Research On Anomaly Monitoring Of The Production Site Based On Context Awareness

In the industrial production, that production mainly depending on the people and taking machine as complement is an usual mode of production. Due to the participation of people, the operating process is different from automated production, whose every step is a mechanical action. Abnormity caused by improper operation may result in decline in product quality and rework; even worse, it may lead to major accidents. Therefore, it’s necessary to do the real-time monitoring of the production process. To ensure production quality, reducing cost, reducing the accident probability, it is of great practical significance to monitor the production process and judge whether an abnormity occurs. This paper mainly studies how to carry out the real-time monitoring for production site according to the perceived contextual information in the pervasive computing environment.Production is a very complex activity, and the monitor of production in pervasive computing environment has the following characteristics: contextual information with multiple sources; abnormity recognition referring to context information in different time; abnormity recognition combined with the phase recognition at different stages. According to these characteristics, this thesis mainly studies two questions:(1) abnormity representation and monitoring:abnormity representation is one of the key issues in the anomaly monitoring of the production site, and the users are the production operators. They do not understand the ubiquitous computing, context awareness and complex event processing technology. There is a need to design an information representation method that is close to the natural language to accurately represent the abnormity information in production. (2) multi-stage abnormity monitoring:according to different monitoring requirements of different production stages, the production process is divided into many stages. Each stage includes transitional phases at the beginning and ending to identify differences and carry out abnormity monitoring.On the basis of analyzing related research, this paper constructs the context model and event model and proposes a Event-Context method to describe abnormal information. It represente anomalies with events and contexts. The paper classifies the different types of anomalies into seven categories, and then converts them into complex events. The Esper, a complex event processing engine, is used to identify anomalies. The paper also puts forward a method to divide multi-stages and conditions of stage conversion by Event-Context. It elaborates on the process of identifying different stages and monitoring anomalies. It tests the sampling process of microstructure analysis experiment to analyze the experimental results and efficiency and proves the feasibility of the method.