Martina Schnellenbach-Held

Evolutionary generation of implicative fuzzy rules for design knowledge representation

In knowledge representation by fuzzy rule based systems two reasoning mechanisms can be distinguished: conjunction-based and implication-based inference. Both approaches have complementary advantages and drawbacks depending on the structure of the knowledge that should be represented. Implicative rule bases are less sensitive to incompleteness of knowledge. However, implication-based inference has not been widely used. This disregard is probably due to the lack of suitable methods for the automated acquisition of implicative fuzzy rules. In this paper a genetic programming based approach for the data-driven extraction of implicative fuzzy rules is presented. The proposed algorithm has been applied to the acquisition of rule bases for the design of reinforced concrete structural members. Finally an outlook on the application of the presented approach within a machine learning environment for evolutionary design and optimization of complex structural systems is given.

A machine learning approach for the support of preliminary structural design

This paper deals with the representation and acquisition of structural design knowledge using fuzzy systems. A new approach for linguistic fuzzy modeling as well as a multi-objective evolutionary algorithm for the data-driven design of fuzzy systems is presented. The developed genetic fuzzy system has been applied to test problems and real-world tasks. Making use of the proposed approaches the interpretability of fuzzy systems can be increased without loss of accuracy. The developed system facilitates the knowledge acquisition and improves the maintainability of the knowledge base.

Knowledge discovery in bridge monitoring data: a soft computing approach

Road and motorway traffic has increased dramatically in Europe within the last decades. Apart from a disproportionate enlargement of the total number of heavy goods vehicles, overloaded vehicles are observed frequently. The knowledge about actual traffic loads including gross vehicle weights and axle loads as well as their probability of occurrence is of particular concern for authorities to ensure durability and security of the road networks structures. The paper presents in detail an evolutionary algorithm based data mining approach to determine gross vehicle weights and vehicle velocities from bridge measurement data. The analysis of huge amounts of data is performed in time steps by considering data of a corresponding time interval. For every time interval a population of vehicle combinations is optimized. Within this optimization process knowledge gained in the preceding time interval is incorporated. In this way, continuously measured data can be analyzed and an adequate accuracy of approximation is achieved. Single vehicles are identified in measured data, which may result from one or multiple vehicles on the bridge at a given point of time.

High Performance Aerogel Concrete

Aerogel Concrete is a promising approach to fulfill the requirements for the thermal insulation of buildings. So far, Aerogel Concretes exhibit very low thermal conductivities of 0.06 ≤ λ ≤ 0.10 W/(mK) and very low corresponding compressive strength of f cm ≤ 2.5 MPa. Thus, the suitability of Aerogel Concrete as a building material for load bearing walls was worthy of discussion. Previous efforts to achieve considerable compression strength in the range of normal strength concretes (f cm ≈ 20 MPa) were accompanied by a significant increase of the thermal conductivity up to λ ≈ 0.55 W/(mK). By embedding silica aerogel granules in a high strength cement matrix, the Institute for Structural Concrete (ISC) at the University of Duisburg-Essen developed in cooperation with the German Aerospace Center (DLR) so called High Performance Aerogel Concrete (HPAC) with an improved correlation between compression strength and thermal conductivity. The thermal conductivities of the new material are in the range of 0.09 W/(mK) ≤ λ ≤ 0.25 W/(mK) for the High Performance Aerogel Concretes considered with compression strength between 2 MPa and 25 MPa. The dry bulk density at the same amounts of aerogel granule is less compared to previous investigations on Aerogel Concrete. The Young’s modulus related to the dry bulk density is comparable to that of Lightweight Aggregate Concrete; the flexural strength is considerable less. The mixtures are flowable, nearly self-compacting and show a short hydration process as well as an increased tendency to shrink. Subsequently, the results of the investigations on HPAC are presented in detail.

Soft-WIM: Vehicle Identification from Bridge Measurement

This paper presents an approach for the identification of operational loads given measured static responses of a bridge structure and its corresponding analytical model. In an optimization process single vehicles and corresponding attributes are identified from data recorded during the presence of one or multiple vehicles on the bridge. The vehicle attributes gross weight, velocity, axle loads and axle spacings are determined. The problem is solved by means of two encapsulated evolutionary optimization routines. The objective function is represented by the calculation of the difference between measured and computed strains. For measurement, only two sensors which are installed in one cross section without any installations on top of the bridge are required. The application to real measurements as well as obtained results will be shown. The results demonstrate that structural health monitoring in conjunction with adequate mechanisms can successfully support the acquisition of additional, valuable information.

