The quality of optical components such as lenses or mirrors can be described by shape errors and surface roughness. With increasing optic sizes, the stability of the polishing process becomes more and more important. If not empirically known, the optical surface must be measured after each polishing step. One approach is to mount sensors on the polishing head in order to measure process relevant quantities. On the basis of these data, Machine Learning algorithms can be applied for surface value prediction. The aim of this work is the stepwise development of an artificial neural network (ANN) in order to improve the accuracy of the models' prediction. The ANN is developed in the Python programming language using the Keras deep learning library. Beginning with simple network architecture and common training parameters. The model will then be optimized step-by-step through the implementation of different methods and Hyperparameter optimization (HPO). Data, which is generated by the sensor-integrated glass polishing head, is used to train the ANN-model. A representative part of these data is held back before, in order to validate the models' prediction. The so-called dataset contains measured values from multiple polishing runs, preceded by a design of experiment. After the model is trained on the dataset, it is able to predict the result of not yet performed polishing runs, with given polishing parameters. Concrete, the ANN is used to predict the resulting glass-surface quality, which includes the surface roughness and the shape accuracy, calculated by the material removal over time. The prediction by artificial neural networks reduces the polishing iterations and thus the production time.
Due to the advantages over conventional polishing strategies, polishing with non-Newtonian fluids are state of the art in precision shape correction of precision optical glass surfaces. The viscosity of such fluids is not constant since it changes as a function of shear rate and time. An example is during the shape correction by polishing with pitch or ice, where pitch flows slowly under its own weight and acts like a solid body during short periods of stress as its viscosity increases. One approach is to use thixotropic fluids like ketchup to reduce the roughness by polishing, without changing the shape of the sample. Tomato ketchup shows a time-dependent change in viscosity: the longer the ketchup undergoes shear stress, the lower is its viscosity. Therefore, in this article, a new processing is put forward to polishing glass surfaces with ketchup containing micro-sized Ce2O. Besides conventional ketchup, curry ketchup and an organic product were tested as well. An industrial robot onto the work piece surface guides the polishing head. The different types of ketchup are compared by means of roughness and shape accuracy and the potential regarding to manufacture high-precise optical glass surfaces.
The quality of optical components such as lenses or mirrors can be described by shape errors and surface roughness. With increasing optic sizes, the stability of the polishing process becomes more and more important. Parameters such as chemical stability of the slurry or tool wear are key elements for a deterministic computer controlled polishing (CCP) process. High sophisticated CCP processes such as magnetorheological finishing (MRF) or the ZEEKO bonnet polishing process rely on the stability of the relevant process parameters for the prediction of the desired material removal. Aim of this work is to monitor many process-relevant parameters by using sensors attached to the polishing head and to the polishing process. Examples are a rpm and a torque sensor mounted close to the polishing pad, a vibration sensor for the oscillation of the bearings, as well as a tilt sensor and a force sensor for measuring the polishing pressure. By means of a machine learning system, predictions of tool wear and the related surface quality shall be made. Goal is the detection of the critical influence factors during the polishing process and to have a kind of predictive maintenance system for tool path planning and for tool change intervals.