From Spatio-Temporal Data to Manufacturing System Model
- PDF / 1,268,560 Bytes
- 9 Pages / 595.276 x 790.866 pts Page_size
- 113 Downloads / 169 Views
From Spatio-Temporal Data to Manufacturing System Model A Data-Knowledge Integration Approach Patrick Charpentier · Andrés Véjar
Received: 3 February 2014 / Revised: 27 March 2014 / Accepted: 8 May 2014 © Brazilian Society for Automatics–SBA 2014
Abstract This paper presents an original approach for automatic construction of a simulation model for complex manufacturing systems. The model generation is based on spatio-temporal product trajectories, and it is detailed by the integration of heterogeneous knowledge & product data flows. The products, therefore, contribute directly to the control of the system. The formal generated model, a queuing network with spatial structure, is a permanent image of the real state of the system to be modelled; it can be described as being auto-adaptive. Keywords Data-driven application · Modeling · Simulation · Product · Trajectories
1 Introduction and General Issues System modeling and identification are an important field of research in automation. The objective is to obtain a mathematical model of a system based on experimental data and available a priori knowledge. This model is intended to provide a faithful approximation of the behaviour of the Electronic supplementary material The online version of this article (doi:10.1007/s40313-014-0133-7) contains supplementary material, which is available to authorized users. P. Charpentier (B) Research Centre for Automatic Control (CRAN), CNRS (UMR 7039), University of Lorraine, Campus Sciences, BP 70239, 54506 Vandœuvre Cedex, France e-mail: [email protected] A. Véjar Systems Research Institute, Polish Academy of Sciences, ul. Newelska 6, 01-447 Warsaw, Poland e-mail: [email protected]
physical system of interest, to estimate the physical parameters or design simulation, forecasting, monitoring or control algorithms. The classical approach consists of formalizing a priori available knowledge, collecting experimental data, then estimating the structure, the parameters and uncertainties of a model, and finally validating (or invalidating) it (Garnier et al. 2007). According to the knowledge available or extracted from the process to be identified, one speaks about parametric identification (the structure of the model is known) or about non-parametric identification (the structure of the model is unknown). The research presented in this paper is a part of this complex dynamic systems modeling and identification problem; its behavior cannot be represented accurately by mathematical equations. The complexity of the systems of interest is related to the large number of entities with nontrivial interactions (nonlinear interactions, feedback loops, etc.), that prevent the observer to predict their behavior or evolution by the calculus, or make impossible implementation of predictive and resolvable equations of the systems (Bar-Yam 2003). Most of the time, these identification and modeling stages require human experts with skills on modeling methods and tools. The building model by this type of approach is usuall
Data Loading...