Bionic Optimization of Concrete Structures by Evolutionary Algorithms

Abstract Floor slabs represent a large volume of concrete in buildings. The goal of this research is to achieve a structure that has an optimized bearing capacity. The optimization implies economic efficiency and sustainability. This paper describes a bionic optimization process that is applied in a project of the German Research Foundation (DFG) Priority Programme called “Concrete light. Future concrete structures using bionic, mathematical and engineering formfinding principles”. The project involves adaption of three different natural structures that lead to a natural flow of forces. These natural structures are (a) spider webs, (b) hollow parts of bones and (c) geometries of structures such as the bottom side of water lilies or seashells. This scientific paper deals with the implementation of an optimization process for a configuration of reinforcement inspired by a spider web. Evolutionary Algorithms (EAs) are used for the development and optimization of an innovative and useful configuration of reinforcement. The EAs use reproduction, mutation and selection as mechanisms, inspired by biological evolution, to solve technical problems gradient-free. In this project the EA is combined with physical nonlinear finite element analyses. The EA is embedded into a C# application, in which the slab structure is generated and the finite element programme is started. The quality of the results is characterized by the fitness of each individual (reinforcement configuration), which is, for this example, the midspan displacement of the generated slab multiplied by the steel volume per slab. Accordingly, the midspan displacement is to be minimized during the process, with the minimum possible amount of reinforcement. The optimization variables are the angles and the number of rebars per slab. Several constrains need to be included to get comparable results between the developed slabs and the conventional slabs with orthogonally configured reinforcement. This paper presents the results of an optimized reinforcement configuration thus found by EA and comparisons with the behaviour of conventional slabs with a similar reinforcement ratio.

Vernetzt-kooperative Gebäudeplanung im Massivbau mit verteilten deklarativen Wissensinstanzen und Fuzzy-Methoden

Die gestiegenen Anforderungen an Planung und Ausfuhrung von Bauprojekten erfordern fur eine wirtschaftliche Projektabwicklung eine koordinierte Kommunikation sowie eine effiziente Informationsverarbeitung (Bretschneider 1998).

Multi-Objective Genetic Programming based Design of Fuzzy Systems

The Multi-Objective Domain Knowledge Augmented Genetic Fuzzy System (MODA-GFS) is a GP based fuzzy system for the data-driven generation of fuzzy rule based systems. The algorithm incorporates domain specific knowledge that is used by human knowledge engineers in the manual fuzzy system design process. The combination of characteristics of two individuals is most interesting if both individuals complement each other. In terms of fuzzy systems this means a potential crossover partner (parent B) has a lower approximation error in an area of the input space, where parent A has a higher error. Within MODA-GFS a method for the determination of feasible crossover mates is implemented. In addition MODA-GFS includes a method for the goal-oriented selection of parent rules that are handed down to the offspring. Especially in the domain of knowledge representation the quality of a fuzzy system is not only determined by its approximation capability but also by its transparancy. In order to assure the automated generation of fuzzy systems that are both accurate and transparent multiobjective optimization methods are implemented. Tests carried out on test functions as well as on real world data sets have shown that the incorporation of domain knowledge significantly speeds up the evolution process. Besides these test results the integration and application of the new methods for automated generation of fuzzy models within a learning expert system environment are described in this paper. Finally an outlook on the current and future work is given, i.e. the transfer of the presented findings to the evolutionary optimization of large-scale structures.

Besonderheiten bei der Bemessung von Bauteilen aus Konstruktionsleichtbeton nach DIN 1045‐1

Der vorliegende Beitrag gibt eine Einfuehrung in die Bemessung von Stahlbetonbauteilen aus Leichtbeton nach DIN 1045-1. Dabei werden auf der Basis der Materialeigenschaften die wesentlichen Unterschiede zur Bemessung von Normalbeton herausgestellt. Es wird verdeutlicht, dass die Bemessung von Bauteilen aus Leichtbeton keine Besonderheit mehr darstellt. In dem nun einheitlichen Bemessungskonzept fuer Leicht- und Normalbeton ist die Bemessung fuer Leichtbeton weitestgehend von einem Faktor abhaengig, der den Einfluss der Rohdichte erfasst. (A) Titel in Englisch: Characteristics on the design of components of lightweight aggregate concrete according to DIN 1045-1.

Anwendung der Fuzzy Logic im konstruktiven Ingenieurbau

In den vergangenen zwei Jahrzehnten hat die Theorie der Fuzzy Logic nahezu alle Disziplinen der Ingenieurwissenschaften erobert. Im konstruktiven Ingenieurbau wird diese neue Technologie jedoch nur sehr zogerlich eingesetzt. Das Ziel dieses Artikels ist, eine kurze Einfuhrung in die Theorie zu geben sowie mogliche Anwendungen im konstruktiven Ingenieurbau aufzuzeigen. Eine dieser Anwendungen liegt im Bereich der wissensbasierten Systeme, wo es gilt, unscharf formuliertes Wissen adaquat zu verarbeiten.