Research On Ubiquitous Sensing In Wireless Industrial Monitoring System

With the development of computer and communication technologies, industrial wireless monitoring system increases the scope of wire monitoring network by wireless technology and data acquisition capability achieves an unprecedented level. But the network size and complexity increase. There are issues which are the lack of system structure and information presentation, data redundancy, unreliable information delivery, information inconsistency in the industrial wireless monitoring system. These issues are the main obstacles to apply wireless technology in industrial monitoring system.In this thesis, ubiquitous computing and context-aware computing are the guide theory. Industrial wireless monitoring system is seen as a ubiquitous computing environment and the condition of the monitoring equipment is a context-aware process. In under the guidance of context-aware computing, the problems of the industrial wireless monitoring system are analyzed. In order to solve the problems, ubiquitous sensing system is proposed in this paper.According to the process of extracting information from industrial environment, the generalized model of ubiquitous sensing system is built. Generalized model will be divided into physical space and three information spaces which are Sensor, System and Service. Generalized volumes are used to analyze the features of data. Generalized operators are used to analyze the relationship between the four spaces. Key issues of ubiquitous sensing system, which are data flow anomaly detection, information transmission mechanism, the consistency of information, are unified into this generalized model. To address these three issues, different theories are used to solve diferent issues.The nature of outlier detection in the industrial wireless monitoring system is the problem of dataflow anomaly detection.A algorithm of outlier detection baesd on windows aggregation is proposed and a data compression algorithm is derived in the thesis. Outlier detection of high-speed sampling data flow in sensor is solved by these algorithms .The algorithms are suitable for use in wireless sensors with limited resources.We conduct a comparative investigation of existing data transport mechanisms of Application Layer in ubiquitous sensing system and points out their deficiencies. We also introduce the closed-loop event-triggered data transport mechanism in ubiquitous sensing system. According to the feedback of destination node, the release node evaluates whether to send data in this mechanism. This mechanism improves the delivery rate and decreases data redundancy and communication traffic in ubiquitous sensing system.Information of industrial wireless monitoring system comes from different physical sensors. However, due to the spatial distribution of differences, the information can not guarantee consistency. In order to resolve inconsistencies of information, a algorithm of distributed data fusion based on Principal Component Analysis is proposed. A data aggregation time control algorithm based on the binomial distribution is proposed. These algorithms can get the consistency of information by different sensors.At last, we built a ubiquitous sensing system in rotogravure printing press with the research results above. Reliable operation of the system verifies the validity of our research.With the guidance of ubiquitous computing and context-aware computing, we build the theoretical framework and analysis the core problems of ubiquitous sensing system. Some algorithms are proposed in order to sovle these core problems which are model of ubiquitous sensing system, outlier detection, information transport mechanism and consistency of information. The results of this thesis provide a useful reference in building an industrial wireless monitoring system. To enhance the development of industrial wireless technology is of great significance.