[110] Applied multi-agent system to study collaborative behaviour of condition monitoring experts during education process
S I Popkov
Moscow State University of Psychology and Education, Russia
Presented in this study is a system based on a mathematical model of Markov random processes, which is intended for the adaptive training process and studying of collaborative behaviour of the groups of operators searching for solutions simultaneously. The set of events provided as the tasks in an isolated and protected training environment use a multi-agent system to control the difficulty level based on an assessment of the abilities and expertise level of the members of the group. The system is used for the assessment of expertise and collaboration abilities for different applied areas including, but not limited to, condition monitoring tasks. The system provides the sandbox environment to implement the possibility to check different solutions in a free form, including programming scripts, data formats and any number of designed solution environment parameter combinations. The system-related mathematical model is able to thoroughly differentiate between time-based mistakes, such as a solution provided out of time, and correctness-related mistakes, such as abnormal free input, abnormal solution and incorrect environment-related initial parameters of input. Propagation errors related to abnormal connections between the communication servers and the sandbox environment are also distinguished and properly handled. The toolset provided by the system as well as its structure allows different kinds of exercise to be implemented, both for skills assessment and expertise improvement related to the condition monitoring subject area. The preferable way of using such a system is to study collaborative behaviour through the set of various exercises in which an adequate solution is only possible through group interconnection and joint efforts. The inability to achieve proper solutions can be explored both on the group level of expertise and through individual mistakes to allow the adaptable system to provide the ways to correct such missteps via a special set of correction exercises. The system described can be used in different training environments for various educational, coaching and team cohesion purposes.
Keywords: adaptive training, Markov random processes, collaboration, expertise, sandbox environment.
Keywords: adaptive training, Markov random processes, collaboration, expertise, sandbox environment.
