Back to home

Operator tracking system using particle filter for skill evaluation in nuclear power plant control room

 HE Aiguo1, KANEMOTO Shigeru1, KAZUMA Takenobu1,
FURUKAWA Takaaki2, WANG Shiyang1, and KAWAI Yo-ichi3
1. The University of Aizu, Ikkimachi, Aizu-Wakamatsu City,Fukushima965-8580, Japan (
2. Rakuten, Inc., 4-12-3 Higashi-shinagawa, Shinagawa-ku,Tokyo 140-0002, Japan (
3. BWR Training Center Corporation, 651 Chuodai Ottozawa Okuma, Futaba,Fukushima 979-1301, Japan (
Abstract: This article proposes an automated operator tracking system by the use of particle filter and image processing technology to help operator skill evaluation in nuclear power plant operator training facilities. In each of the control room of the training facilities, a full-scope plant simulator with mock-up control panels are used for real-time operator training. At this point, the operators’ behaviors and plant’s events are recorded as video-log and event-log, by a multi-media recording system to evaluate operators’ skills at training review meetings in which the instructors and trainees discuss each of their action during the training process. Multiple cameras that are placed on the control room ceiling are used in the recording system. However, the views from these cameras are limited and therefore it is not possible to thoroughly check how each of the operator approaches the target panel within an appropriate timing that corresponds to the plant’s event. For example, the instructors have to estimate operators’ real-world position from the views; and in some cases other operators might conceal the target operators from the cameras. The purpose of the proposed system is to help checking whether the position and timing of each operator is appropriate during each event or alarm occurrence, by tracking the operators from the recorded video.
To achieve this objective, the real-time image processing technology is newly introduced in this study, where particle filter is one of convenient algorithms for operator tracking. In this algorithm, the main issue is how to recognize multiple operators from the background and to get their positions within the coordinates of the control room. For this purpose, one 3-D particle filter is used for each operator wearing colored vests and the similarity calculation algorithm is based on color histogram. The particles are directly placed inside the control room. By converting the particle coordinates into camera coordinates and taking into account the distance between particles and the camera, the position and size of candidate regions on the video frame for similarity calculation can be easily calculated. The above approach is verified by using actual training environment in the Boiling Water Reactor (BWR) operator training center corporation facility.
Keyword: image processing; operator tracking; particle filter; operator training center
      Click here for viewing full text.