PROBLEM SELECTION TO CONTROL CENTRE OPERATORS TRAINING

摘要

Usually, the training programs available in industrial environment do not consider the particular training needs of each trainee. Individualized instruction is the main goal of an ITS. To achieve this, an ITS maintains a trainee model, which models the trainee's understanding of domain concepts. In a learning-by-doing system, one way to achieve this individualization is to provide the system with means to choose the next problem for the trainee. The chosen problem must be the most appropriate to the trainee's knowledge status. This paper presents a Curriculum Planning module used in an ITS to train Control Centre operators of the Portuguese electrical transmission network. The Curriculum Planning module involves two stages. First, the tutor must define the difficulty level of the problem, considering the trainee's progression. Heuristic rules are used to calculate the difficulty level. In the second phase, the tutor must use the contents of the user model to choose the type or types of the most suitable incidents to be included in the problem. A classification is performed in order to get the types of incidents that better match the trainee needs. The classification algorithm is based on a simple neural network in which activation values of the output units represent a measure of the adequacy of each type of incident to the trainee.