Study of diagnostic system using machine learning to reduce the incidents of the plant is in advan because an accident causes large damage about human, economic and social loss. There is a proble m that 2 performances between a classification performance and generalization performance on the m hine diagnostic machine is exclusive. However, multi agent diagnostic system makes it possible to use a diagnostic machine specialized either performance by multi diagnostic machines can be used. We propose method t...
英字タイトル:
Method to predict process signals to learn using SVM

For accomplishing the safety of a large-scale engineering plant, the common approach is to suppose various emergency situations and prepare in advance effective measures. However, there may happen an unanticipated event. Therefore, it is very important to generate support information of operators to take responsive actions. The previous study developed a technique to generate plausible operation procedures for an emergency situation based on a functional model. Although the algorithm produces some plausi...
英字タイトル:
Investigation of indices to prioritize counter operation procedures in an emergency plant situation

Abstract : A hybrid diagnostic agent system is developed to detect and identify early an anomaly that happens in the fast-breeder reactor “Monju”. The system outputs a diagnostic result by integrating the results of diagnosis by four diagnostic software agents. They are (1) an estimation agent of overall heat transfer coefficient of evaporator and superheater, (2) a state identification agent based on SVM (Support Vector Machine), (3) an anomaly detection agent by WT (Wavelet Transformation), and (4) a...
英字タイトル:
Whole System and Estimation Technique of Overall Heat Transfer Coefficient of Evaporator and Superheater

In the present study, early detection of anomaly and failure identification have been studied based on the signals simulated based on the real plant parameters. The specific feature of the present method is the use of similarity defined based on the multi-attributes of measured parameters. The frequency and time domain characteristics have been adopted as the attributes showing the symptom of failures. The five failure transients and four kinds of specific signal anomaly have been simulated and used as...
英字タイトル:
Anomaly Diagnosis based on Multi-attribute Similarity

In recent years, human resource development in the field of plant operation has become one of the major problems. The development of novice plant operators and core human resources is an important matter. However, it takes a large time and effort. This study aims at saving time and effort required for the human resource development by identifying requirements for novice plant operators to carry out exhaustive operation. In accordance with that, this study analyzes omissions of tasks in an operation experi...
英字タイトル:
Requirements of a training manual for acquiring basic ability with the exhaustive operation for novice plant operators : an exploratory study