Search

If safety injection systems (SISs) do not work in the event of a loss-of-coolant accident (LOCA), the accident can progress to a severe accident in which the reactor core is exposed and the reactor vessel fails. Therefore, it is considered that a technology that provides recoverable maximum time for SIS actuation is necessary to prevent this progression. In this study, the corresponding time was defined as the golden time. To achieve the objective of accurately predicting the golden time, the prediction was performed using the deep fuzzy neural network (DFNN) with rule-dropout. The DFNN with rule-dropout has an architecture in which many of the fuzzy neural networks (FNNs) are connected and is a method in which the fuzzy rule numbers, which are directly related to the number of nodes in the FNN that affect inference performance, are properly adjusted by a genetic algorithm. The golden time prediction performance of the DFNN model with rule-dropout was better than that of the support vector regression model. By using the prediction result through the proposed DFNN with rule-dropout, it is expected to prevent the aggravation of the accidents by providing the maximum remaining time for SIS recovery, which failed in the LOCA situation.

In this study, we present a preprocess method using radiation energy intervals on a gamma-ray spectrum based on a deep learning algorithm to achieve real-time radionuclide identification. Data preprocessing is performed by classifying energy intervals, distinctly corresponding to pulse amplitudes of each radiation measurement system. Since the energy intervals are distinguished with noise, backscatter area, Compton edge, and photopeaks depending on radionuclides, raw data are sorted in each interval in preprocessed dataset using a deep learning algorithm. Using60Co,137Cs, and the energy interval preprocessing, the multi-source identification shows 100% accuracy in 2000 measured data compared with 70% accuracy for those without the preprocessing method. The measured time is 72 s for 2000 test data, dramatically reduced from the conventional data collection time of 60 min for 100 000 data. The proposed approach reduces the minimum number of data to identify radionuclides before visualizing the spectrum. With the preprocess method, radionuclide identification is completed in tens of seconds, applicable for low radiation activity areas such as decommissioning reactor sites.

Acquiring instrumentation signals generated from nuclear power plants (NPPs) is essential to maintain nuclear reactor integrity or to mitigate an abnormal state under normal operating conditions or severe accident circumstances. However, various safety-critical instrumentation signals from NPPs cannot be accurately measured on account of instrument degradation or failure under severe accident circumstances. Reactor vessel (RV) water level, which is an accident monitoring variable directly related to reactor cooling and prevention of core exposure, was predicted by applying a few signals to deep neural networks (DNNs) during severe accidents in NPPs. Signal data were obtained by simulating the postulated loss-of-coolant accidents at hot- and cold-legs, and steam generator tube rupture using modular accident analysis program code as actual NPP accidents rarely happen. To optimize the DNN model for RV water level prediction, a genetic algorithm was used to select the numbers of hidden layers and nodes. The proposed DNN model had a small root mean square error for RV water level prediction, and performed better than the cascaded fuzzy neural network model of the previous study. Consequently, the DNN model is considered to perform well enough to provide supporting information on the RV water level to operators. Keywords: Deep neural networks, Genetic algorithm, Reactor vessel water level, Safety-critical instrumentation signals, Severe accident

A serious threat to the integrity of the reactor core, reactor coolant system, or containment is incurred if proper and essential actions to mitigate accidents cannot be taken owing to insufficient information about the internal states of the nuclear power plant (NPP). Therefore, this study was carried out to develop a model capable of mitigating the risk of severe accidents by accurately predicting the internal states of an NPP containment. A deep fuzzy neural network (DFNN) is a method in which syllogistic fuzzy reasoning is relatively efficient and inference capability is enhanced. In this study, the internal states of an NPP containment, hydrogen concentration and pressure, are predicted using a rule-dropout DFNN, as little NPP information is available under the circumstances of severe accidents. In addition, the performance of the proposed rule-dropout DFNN model is compared with that of other fuzzy neural network variations to verify the enhancement in the accuracy of the DFNN. The developed rule-dropout DFNN model is expected to be capable of providing accident monitoring information in advance for accident mitigation, as its prediction error for the hydrogen concentration and pressure in the containment is low.

