文章基本信息

标题：Context-Aware Recommendation for Industrial Alarm System
本地全文：下载
作者：Márcio J. da Silva ; Carlos E. Pereira ; Marcelo Götz 等
期刊名称：IFAC PapersOnLine
印刷版ISSN：2405-8963
出版年度：2018
卷号：51
期号：11
页码：229-234
DOI：10.1016/j.ifacol.2018.08.266
语种：English
出版社：Elsevier
摘要：AbstractThe selection of alerts within an industrial plant is not an easy task, due to the huge amount of variables involved. By this fact, it is necessary to have the means to easiest the acquirement of necessary information throughout the extraction of the knowledge in order to assist operator in the decision-making process. In this sense, a context-sensitive recommendation system can identify a situation from patterns of events. From the recognition of a situation in a particular ontological context, a proactive approach based on knowledge seems to be more effective for this task. The industrial alarm management system, generally used in supervision systems and process controls, has a problem related to the cognitive overload of their operators. Unfortunately, this occurs mainly during perturbations in the process. In this context, this work aims to investigate and develop methods to adopt proper actions sensitive to the present situation in an industrial alarm management system through semantic web technology and machine learning techniques. This problem involves questions related to the way of obtaining the context data, the context model that describes the situation and the way in which the context data of the plant are related, which allows to infer a situation. In this sense, it is proposed the appliance of data mining techniques to obtain knowledge of the process, taken into consideration an adaptive interface that makes non-intrusive recommendations and help the operator in an industrial plant. To combine deterministic knowledge with probabilistic knowledge, a Bayesian Networks is used. For the concept evaluation, a case study was conducted based on a real historical database, which shows sound and promising results.
关键词：KeywordsRecommendationContextSemantic RulesBayesian Network