The Nuclear Export and Import Control System (NEPS) is currently in operation for nuclear export and import control. To ensure consistent and efficient control, various computational systems are either already in place or being developed. With numerous scattered systems, it becomes crucial to integrate the databases from each to maximize their utility. In order to effectively utilize these scattered computer systems, it is necessary to integrate the databases of each system and develop an associated search system that can be used for integrated databases, so we investigated and analyzed the AI language model that can be applied to the associated search system. Language Models (LM) are primarily divided into two categories: understanding and generative. Understanding Language Models aim to precisely comprehend and analyze the provided text’s meaning. They consider the text’s bidirectional context to understand its deeper implications and are used in tasks such as text classification, sentiment analysis, question answering, and named entity recognition. In contrast, Generative Language Models focus on generating new text based on the given context. They produce new textual content continuously and are beneficial for text generation, machine translation, sentence completion, and storytelling. Given that the primary purpose of our associated search system is to comprehend user sentences or queries accurately, understanding language models are deemed more suitable. Among the understanding language models, we examined BERT and its derivatives, RoBERTa and DeBERTa. BERT (Bidirectional Encoder Representations from Transformers) uses a Bidirectional Transformer Encoder to understand the sentence context and engages in pre-training by predicting ‘MASKED’ segments. RoBERTa (A Robustly Optimized BERT Pre-training Approach) enhances BERT by optimizing its training methods and data processing. Although its core architecture is similar to BERT, it incorporates improvements such as eliminating the NSP (Next Sentence Prediction) task, introducing dynamic masking techniques, and refining training data volume, methodologies, and hyperparameters. DeBERTa (Decoding-enhanced BERT with disentangled attention) introduces a disentangled attention mechanism to the BERT architecture, calculating the relative importance score between word pairs to distribute attention more effectively and improve performance. In analyzing the three models, RoBERTa and DeBERTa demonstrated superior performance compared to BERT. However, considering factors like the acquisition and processing of training data, training time, and associated costs, these superior models may require additional efforts and resources. It’s therefore crucial to select a language model by evaluating the economic implications, objectives, training strategies, performance-assessing datasets, and hardware environments. Additionally, it was noted that by fine-tuning with methods from RoBERTa or DeBERTa based on pre-trained BERT models, the training speed could be significantly improved.

About 83% of the information systems of administrative and public agencies are operated by agencies, and most of them are vulnerable to security due to the small scale of operation, insufficient facilities, and lack of dedicated personnel. To address these issues, the Ministry of the Interior and Safety announced in June 2021 that, as part of the “Second Basic Plan for E-Government,” all information systems of administrative and public agencies will be converted to a cloud-based integrated management operating environment by 2025 to provide stable public services. Accordingly, relevant laws and guidelines should be researched and analyzed to prepare for the cloud conversion of the Nuclear Export and Import Control System (NEPS) operated by the Export and Import Control Office of the Korea Institute of Nuclear Nonproliferation and Control (KINAC). The Cloud Computing Act defines cloud computing, establishes a basic plan and implementation plan, provides support for promoting the adoption of cloud computing by state institutions, supports the construction of integrated information and communication facilities based on cloud computing technology, provides cost and technical support, and regulates cloud security certification, and applies the Personal Information Protection Act and the Act on Promotion of Information and Communication Network Utilization and Information Protection to protect personal information. The E-Government Act defines integrated standards and principles for information resources, support for the use of cloud computing services, classification standards for information resources, and integrated standards for calculating the size and capacity of information systems. The Notice on Standards for Using Cloud Computing Services and Securing Safety for Administrative Agencies and Public Institutions specifies the standards for using cloud computing services and measures to secure stability for administrative agencies, contracts for using cloud computing services, and ensuring continuity of cloud computing services. The Basic Guidelines on National Information Security stipulate the establishment and implementation of security measures, system security, user security, security management, information and communication network security separating internal network and internet network, and cloud computing security measures, and stipulate the NIS security review when introducing private cloud services. In order to convert NEPS to the cloud computing services, network, and software design plans, transfer plans, and cloud operation plans will be established in compliance with the relevant laws and guidelines. And future research will include researching the system status of major public and private cloud service providers and analyzing their advantages and disadvantages.

