A computer science perspective of bridge design
Communications of the ACM
The design and implementation of an object-oriented toolkit for 3D graphics and visualization
Proceedings of the 7th conference on Visualization '96
Issues and Applications of Case Based Reasoning to Design
Issues and Applications of Case Based Reasoning to Design
Formal Engineering Design Synthesis
Formal Engineering Design Synthesis
Simulation data mining: a new form of computer simulation output
WSC '05 Proceedings of the 37th conference on Winter simulation
Proceedings of the 38th conference on Winter simulation
Active learning for class imbalance problem
SIGIR '07 Proceedings of the 30th annual international ACM SIGIR conference on Research and development in information retrieval
An Algorithm Selection Approach for Simulation Systems
Proceedings of the 22nd Workshop on Principles of Advanced and Distributed Simulation
Integrating Data Mining and Agent Based Modeling and Simulation
ICDM '09 Proceedings of the 9th Industrial Conference on Advances in Data Mining. Applications and Theoretical Aspects
Data mining application on crash simulation data of occupant restraint system
Expert Systems with Applications: An International Journal
Data mining on crash simulation data
MLDM'05 Proceedings of the 4th international conference on Machine Learning and Data Mining in Pattern Recognition
| Hi-index | 0.00 |
We introduce simulation data mining as an approach to extract knowledge and decision rules from simulation results. The acquired knowledge can be utilized to provide preliminary answers and immediate feedback if a precise analysis is not at hand, or if waiting for the actual simulation results will considerably impair the interaction between a human designer and the computer. This paper reports on a bridge design project in civil engineering where the motivation to apply simulation data mining is twofold: (1) when dealing with real-world bridge models the simulation efficiency is inadequate to gain true interactivity during the design process, and (2) the designers are confronted with a parameter space (the design space) of enormous size, from which they can analyze only a small fraction. To address both issues, we propose that a database of models (the design variants) should be pre-computed so that the behavior of similar models can be used to guide decision making. In particular, simulation results based on displacement, strain, and stress analyses are clustered to identify models with similar behavior, which may not be obvious in the design space. By means of machine learning, the clustering results obtained in the simulation space can be transferred back into the design space in the form of a highly non-linear similarity measure that compares two design alternatives based on relevant physical connections. If the assessments of the measure are reliable, it will perfectly address the mentioned issues above. With this approach we break new ground, and our paper details the technology and its application for a real-world design setting.