filtered back projection with nearest neighbor interpolation). Thus ANN provides 63% better peak signal to noise ratio (PSNR) when compared with FBP‐NN and 47% better when compared to MAP. Thus, ANN is better than FBP and MAP algorithm, by providing better PSNR. © 2014 Wiley Periodicals, Inc. Int J Imaging Syst Technol, 24, 249–255, 2014     

目标地震动信号的特征提取及识别研究

地面目标的地震动信号是目标识别的关键,本文研究了机动目标地震动特性,总结出目标的地震动信号的特征提取规律,并将神经网络方法用于目标的地震动信号的分类识别中。给出了识别结构及改进的BP算法,并将改进的BP算法用于实际目标的地震动信号的分类识别,得到令人满意的结果。     

A systematic approach to analysing errors of commission from diagnosis failure in accident progression

The accident scenarios of a nuclear power plant are composed of an initiating event (IE), additional events/failures and human inappropriate actions, the combinations of which lead to irreversible consequences. In such a dynamic situation, operators should diagnose the occurring events/failures (including an initiating event and additional events) and assess the related situations utilising the available resources such as operating procedures or human–machine systems to control and maintain the plant in a stable condition. The misdiagnosis or diagnosis failure of the occurring events could cause critical human inappropriate actions that aggravate the plant condition, which is termed as errors of commission (EOCs). This paper presents a methodology for analysing the potential for diagnosis failure of the initiating and additional events and the consequent EOC events, based on the operating procedures, in the accident scenarios of nuclear power plants. The method to be presented categorizes the diagnostic situations in the accident scenarios into three cases according to the structure of the emergency operating procedures (EOPs) and the time of the occurring events: (1) the diagnosis of an initiating event, (2) the diagnosis of both an initiating event and an additional event when an additional event occurs prior to the performance of the diagnosis procedure, and (3) the diagnosis of an additional event when an additional events occurs after the performance of the diagnosis procedure. The application of the method is illustrated through three case example scenarios: (1) the power-operated relief valve (PORV) or the pressurizer safety valve (PSV) LOCA, (2) the loss of all feedwater (LOAF) event (loss of main feedwater*loss of auxiliary feedwater), (3) the sequence of<the station blackout (SBO)*loss of turbine-driven (or diesel-driven) auxiliary feedwater*PSV stuck-open*recovery of AC power>.     

The MDTA-based method for assessing diagnosis failures and their risk impacts in nuclear power plants

In the emergency situations of nuclear power plants (NPPs), a diagnosis of the occurring events along an accident progression or as initiating events is crucial for managing or controlling a plant to a safe and stable condition. If the operators fail to diagnose the occurring event(s), their responses to a given event can eventually become inappropriate or inadequate. This paper presents an analytical method for assessing the potential for a diagnosis failure (or misdiagnosis) and its consequences for human behaviour and plant safety. The method largely comprises of three steps as follows: (1) Analysis of the potential for a diagnosis failure, (2) Identification of the human failure events (HFEs) that might be induced due to a diagnosis failure, and (3) Quantification of the HFEs and their modeling into a PSA model. The paper also presents a pilot application of the proposed method to the small loss of coolant accident of a Korean NPP.     

A computational model for knowledge-driven monitoring of nuclear power plant operators based on information theory

To develop operator behavior models such as IDAC, quantitative models for the cognitive activities of nuclear power plant (NPP) operators in abnormal situations are essential. Among them, only few quantitative models for the monitoring and detection have been developed. In this paper, we propose a computational model for the knowledge-driven monitoring, which is also known as model-driven monitoring, of NPP operators in abnormal situations, based on the information theory. The basic assumption of the proposed model is that the probability that an operator shifts his or her attention to an information source is proportional to the expected information from the information source. A small experiment performed to evaluate the feasibility of the proposed model shows that the predictions made by the proposed model have high correlations with the experimental results. Even though it has been argued that heuristics might play an important role on human reasoning, we believe that the proposed model can provide part of the mathematical basis for developing quantitative models for knowledge-driven monitoring of NPP operators when NPP operators are assumed to behave very logically.     

Hybrid neural network model for the design of beam subjected to bending and shear

There is no direct method for design of beams. In general the dimensions of the beam and reinforcement are initially assumed and then the interaction formula is used to verify the suitability of chosen dimensions. This approach necessitates few trials for coming up with an economical and safe design. This paper demonstrates the applicability of Artificial Neural Networks (ANN) and Genetic Algorithms (GA) for the design of beams subjected to moment and shear. A hybrid neural network model which combines the features of feed forward neural networks and genetic algorithms has been developed for the design of beam subjected to moment and shear. The network has been trained with design data obtained from design experts in the field. The hybrid neural network model learned the design of beam in just 1000 training cycles. After successful learning, the model predicted the depth of the beam, area of steel, spacing of stirrups required for new problems with accuracy satisfying all design constraints. The various stages involved in the development of a genetic algorithm based neural network model are addressed at length in this paper.     

