A Network-based Distributed System For Glass Quality On-line Detection

As the rapid progress of industrial automation and the wide-spread application of computer, the technology of on-line quality detection has been changed radically. The higher quality of float glass, its functional-diversification and the internationalization of its production order require higher technology of detection to achieve higher precision. The way of measure has changed into 100-percent-detection from spot-inspection; the items of detection have been driven from simple to sophisticated, from single to Multi-comprehensive. The distributed approach of quality detection based on multicamera-vision can reduce the expense on detection, improve the quality of the production and increase the efficiency of producing. All these guarantee the innovated way of detection an extremely extensive prosperity.In this thesis, the background, the current condition, the future trend, the purpose and the significance of the research is firstly reviewed.Secondly, through analyzing the supposed fundamental functions which might be carried out by the glass quality on-line detection system, it introduces the process of implementation in coordination between each constituent of the system. The detection tasks are real-time and distributed ones, so we adopt C/S model to construct a distributed network prototype so as to match these characteristics. The required key techniques meant to accomplish the distributed tasks of detection are also given.Thirdly, as to fulfill the detection tasks, the most significance is to resolve the problems of synchronization in the progress of image acquisition and to perform distributed synchronous processing techniques. In this thesis the detailed analysis of above-mentioned techniques provides the implementation principles and strategies for the synchronous image acquisition and distributed processing: impulse signals from encoder as external triggers synchronize the image acquisition in 8 CCD cameras; the cutting signals from the proximity switch located on the glass cutting apparatus synchronize the image acquisition and glass quality classification by ink-jet marker.In addition, it discusses on the realization of communications within the online detection system and two main threads are demonstrated: the client process and the server process. Also a probability analysis on network congestion was taken and compared with other congestion control approaches, a proper congestion-controlling mechanism based on the predictable RTT is employed for detecting and resolving congestion by which mainly to achieve congestion avoidance under the particular conditions in this network-based system. Furthermore, a method is proposed for a client to automatically diagnose the connection status between the server and the client and to re-connect to the server when the disconnection is detected.In the development implementation, a practical on-line quality detection system for float glass which is developed with the distributed network platform is presented, and the proper performance has also been demonstrated in this thesis.In the end, all works of the thesis are summarized and all the problems that need to be solved in the future are also pointed out.