Nuclear power plants (NPPs) are designed in consideration of design basis accidents (DBAs). However, if the safety injection system (SIS) is not working properly in a loss-of-coolant-accident (LOCA) situation, it can induce a severe accident that exceeds DBAs. Therefore, it is important to properly actuate the SIS before a DBA becomes a severe accident. If the SIS is not working in time, the reactor core may be uncovered and the reactor vessel (RV) may be damaged. In this paper, we defined the golden time as the available time from an initial SIS malfunction for actuating the SIS to prevent reactor core uncovery and RV failure. A support vector regression (SVR) model was applied to predict the golden time. The input variables and parameters of the SVR model were selected and optimized by using a genetic algorithm. The data set of severe accident scenarios was obtained by using the Modular Accident Analysis Program (MAAP) code. An optimized power reactor (OPR1000) was used for the simulations. It was shown that that the proposed SVR model could predict the golden time accurately.

This work presents a detector capacitance compensation circuit for array-type SiPM-based radiation detectors while maintaining high-quality pulse information, including energy and time resolution. The major component of the proposed configuration, a unity-gain amplifier exploiting the Miller effect, maintains the same AC voltages on two sides of a detector, resulting in no effective detector capacitance between the nodes. This technique ideally allows combining channels without performance degradation in terms of rise time and amplitude of signals that can be caused due to the accumulated capacitance when SiPM cells are shorted in parallel in a signal multiplexing scheme. In order to verify the configuration, we designed and built a test board inserting a unity-gain amplifier between a 16-pixel SiPM array and a conventional charge-sensitive amplifier as a preamplifier . The measured pulses at the output of the front-end system show that the detector compensation technique can successfully mitigate performance degradation of rise time and amplitude as well as coincidence time. This technique can be adopted in many applications that require a multiplexed readout of a large number of array-type detectors, e.g., in cargo scanning technologies, radiation monitoring in nuclear plant facilities, and advanced medical imaging instruments such as PET and SPECT.

The Fukushima accident induced by the great earthquake and tsunami reveals the vulnerability of I&C System. In the severe environment, the normal I&C system did not work properly and results in false information about the internal situation in NPP. Eventually the accident was not properly handled at the early stage. Therefore advanced emergency response system using a wireless channel is necessary to cope with the severe accident. In this paper, we introduce the ERS consisting of the HMS and MCS the ECS linking the HMS with MCS and the performance requirement of the ECS is analyzed. The ECS satisfying the requirement is designed conceptually and the performance of the ECS is evaluated through analysis and simulator. To secure a reliable and diverse configuration, the ECS is configured as the dual system which consists of the terrestrial communication and satellite communication. The terrestrial communication system is designed based on the IEEE 802.11. Analyzed performance results prove that the performance requirement can be sufficiently achieved. But if the scalability of data capacity is considered later, use of the advanced 802.11 standard such as 802.11n and multiple signal paths between the HMS and MCS are necessary. In the satellite communication system, the FDMA is used in the status link and the DSSS is used in the control link. The network supporting various data rates is designed and the communication link budget is analyzed considering the link availability of 99%. The designed satellite communication system can secure the link margin of 1.3 – 15.7 dB.

Fukushima accident was worse by instrument inability. Eventually, the accident was not mitigated and keeping the integrity of reactor was failed since the operators was not able to quickly understand the situation and take necessary actions. Therefore, in this study, reactor vessel (RV) water level considered as one of the parameters to keep the integrity of reactor is predicted in loss of coolant accident (LOCA) situation using the deep neural network (DNN) method. This is in an effort to provide supporting information under the severe circumstance. The simulation data obtained from modular accident analysis program (MAAP) are applied to the DNN method to check the prediction performance of the RV water level. The prediction performance of RV water level using the proposed DNN model is presented as root mean square error (RMSE). Although the data of several circumstances among a variety of LOCAs are applied, good prediction performance is shown using the proposed DNN method.

Providing information about the leak flow rate caused by a loss-of-coolant accident (LOCA) to nuclear power plant (NPP) operation personnel is a key to the management and mitigation of severe post-LOCA circumstances at NPPs where active safety injection systems do not actuate. The leak flow rate is a function of break size, differential pressure (i.e., difference between internal and external reactor vessel pressure), temperature, and so on. In this study, the break position and size were first identified and predicted, and then, the leak flow rate was predicted using a fuzzy neural network (FNN). The FNN was developed using training data and validated using independent test data. The data were generated from simulations of the optimized power reactor 1000 (OPR1000) using MAAP4 code. The data for training the FNN model were selected among the acquired data using the subtractive clustering method, and FNN performance was improved. The developed FNN model was sufficiently accurate to be used for predicting leak flow rate, which is useful information for managing severe post-LOCA situations.

