Nikos D. Lagaros

Structural reliability analyis of elastic-plastic structures using neural networks and Monte Carlo simulation

Abstract This paper examines the application of Neural Networks (NN) to the reliability analysis of complex structural systems in connection with Monte Carlo Simulation (MCS). The failure of the system is associated with the plastic collapse. The use of NN was motivated by the approximate concepts inherent in reliability analysis and the time consuming repeated analyses required for MCS. A Back Propagation algorithm is implemented for training the NN utilising available information generated from selected elasto-plastic analyses. The trained NN is then used to compute the critical load factor due to different sets of basic random variables leading to close prediction of the probability of failure. The use of MCS with Importance Sampling further improves the prediction of the probability of failure with Neural Networks.

Structural optimization using evolution strategies and neural networks

The objective of this paper is to investigate the efficiency of combinatorial optimization methods, in particular algorithms based on evolution strategies (ES) when incorporated into the solution of large-scale, continuous or discrete, structural optimization problems. Two types of applications have been investigated, namely shape and sizing structural optimization problems. Furthermore, a neural network (NN) model is used in order to replace the structural analysis phase and to compute the necessary data for the ES optimization procedure. The use of NN was motivated by the time-consuming repeated analyses required by ES during the optimization process. A back propagation algorithm is implemented for training the NN using data derived from selected analyses. The trained NN is then used to predict, within an acceptable accuracy, the values of the objective and constraint functions. The numerical tests presented demonstrate the computational advantages of the proposed approach which become more pronounced in large-scale optimization problems.

STRUCTURAL SHAPE OPTIMIZATION USING EVOLUTION STRATEGIES

The objective of this paper is to investigate the efficiency of combinatorial optimization methods and in particular algorithms based on evolution strategies, when incorporated into shape optimization problems. Evolution strategy algorithms are used either on a stand-alone basis, or combined with a conventional mathematical programming technique. The numerical tests presented demonstrate the computational advantages of the proposed approach which become more pronounced in large-scale optimization problems and/or parallel computing environment.

Multicomponent incremental dynamic analysis considering variable incident angle

Performance-based earthquake engineering (PBEE) is the current trend in designing earthquake-resistant structures. The implementation of the PBEE framework requires the assessment of the structural capacity in multiple earthquake hazard levels. Incremental dynamic analysis (IDA) is considered to be one of the most efficient computational tools for estimating structural capacity; therefore, it is often incorporated into the PBEE framework. Most real world reinforced concrete (RC) buildings can only be represented accurately with three-dimensional (3D) models; hence, a multicomponent incremental dynamic analysis (MIDA) is required in order to carry out an IDA-based PBEE framework. In this work, the implementation of IDA studies in 3D structures is examined, where a two-component seismic excitation is applied, and a new procedure for performing MIDA is proposed. According to the proposed procedure, the MIDA is performed over a sample of record-incident angle pairs that are generated using Latin hypercube sampling (LHS).

Life-cycle cost assessment of optimally designed reinforced concrete buildings under seismic actions

Life-cycle cost analysis (LCCA) is an assessment tool for studying the performance of systems in many fields of engineering. In earthquake engineering LCCA demands the calculation of the cost components that are related to the performance of the structure in multiple earthquake hazard levels. Incremental static and dynamic analyses are two procedures that can be used for estimating the seismic capacity of a structural system and can therefore be incorporated into the LCCA methodology. In this work the effect of the analysis procedure, the number of seismic records imposed, the performance criterion used and the structural type (regular or irregular) is investigated, on the life-cycle cost analysis of 3D reinforced concrete structures. Furthermore, the influence of uncertainties on the seismic response of structural systems and their impact on LCCA is examined. The uncertainty on the material properties, the cross-section dimensions and the record-incident angle is taking into account with the incorporation of the Latin hypercube sampling method into the incremental dynamic analysis procedure. In addition, the LCCA methodology is used as an assessment tool for the designs obtained by means of prescriptive and performance-based optimum design methodologies. The first one is obtained from a single-objective optimization problem, where the initial construction cost was the objective to be minimized, while the second one as a two-objective optimization problem where the life-cycle cost was the additional objective also to be minimized.