Hydrogen isotopes (H, D, T) separation technologies have received great interest for treatments of tritiated liquid waste produced in Fukushima. In addition, the separated deuterium and tritium can be utilized in various industries such as semiconductors and nuclear fusion as expensive and rare resources. However, separating hydrogen isotopes in gas and liquid forms still requires energyintensive processes. To improve efficiency and performance of hydrogen isotope separation, we are developing water electrolysis, cryosorption, distillation, isotope exchange, and hydrophobic catalyst technologies. Furthermore, an analytical method is studied to evaluate the separation of hydrogen isotopes. This presentation introduces the current status of hydrogen isotope research in this research group.

With the rapid improvement of hardware and software-related IT technology, applying A.I. to the private and public sectors, such as the Food Poisoning Prevention Program in Nevada and Smart City based on big data in Boston, is steadily increasing. However, the cases of application to the regulation sector of government are still insufficient. The Korea Institute of Nonproliferation and Control (KINAC) is studying to apply A.I. technology to the regulation to improve the objectivity, consistency, and efficiency of classification and export licensing review. The KINAC developed the Nation Nuclear Technology Information Collection and Analysis System using A.I. techniques such as machine learning and deep learning techniques. KINAC and FNC Technology are developing the Export Risk Assessment System using A.I. modules and Bayesian Networks. The KINAC and Korea Atomic Energy Research Institute (KAERI) are developing an inventory history management system subject to the Nuclear Cooperation Agreements. The Nuclear Safety and Security Commission (NSSC) and KINAC operate the Nuclear Import and Export Control System (NEPS) for application and export license review according to relevant laws such as the Foreign Trade Act. Therefore, preparing an integration plan for the existing NEPS and the new systems is necessary. Since the NEPS has to be operated and accessible at all times, so the stability of the NEPS is the most important when integration and linking. So, it is suggested that the Collection and Analysis System and the Risk Assessment System, which require a lot of data traffic, are configured in a server separate from the NEPS, and the new DB and the NEPS DB are only linked. An inventory history management system is also suggested to be integrated and configured into the NEPS. Third, it is recommended that each system lists the information provided to or received from the NEPS in advance, and one-way communication should be performed basically. Two-way communication should be performed when necessary. Finally, against various cyber accidents and information leakage, it is proposed to review security vulnerabilities and apply essential security measures and guidelines. Through the integration and linkage of these systems, it is expected that the objectivity, consistency, and efficiency of classification and export licensing review of the KINAC are strengthened, and national transparency of development, production, and use of nuclear material is enhanced. It can be satisfied with the increasingly strengthened demands of the international community on duty for strategic item management.

2020년 1월 1일부터 국제해사기구(IMO)는 전 세계 모든 해역을 지나가는 선박을 대상으로 선박연료유의 황 함유량 상한선을 3.5 %에서 0.5 %로 낮춰 선박으로 인해 발생하는 대기오염을 줄이기 위한 강력한 규제를 실시한다. 황 함유량이 낮은 연료유를 사용하여 대기오염 물질을 줄이는 것도 중요하지만 선박을 경제적으로 운영하여 불필요한 에너지 낭비를 줄이는 것 또한 대기오염 물질을 줄이는 데 큰 도움이 된다. 따라서 선박은 잡음의 영향을 받더라도 항로를 정확하게 유지하여야 한다. 항로를 정확하게 추종하기 위해 오토파일럿 시스템이 사용되지만 오토파일럿 시스템의 성능이 아무리 우수하다 하더라도 잡음의 영향을 받게 된다면 성능에 한계를 가진다. 실제 환경에서는 자이로스코프에서 측정잡음이 더해진 회두각이 오토파일럿 시스템의 입력으로 들어가 오토파일럿 시스템의 성능을 저하시킨다. 이와 같은 문제를 해결하기 위해 상태추정에 쓰이는 Kalman Filter를 적용하여 잡음의 영향을 줄여주는 기법이 있지만 이 또한 역시 잡음의 영향을 완전히 제거시키는 것이 불가능하다. 따라서 본 논문에서는 잡음제거 성능을 더욱 더 개선시키기 위해 전진방향 구간에서는 인공지능 기술 중 하나인 다층퍼셉트론(Multi-Layer Perceptron; MLP)를 적용하고, 회전구간에서는 Kalman Filter를 적용하여 Kalman Filter만을 사용한 경우보다 우수한 잡음제거 기법을 제안한다. 시뮬레이션을 통해 제안한 방법이 Kalman Filter만을 사용한 경우보다 조타기의 오동작을 방지하여 선박의 전진방향 운동이 개선됨을 확인할 수 있다.