2304 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 55, NO. 6, DECEMBER Soft-Computing-Based Car Body Deformation EES Determination for Car Crash Analysis Sy Annamária R. Várkonyi-Kóczy, Senior Member, IEEE, András Rövid, and Maria da Graça Ru Abstract—Car body deformation modeling plays a very impor- tant role in crash accident analyses, as well as in safe car body design. The determination of the energy absorbed by the defor- mation and the corresponding energy equivalent speed can be of key importance; however, their precise determination is a very difficult task. Starting from the results of crash tests, intelligent and soft methods offer a way to model the crash process itself, as wellas to determine the absorbed energy, the before-crash speed of the car, etc.In this paper, a modeling technique and an intelligent expert system are introduced, which, together, are able to follow the deformation process of car bodies in car crashes and analyze the strength of the different parts, which can significantly contribute to the improvement of the safety of car bodies. Index Terms—Car body deformation, crash analysis, energy equivalent speed (EES) determination, fuzzy and neural network (NN)-based modeling, fuzzy corner detection, fuzzy edge detec- tion, intelligent systems, three-dimensional (3-D) modeling. I. INTRODUCTION C RASH and catastrophe analysis has been a rather seldom discussed field of traditional engineering in the past. In recentime,both the research and theoretical analyses have become part of the everyday planning work [1]–[3]. The most interesting point in crash analysis is that, although the crash situations are random probability variables, the deterministic view plays an important role in them. The stochastic view, statistical analysis, and frequency testing all concern past acci- dents. Crash situations, which occur the most frequently (e.g., the characteristic features of the crash partner, the direction of the impact, and the before-crash speed) are chosen from these statistics and are used as initial parameters of crash tests. These tests are quite expensive; thus, only some hundred tests perfactory are realized annually, which isnot a sufficient amountfor accident safety.Forthe construction of optimal Manuscript received June 15, 2004;revised December 16, 2005.This work was supported by the Hungarian Fund for Scientific Research (OTKA T 035190)and theHungarian–Portuguese Intergovern S&T Cooperation Programme (P-24/03). A. R. Várkonyi-Kóczy iswith theDepartment of Measurement and Information Systems, Budapest University of Technology and Economics, Budapest 1111,Hungary, and also with the Integrated Intelligent Systems Japanese–Hungarian Laboratory, Budapest University of Technology and Eco- nomics, Budapest 1111, Hungary (e-mail: koczy@mit.bme.hu). A. Rövid is with the Integrated Intelligent Systems Japanese–Hungarian Laboratory, Budapest University of Technology and Economics, Budapest 1111,Hungary, and also with the Department of Automobiles, Budapest University of Technology and Economics, Budapest 1111,Hungary (e-mail: andras.rovid@auto.bme.hu). M. da Graça Ruano is with the University of Algarve, Campus de Gambelas, Faro 8000, Portugal (e-mail: mruano@ualg.pt). Digital Object Identifier 10.1109/TIM.2006.873796 car body structures, more crash tests are needed. Therefore, real-life tests are supplemented by computer-based simul which increase the number of analyzed cases to 1000–200 The computer-based simulations—like the tests—are limit to precisely defined deterministic cases. The statistics are for the strategy planning of the analysis. The aforementioned example clearly shows that the stochastic view does not e the deterministic methods [4], [5]. Crash analysis is also very helpful for experts of road ve accidents because their work requires simulations and da are as close to reality as possible. By developing the applied methods and algorithms, we can make the simulations more precise and thus contribute toward the determination of the factors causing the accident. The results of the analysis of crashed cars, among whic energy absorbed by the deformed car body is one of the m important, are of significance in other fields as well. They information about the deformation process itself and have a direct effect on the safety of the persons sitting in the car through the analysis of traffic accidents and car crash tes can obtain information concerning the vehicle, which can be of help in modifying the structure/parameters to improve future safety. There is an ever-increasing need for more c techniques, which need less computational time and can m widely be used. Thus, new modeling and calculating meth are highly welcome in deformation analysis. The techniques of deformation energy estimation used u until now can be classified into two main groups. The first applies the method of finite elements [6]. This procedure is accurate enough and is suitable for simulating the deform process, but this kind of simulation requires very detailed knowledge about the parameters of the car body and its e absorbing properties, which in most of the cases are not a able. Furthermore, if we want to get accurate enough res complexity can be very high. The othergroup covers the so-called energy grid-based methods, which start from known crash test data and from the shape of the deformation, or from the maximal car body deformation [7], [8]. The distribution of the energy, which be absorbed by the cells, is considered just in two dimens (2-D),and the shape of the deformation is described by a 2-D curve. This curve is the border of the deformation vis from the top view of the car body, although there are a lot of cases when the shape of the deformation cannot be descr 2-D curves. Furthermore, in many of the cases, considerin energy distribution in 2-D is not precise enough because the absorbing properties of the car body change along the ve axis as well, which is not considered by these methods. 0018-9456/$20.00 © 2006 IEEE