Soft computing methodologies for structural optimization

The paper examines the efficiency of soft computing techniques in structural optimization, in particular algorithms based on evolution strategies combined with neural networks, for solving large-scale, continuous or discrete structural optimization problems. The proposed combined algorithms are implemented both in deterministic and reliability based structural optimization problems, in an effort to increase the computational efficiency as well as the robustness of the optimization procedure. The use of neural networks was motivated by the time-consuming repeated finite element analyses required during the optimization process. A trained neural network is used to perform either the deterministic constraints check or, in the case of reliability based optimization, both the deterministic and the probabilistic constraints checks. The suitability of the neural network predictions is investigated in a number of structural optimization problems in order to demonstrate the computational advantages of the proposed methodologies.

Optimization of Large-Scale 3-D Trusses Using Evolution Strategies and Neural Networks

The objective of this paper is to investigate the efficiency of optimization algorithms, based on evolution strategies, for the solution of large-scale structural optimization problems. Furthermore, the structural analysis phase is replaced by a neural network prediction for the computation of the necessary data for the evolution strategies (ES) optimization procedure. The use of neural networks (NN) was motivated by the time-consuming repeated analyses required by ES during the optimization process. A back propagation algorithm is implemented for training the NN using data derived from selected analyses. The trained NN is then used to predict, within an acceptable accuracy, the values of the objective and constraint functions. The proposed methodology has been applied in sizing structural optimization problems of large-scale three dimensional roof trusses. The numerical tests presented demonstrate the computational advantages of the proposed approach which become more pronounced for large-scale optimization problems.

MULTI-OBJECTIVE OPTIMIZATION OF SKELETAL STRUCTURES UNDER STATIC AND SEISMIC LOADING CONDITIONS

This chapter presents a evolution strategies approach for multiobjective design optimization of structural problems such as space frames and multi-layered space trusses under static and seismic loading conditions. A rigorous approach and a simplified one with respect to the loading condition are implemented for finding optimal design of a structure under multiple objectives.

Neural network based prediction schemes of the non-linear seismic response of 3D buildings

Since early 1980s new families of computational methods, termed as soft computing (SC) methods, have been proposed. SC methods are based on heuristic approaches rather than on rigorous mathematics while their use in various areas of computational mechanics is continuously growing. Artificial neural networks (ANNs), which have been applied in many engineering fields, are among the most popular SC methods. Computational earthquake engineering is a computationally intensive field where ANNs have been used for the simulation of the structural behaviour under static or dynamic loading. Performance-based design (PBD) is the current trend for the seismic design of structural systems where the structural performance is assessed for multiple hazard levels, requiring significant computational effort. In this work a new adaptive scheme is proposed in order to predict the structural non-linear behaviour when earthquake actions of increased severity are considered. The predicted structural response by ANNs can be used in the PBD framework when dynamic analyses are performed, aiming at reducing the excessive computational cost.

A critical assessment of metaheuristics for scheduling emergency infrastructure inspections

Abstract Search and rescue operations following natural disasters have become the cornerstone for minimizing the adverse social effects and the impact of these hazards. Despite their importance, the literature has not adequately dealt with post disaster operations–at least in part–because of the difficulty in rapidly solving the mathematically complex problems involved. In this paper we consider two important issues within the scope of post natural disaster actions; first, we develop deterministic and probabilistic districting and routing problems for scheduling infrastructure inspection crews following a natural disaster in urban areas; second, we assess and compare five metaheuristic optimization algorithms for solving these districting and routing problems. Results suggest that the five approaches examined offer applicable as well as fast solutions but with varying qualitative characteristics; selection of the preferred approach for practical applications will largely depend upon the network’s characteristics.