Development Of A Pneumatic Compliant Polishing Tool System And Research On Edge Effect In Polishing

The requirements of surface quality, production efficiency and relative apertureof optical aspheric and freeform surface in modern optical system became higher andhigher. Nowadays, the manufacturing of optical aspheric and freeform surface hasbeen a key problem, and its processing efficiency and cost has become an importantcriterion reflecting the national level of the optical industry or even the manufacturing.The polishing process is a pivotal step of the manufacturing of the optical surface,directly determining the final quality and duration of the surface. From Prestonequation, polishing force is an important parameter in the polishing process, whichinfluences the surface roughness and accuracy. In order to realize the deterministicpolishing, the polishing force needs to be controlled exactly. When the polishing toolpasses by the edge, the surface quality would be reduced due to the edge effect.Therefore, this paper would focus on the polishing force control and the edge effect.The control of polishing force is usually realized by the position control ofmachine tool or robot, which results the problem of force-position coupling. Apneumatic compliant polishing tool mounted on the four-axis precision CNC machinetool is developed in this paper, which is responsible for the force control, while themachine tool is only responsible for the position control. Obviously, the force controlis decoupled from the position control, which makes the control relatively easy.The accurate model is helpful for the analysis and design of the control systemand control strategy. It’s hard to build the accurate model using the traditional linearmodeling method because of ignoring many factors. The system identification methodis a “black box” modeling method based on the recorded input and output data, whichis especially suitable for modeling of the complex pneumatic nonlinear system. So inthis paper, the identification method is adopted to build the accurate system model. Meanwhile, the accurate system model can be obtained from the online identificationexperiments that are carried out in the real-time system platform, which is developedwith the data acquisition card (Advantech, PCI-1710U) and RTW toolbox inMatlab/Simulink, and can be verified by the validation experiments.The closed-loop control experiments of pneumatic servo control system forconstant force are carried out by adopting PID control, feed-forward PID control andsingle neuron PID control strategies within the real-time control system platform. Theexperimental results show that the better control can be achieved by single neuronPID control strategy. Considering the strong nonlinearity of pneumatic system and themore uncertainty resulting from the rotating spindle, the control performance may notbe excellent by using the above-mentioned control methods, especially for the variedforce. The neural network has the good ability of learning, adapting and nonlineartracking, meanwhile the constant force control experiments have verified that theneural network is more suitable for the control of nonlinear pneumatic system. So inthis paper, an online tuning neural PID control strategy is proposed to control theconstant and varied force with different spindle speed and it has strong robustness andtracking ability, especially when the polishing force is varying. In addition, the onlinetuning neural PID controller is used for the polishing force control in the polishingprocess of three-mirror and its effectiveness is verified by the surface quality afterpolishing.The method to build the above-mentioned real-time system is easy and cheapbecause the toolbox RTW can connect to the date acquisition card directly. Thereal-time system platform is convenient for development of the control system,simulation of the controller and field test, which can reduce the time for developingcontrol system greatly. Meanwhile, the results of the identification and forcecontrolled experiments show that the developed real-time system can provide therequired sampling frequency range for the pneumatic servo control system.In order to realize the deterministic polishing, the research on edge effect needsto be carried out considering the particularity of edge polishing. Edge effect wouldreduce the accuracy and effective aperture, which influences the resolution and focusing capability of optical system. In this paper, the edge effect is analyzedtheoretically and experimentally, and the accurate theoretical material removal profilefor different polishing path is modeled. In addition, some qualitative analyses abouthow to reduce the edge effect are also given.