Assessment of fire vulnerability for nuclear power plant safety systems by combining fault-and success-oriented logic with physical models

The time behaviour of potential accident sequences may carry important information regarding nuclear power plant (NPP) safety operation and shutdown. In the case of external and environmental events, the ability of NPP components to operate correctly can be changed dramatically in a short time. In contrast to the failures caused by internal events, these two groups of undesirable events may lead to dynamic dependent failures among components of one or several systems. Such kinds of failure should be taken into account in the models of NPP behaviour. To evaluate how successfully the tasks of the safety systems will be carded out, logical models such as fault trees are usually used. The fault trees are not efficient at describing the short-term changes of the failure probabilities for system components. A method that has some advantages over the pure fault tree logic is proposed. The main features of the method are demonstrated by using examples.     

Assessment of human–machine interface design for a Chinese nuclear power plant

A nuclear power plant (NPP) is a complex system but requires high reliability. The human–machine interface (HMI) design plays very important role in reactor safety. This paper describes an assessment on HMI design of a Chinese NPP, using a software system named Dynamic Interaction Analysis Support (DIAS). DIAS can give not only quantitative indices for dynamically assessing the HMI design, but also allow modify the values of these indices by taking into account human error probability during specified emergent operation procedures. The operation procedures dealing with postulated accidents and transients recorded from a full-scale plant simulator in the training center of a Chinese NPP were selected as references. According to the results of simulation and analysis, the potential problems in the HMI design and the operation procedures were detected. Suggestions to improve the HMI design and the operation procedures were addressed.     

A methodology for formulating, formalizing, validating, and evaluating a real-time process control advisor

This research presents guidelines to design control processes where improving quality is achieved by improving the manufacturing consistency through the use of intelligent process control. Conventional control processes cannot include the theoretical knowledge, experimental knowledge, and expert knowledge available concerning the product. A hybrid intelligent process control (IPC) combining a continuous simulation (CS) and an artificial neural network (ANN) can make this knowledge available to the operator for process control. This paper presents a methodology for combining the CS and ANN to achieve real-time process control. A human-machine interface (HMI) is included in the process to aid operators in communication with the CS/ANN hybrid IPC. The result of the new process is a real-time process control advisor (RTPCa). A case example for the methodology of formulating, formalizing, validating, and evaluating the RTPCa is given. The case studied concerns galvanizing continuous sheet steel at a steel plant. The CS is written in SIM AN, and the ANN in C. The research validates and evaluates the RTPCa using plant data, simulation output, and face validation by plant personnel. The authors conclude that the benefits of the RTPCa over other forms of IPC include better process communication to the operator, robustness to moderate changes in system parameters, the flexibility to retrain the ANN if conditions change dramatically, and the computation speed necessary for real-time process control. This methodology has further applications to other continuous processes where quality is determined by manufacturing consistency of the product, such as in the pulp paper and film processing industries.     

模糊聚类和混沌自适应粒子群的神经网络色彩匹配

目的研究基于混沌理论、粒子群算法、模糊聚类和人工神经网络的色彩匹配模型。方法结合混沌理论和动态自适应策略,对粒子群算法进行改进,得到混沌自适应粒子群算法,并应用于径向基人工神经网络的基函数中心,以及扩展常数和网络权值的优化中;通过模糊聚类分类样本数据,得到混沌自适应粒子群径向基人工神经网络色彩匹配模型,并将模型与其他色彩匹配方法进行比较。结果CSAPSO RBF ANN色彩匹配模型的平均绝对误差、均方根误差和色差平均值分别为0.0106,0.000 96和1.9122。结论 CSAPSO RBF ANN色彩匹配模型具有良好的普遍性、通用性和泛化能力。     

COPILOT—A PC-based expert system for reactor operational assistance using a Bayesian diagnostic module

This paper describes an expert system for nuclear plant operational assistance built around the idea of Bayesian diagnosis. A key feature of the system is its ability to handle uncertainties, human operator actions, and time-dependent or trajectory-type input information. The hierarchical structure of a sample knowledge base is shown for a typical pressurized water reactor, and numerical examples are presented.