Chaotic Inventory Management as Benchmarking for Maritime Supply Chain Performance Evaluation
Agostino Bruzzone
DIP University of Genoa
via Opera Pia 15 16145 Genova, Italy
agostino@itim.unige.it
http://st.itim.unige.it
Alessandra Orsoni, Simone Viazzo
Liophant Simulation Club
info@liophant.org
www.liophant.org
Abstract
The Chaotic Systems represent a challenge as application; the study and identification of chaos in industrial systems based on measured data resulted to be very hard due to the consequence that these phenomena generates in real applications.
Charme Project faces with an industrial application where the data measured resulted affect by chaotic characteristics and phase space analysis provided the identification of strange attractors; the analysis of variance on the measured data allows to classify the strange attractors and to support the system study. In effect one of the most critical aspect related to the industrial application is the presence of sensitive independent variables affected by stochastic behaviour; unfortunately for many applications (i.e. planning) the statistical analysis is unable to provide useful solutions due to the fact that a point forecast of the real system is required for a proper conctrol.
For instance the supply chain sector is a very interesting application area with special attention to: operative management as well as risk analysis and decision support. In this area the stochastic components provide a set of non predicable behaviours in very critical variables; one interesting aspect will be to avoid the appearance of such non predicable behaviours, by re-engineering/redesign activities as well as operative control. Charme project applies experiences and techniques used to study deterministic chaos in order to identify chaotic causes in industrial stochastic systems; so Charme represents an innovative approach devoted to determine the actions for preventing such behaviours in supply chain management applications for process industry.
Charme considers the case of process production plants that, in specific boundary conditions corresponding to real situations, start to perform in a very strange way (i.e.inventory levels). The phase space analysis based on filtered data provided a confirmation of presence of a strange attractor. Charme applies critical threshold analysis in bifurcation diagrams based on the boundary conditions and parameters for the identification of analysis range. Simulation was used in order to reproduce such phenomena while clustering analysis was devoted to classification of system behaviours. Based on the classification it is possible to obtain an effective support for predictive control in order to be able to anticipate the chaotic behaviour and strange attractor appearance. This approach was already successful used on other industrial plant problems and it is including the stochastic component filtering that represent a complex component in real industrial chaotic system. Charme presentation includes experimental analysis as well as comparative results obtained by different clustering techniques used in the study in order to provide a validation of the proposed methodology as well as a performance estimation.
Back to HMS2002 Program