Search

Multi-layer neural networks have been used in this paper for modeling nonlinear behaviour of concrete gravity dams under earthquake excitation. Koyna dam which has been studied extensively by other authors in the past has been studied as test example in this paper too, where the nonlinear response of its crest has been modelled by the proposed algorithm. The main steps of the algorithm are as follows: First the concrete gravity dam has been numerically analyzed for its nonlinear behaviour under earthquake excitation to generate numerical data to be used in the training of the neural networks. To this end the dam has been subjected to a white noise excitation so that the generated data could be rich enough for the training of a general neuro-modeller of the dam response. The neuro-modeller has then been trained on the generated data to learn the hysteretic behaviour of the dam implicitly. Then the neural network has been tested on a number of earthquakes including near field as well as very strong earthquakes for verification. The results obtained in this study prove that the method has been successful regarding the generalization capabilities of the trained neuro-modeller where other earthquakes than those used in its training have been used in its testing. In the tests, the neuro-modeller could predict the response with high precision. One significant benefit of using this algorithm is in cases where it is desired to use collected data from tests on experimental models or through monitoring of the response of a dam to prepare a suitable model for predicting its response under any earthquake. Another benefit is the time of analysis which can be reduced by this method. Once the neuro-modeller is trained, it can predict the response of the dam to any earthquake without the need to be updated. [ABSTRACT FROM AUTHOR]

Springback is a very important factor to influence the quality of sheet metal forming. Accurate prediction and controlling of springback is essential for the design of tools for sheet metal forming. Several approaches have been proposed for springback compensation by modification of the tooling shape. These approaches are iterative finite element methods. In this paper an analytical approach is presented for one step modification of the tooling shape in channel forming process to compensate the springback error. With the help of this approach, the optimum die shape for producing the target shape can be obtained in a few seconds. The algorithm of springback compensation by inverse FE modelling is also presented. The results of the analytical approach coincide with those of FE approach. The accuracy of the obtained results is verified by the experimental results and high precision is achieved. [ABSTRACT FROM AUTHOR]

In this paper, we study the problem of computing tail conditional expectations (TCE) or portfolio gains at specific times (T) in the future. We present efficient algorithms to handle the following two cases: (1) we have only one option (call or put) in our portfolio, denoted as a single-stock-single-option portfolio (SSSO); (2) we have a stock and some of its options in our portfolio, denoted as a single-stock-multiple-option portfolio (SSMO). Compared with previous simulation algorithms, our algorithms compute TCE of a given portfolio more efficiently and still maintain the same degree of accuracy. In the SSSO case, we reduce the computational time complexity from the SSSO-Naïve Algorithm's O(m*n) to SSSO algorithm's O(m+n), where m is the number of possible price outcomes for an underlying stock at time T and n is the number of possible price outcomes for an underlying stock at time U (maturity) in respect to each possible price outcome for an underlying stock at time T. In the SSMO case, we provide two algorithms to compute the TCE. The computational time complexity of the SSMO-Naive Algorithm is O(q*m*n+m*log(m)), where q is the number of options. The computational time complexities of our two algorithms are O(q*(m+n)+m*log(m)), and O(q*log(q)+n+m*q+ m*q*log(q)+m*log(m)) or O(q*log(q)+n+m*q+m*q*log(m)+m* log(m)). In both cases, our experiments show that when m and n are greater than five thousand, our algorithms run thousands of times faster than the naive algorithms. [ABSTRACT FROM AUTHOR]

In this paper, we consider two finite element model updating problem which incorporate the measured modal data into the analytical finite element model, producing an adjusted model on the damping and stiffness, that closely match the experimental modal data. We develop an efficient numerical algorithm for solving this problem. The new algorithm is direct methods which require O(nk²) flops. Here n is the dimension of the coefficient matrices defining the analytical model and k is the number of measured eigenpairs. [ABSTRACT FROM AUTHOR]

In this paper an algorithm for minimization of a nondifferentiable function is presented. The algorithm uses the Moreau-Yoshida regularization of the objective function and its second order Dini upper directional derivative. It is proved that the algorithm is well defined, as well as the convergence of the sequence of points generated by the algorithm to an optimal point. An estimate of the rate of convergence is given, too. [ABSTRACT FROM AUTHOR]

This paper is concerned with adaptive numerical methods for initial value problems governed by systems of ordinary differential equations. Here we introduce a novel step selection algorithm based on the simple idea that locally all continuous functions can be suitably approximated by a straight line. Finally we present two sample numerical computations performed by our step selection algorithm. [ABSTRACT FROM AUTHOR]

Since the evaluation of each individual is based on the time-consuming structural analysis, the computational efficiency of truss topology optimization using genetic algorithm is very low. The paper focuses on this challenging problem. It is observed that there are a number of duplicate individuals appearing repetitively in the evolutionary process. Therefore, an individual identification technique is introduced to avoid evaluating the duplicate individuals by the time-consuming structural analysis but by searching the evolutionary history data to save computing time, the computational complexity of this technique is deduced. The results of two truss examples verify that the technique can effectively improve the efficiency of the algorithm. Based on this identification technique, numeric experiments are implemented to study the influence of several factors, i.e., the population size, the max generation, and the scale of problems, on the proportion of duplicate individuals. Results show that the population size has a significant impact on the proportion, and that both the max generation and the scale of problems have little influence. [ABSTRACT FROM AUTHOR]