Research On Shape Measurement And Shape Control

This thesis is mainly concerned with problems of shape measurement and shape control.First introducing shape and its relevant concepts, the thesis investigates the existing methods of shape measurement and shape control inland and abroad and summarizes a variety of measurement methods. Shape functions, control methods, control models as well as regulation and control efficiency are analyzed in the view of automatic control.Taking various factors into account, aiming at contacting-shape measurement mainly studied in the thesis, we select electric resistance strain gage to form pressure sensor that is used as shape measurement element. A feasible sensor structure is formulated through experiments on such factors as affixing technology of strain gage and distance from strain gage to the surface of detecting roller. The following experimental results and the characteristics of the sensor show that the selection of the measuring elements fully meets the requirements of shapemeasurement.Wireless transmission of shape signals, which transmit signals from inside the measuring roller to the outside receiving set by infrared, is implemented by the design of dynamic measurement device and dynamic measurement circuit, which overcomes a lot of shortcomings caused by slip-ring devices used before. In addition, chip ADuC812 produced by Analog Device Co. Ltd. of the United States is adapted in the measurement circuit so that data collection of high speed and high accuracy is realized with low cost.The thesis presents a neural network-fuzzy system scheme for shape control in terms of characteristics of Sendzimir rolling mill and operation process of manual control shape, and has a deep research on the problem of neural network shape pattern recognition. Finally, large quantities of computer simulations illustrate that application of neural network into shape pattern recognition can achieve satisfactory effects.