Jonghyun Kim

Systematic development of scenarios caused by cyber-attack-induced human errors in nuclear power plants

The digitalization of the instrumentation and control (I&C) systems in nuclear power plants has increased the threat from cyber-attack. This paper addresses cyber-attack vulnerability from potential attacks on non-safety I&C systems that result in a hazardous plant status. In safety-critical applications such as nuclear power plants, safety systems are separated and isolated from non-safety systems by design. Cyber-attacks on the non-safety systems can however escalate into plant safety threats by inducing wrong operator actions. We focus here on the operator actions that lead to the unavailability of the safety system and cause the failure of accident mitigation. In this study, the effect of safety system unavailability on plant safety is carefully modeled by using n conventional fault tree (FT) analyses, and human actions are analyzed based on emergency operating procedures. Based on the results of the FT analyses in combination with human action analyses, we suggest a novel method to systematically develop cyber-attack propagation scenarios, where a cyber-attack is linked to its consequences. As a case study, the feed-and-bleed operation is analyzed with the developed FT, producing multiple scenarios that lead to core damage. The results of this study are expected to be useful in establishing preventive measures

Autonomous Algorithm for Start-Up Operation of Nuclear Power Plants by Using LSTM

Autonomous operation is one of the technologies of the forth-industrial revolution that is attracting attention in the world due to the development of new computer algorithms and the hardware performance. Its main core technology is based on artificial intelligent (AI). Autonomous control, which is a high level of automation, is having the power or ability of self-governance in the overall system without human intervention. This study aims to develop an autonomous algorithm to control the NPPs during start-up operation by using Long-Short Term Memory (LSTM) that is one of the recurrent neural network (RNN) methods. RNN, which is a type of AI method, is suitable for application to the NPP system because it can help to calculate the interaction of non-linear parameters as well as to capture the pattern of time series parameters. This study suggests a conceptual design for autonomous operation during start-up operation from 2% power to 100% power in nuclear power plants.

Accident Diagnosis and Autonomous Control of Safety Functions During the Startup Operation of Nuclear Power Plants Using LSTM

Accident diagnosis is regarded as one of the complex tasks for nuclear power plant (NPP) operators. In addition, if the accident occurs during the startup operation, it is hard to cope with the situation appropriately because the initial conditions are different from the normal operation mode. Although operating procedures are provided to operators, accident diagnosis and control for recovery are difficult tasks under extremely stressful conditions. In order to achieve safe operation during the startup operation, this study proposes algorithms not only for accident diagnosis but also for protection control using long short-term memory (LSTM), which is an advanced version of recurrent neural networks, and functional requirement analysis (FRA). Using the LSTM, the network structures of algorithms are built. In addition, FRA is performed to define the goal, functions, processes, systems, and components for protection control. This approach was trained and validated with a compact nuclear simulator for several accidents to demonstrate the feasibility of diagnosis and correct response under startup operation.

Autonomous Algorithm for Safety Systems of the Nuclear Power Plant by Using the Deep Learning

This study aims to develop an autonomous algorithm to control the safety systems of nuclear power plant (NPP) by using the deep learning that is one of machine learning methods. The autonomous algorithm has two main goals. First, it achieves a high level of automation for nine safety functions of NPP. Second, the algorithm controls the nine safety functions in an integrated way. The function-based hierarchical framework is suggested to represent the multi-level structure that models NPP safety systems with the levels of goal, function and system. The function-based hierarchical framework is used to model the NPP for the application of the multi-system deep learning network. Multi-system deep learning network is applied to develop the algorithm for autonomous control. This approach enables the systematic analysis of power plant system and development of the database for the deep learning network.