Application Capabilities of a General, ANN-based Cutting Model in Different Phases of Manufacturing through Automatic Determination of its Input-Output Configuration
AbstractReliable process models are extremely important in different fields of computer integrated manufacturing. Outlying the fact that closely related assignments require different model settings, the paper addresses the problem of automatic input-output configuration and generation of ANN-based process models with special emphasis on modelling of production chains. Production operations have several input and output parameters and the dependencies among them are usually non-linear, consequently, related operation models have to handle multidimensionality and non-linearity. Artificial neural networks (ANNs) can be used as operation models because they can handle strong non-linearities, large number of parameters, missing information. A lot of efforts have been made to apply ANNs for modelling manufacturing operations. The assignments to be performed determined the input-output configurations of the models, i.e. the parameters to be considered as inputs and the ones as output. Considering the input and output variables of a given task together as a set of parameters, the ANN model estimates a part of this parameter set based on the remaining part. This selection strongly influences the accuracy of the developed model, especially if dependencies between parameters are non-invertable. In different stages of production (e.g. in planning, optimisation or control) tasks are different, consequently, the estimation capabilities of the related applied models are different even if the same set of parameters is used. Based on their inherent learning capabilities, ANNs can adapt themselves to changes in the production environment and can also be used in cases where no exact knowledge is available about the dependencies among the various parameters of manufacturing. In the case if there is no exact knowledge, it is unknown which input-output configuration of an ANN can satisfy the accuracy requirements of the model building, consequently, a method is needed for automatic input-output configuration of the applied ANN model. One of the main goals of the research to be reported here was to find a general model for a set of assignments, which can satisfy accuracy requirements. Research was also focused on how to apply the general model for various tasks. The suggested optimisation procedure results in compromises among different viewpoints fulfilling daily requirements. Experiments show the applicability of this method.
Keywords: cutting press, modelling, artificial neural networks, solution of, assignments, production line optimisation
How to Cite
János Viharos, Z. (no date) “Application Capabilities of a General, ANN-based Cutting Model in Different Phases of Manufacturing through Automatic Determination of its Input-Output Configuration”, Periodica Polytechnica Mechanical Engineering, 43(2), pp. 189-196. doi: https://doi.org/N/A.