This study suggests a generalized availability and safety evaluation model to evaluate the influences to the system`s fault tolerant capabilities depending on automatic fault detection function such as the automatic periodic testings. The conventional evaluation model of automatic fault detection function deals only with the self diagnostics, and supposes that the fault detection coverage of self diagnostics is always constant. But all of the fault detection methods could be degraded. For example, the periodic surveillance test has the potential human errors or test equipment errors, the self diagnostics has the potential degradation of built-in logics, and the automatic periodic testing has the potential degradation of automatic test facilities. The suggested evaluation models have incorporated the loss or erroneous behaviors of the automatic fault detection methods. The availability and the safety of each module of the safety grade platform have been evaluated as they were applied the automatic periodic test methodology and the fault tolerant evaluation models. The availability and safety of the safety grade platform were improved when applied the automatic periodic testing. Especially the fault tolerant capability of the processor module with a weak self-diagnostics and the process parameter input modules were dramatically improved compared to the conventional cases. In addition, as a result of the safety evaluation of the digital reactor protection system, the system safety of the digital parts was improved about 4 times compared to the conventional cases.

This paper presents a self-tuning predictive control algorithm for steam generator. The control algorithm is derived by suitably modifying the generalized predictive control algorithm. The main feature of the proposed method is to take the measurable disturbance into problem consideration and a simple adaptive scheme for obtaining the controller gain. In order to evaluate the performance of the proposed algorithm, computer simulations are done for an PWR steam generator model. Simulation results show satisfactory performances against load variations, steam flow rate estimation errors, and feedwater temperature variations. It can be also observed that the proposed algorithm exhibits better responses than a conventional PID controller.

The reactor protection system (RPS) is designed to minimize a propagation of the abnormal or accident conditions of nuclear power plants. We have developed a KINCS-RPS prototype by using POSAFE PLC. The main design feature of the KNICS RPS is to satisfy the APR 1400 design requirements and to improve the availability and maintainability. The application software and hardware of the RPS are being developed according to the safety critical software life cycle and the safety H/W criteria. Especially, the software requirement specification of the RPS application software is written in a formal language for a correct software verification and validation. Also, the KNICS RPS has undergone an equipment qualification (EQ) test. The RPS prototype satisfied the nuclear EQ criteria except for some minor points. It is expected that the digital reactor protection system being developed through the 2nd phase of the KNICS project will be able to be applied to the existing NPPs for a replacement or to a new nuclear power plant.

In this paper, the design concept of the smart type motor operator valve for nuclear power plant was described. The development objective of the smart valve is to achieve superior accuracy, long-term reliability, and ease of use. In this reasons, developed smart valve has fieldbus communication such as deviceNet and Profibus-DP, auto-tuning PID controller, self-diagnostics, and on-line calibration capabilities. And also, to achieve pressure, temperature, and flow control with internal PID controller, the pressure sensor and transmitter were included in this valve. And, temperature and flow signal acquisition port was prepared. The developed smart valve will be performed equipment qualification test such as environment, EMI/EMC, and vibration in Korea Test Lab. And, the valve performance is tested in a test loop which is located in Seoul National University Lab. To apply nuclear power plant, the software is being developed according to software life cycle. The developed software is verified by independent software V and V team. It is expected that the smart valve can be applied to an existing NPPs for replacing or to a new nuclear power plants. The design and fabrication of smart valve is now being processed.

Valves integrated with a smart positioner have been developed at many sites. We have been developing a smart valve with a sensor for a nuclear power plant, which is performed a process sensing function and a process control function automatically. The smart valve has the functions as follows; self-sensing, auto-tuning, multi-communication, self-diagnosing, etc. The smart valve should satisfy the nuclear criteria for a nuclear application. In this paper, we developed the requirements and criteria for the valve, which contain the requirements and criteria for the valve body, the sensing part, the control part, and the communication part. We produced a prototype of the smart valve which could control the pressure on the basis of the requirements. Additionally, we established the test criteria such as the environment test, the seismic test, and the EMI test. On the basis of the requirements and criteria, we are developing a prototype of the gate type-smart valve which contains the valve body, the control circuit, the communication circuit, and the application software. The smart valve can reduce the installation costs and the maintenance costs of nuclear power plants.

This paper proposes an effective selection method of the control weighting for constrained GPC. Considered are the constraint in the control and the constraint in the rate of the control as well. The effectiveness of the proposed method is shown through an illustrative example of level control of the steam generator in low power operation.