Search

Abstract: Part 1 of this paper proposed a methodology for the efficient determination of gradient information, when optimising a vehicle’s suspension characteristics for ride comfort and handling. The non-linear full vehicle model, and simplified models for gradient information has been discussed, and validated. In this paper, the simplified models presented in Part 1 are used for gradient information simulations. The convergence histories of the optimisation are compared to those obtained when only the full, computationally expensive, vehicle model is used. For illustration of the proposed gradient-based optimisation methodology, up to four design variables are considered in modelling the suspension characteristics. The proposed methodology is found to be an efficient alternative for the optimisation of the vehicle’s suspension characteristics. The undesirable effects associated with noise in the gradient information is effectively reduced, using the simplified models. Substantial benefits are achieved in terms of computational time needed to reach a solution. [Copyright &y& Elsevier]

In probabilistic model analysis and optimization, expected values of a model output f ( x ) in face of continuous random inputs x are estimated through n -dimensional integrals, where n = d i m ( x ) . Cubature formulae are approximations of these integrals by a weighted sum of function evaluations at carefully chosen points. When each function evaluation corresponds to a heavy computational simulation, and particularly in optimization problems, one needs very efficient formulae with few integration points, even though only having modest accuracy. In this paper, we evaluate the performance of several cubature formulae with few points, including Smolyak type formulae, also known as sparse grid integration, and recently proposed thinned cubatures, constructed using orthogonal arrays. Tests are made for a wide family of smooth and non-oscillatory functions f ( x ) , possibly with significant anisotropy, and covering both normal and uniform input probability distributions. Two practical case studies are also presented, one of analysis of a large scale mass transfer model with uncertain parameters and a second one of optimal production planning under uncertain market conditions. Results clearly indicate that cubatures with large negative weights, including Smolyak type formulae, are not reliable, contrary to positive thinned cubatures that produce very reasonable estimates up to dimension 24. These thinned cubatures may also surpass quasi-Monte Carlo methods also up to dimension 24. [ABSTRACT FROM AUTHOR]

Abstract: Metamodelling is an increasingly more popular approach for alleviating the computational burden associated with computationally intensive optimization/management problems in environmental and water resources systems. Some studies refer to the metamodelling approach as function approximation, surrogate modelling, response surface methodology or model emulation. A metamodel-enabled optimizer approximates the objective (or constraint) function in a way that eliminates the need to always evaluate this function via a computationally expensive simulation model. There is a sizeable body of literature developing and applying a variety of metamodelling strategies to various environmental and water resources related problems including environmental model calibration, water resources systems analysis and management, and water distribution network design and optimization. Overall, this literature generally implies metamodelling yields enhanced solution efficiency and (almost always) effectiveness of computationally intensive optimization problems. This paper initially develops a comparative assessment framework which presents a clear computational budget dependent definition for the success/failure of the metamodelling strategies, and then critically evaluates metamodelling strategies, through numerical experiments, against other common optimization strategies not involving metamodels. Three different metamodel-enabled optimizers involving radial basis functions, kriging, and neural networks are employed. A robust numerical assessment within different computational budget availability scenarios is conducted over four test functions commonly used in optimization as well as two real-world computationally intensive optimization problems in environmental and water resources systems. Numerical results show that metamodelling is not always an efficient and reliable approach to optimizing computationally intensive problems. For simpler response surfaces, metamodelling can be very efficient and effective. However, in some cases, and in particular for complex response surfaces when computational budget is not very limited, metamodelling can be misleading and a hindrance, and better solutions are achieved with optimizers not involving metamodels. Results also demonstrate that neural networks are not appropriate metamodelling tools for limited computational budgets while metamodels employing kriging and radial basis functions show comparable overall performance when the available computational budget is very limited. [Copyright &y& Elsevier]

Abstract: This work develops a novel two-stage fuzzy optimization method for solving the multi-product multi-period (MPMP) production planning problem, in which the market demands and some of the inventory costs are assumed to be uncertainty and characterized by fuzzy variables with known possibility distributions. Some basic properties about the MPMP production planning problem are discussed. Since the fuzzy market demands and inventory costs usually have infinite supports, the proposed two-stage fuzzy MPMP production planning problem is an infinite-dimensional optimization problem that cannot be solved directly by conventional numerical solution methods. To overcome this difficulty, this paper adopts an approximation method (AM) to turn the original two-stage fuzzy MPMP production planning problem into a finite-dimensional optimization problem. The convergence about the AM is discussed to ensure the solution quality. After that, we design a heuristic algorithm, which combines the AM and simulated annealing (SA) algorithm, to solve the proposed two-stage fuzzy MPMP production planning problem. Finally, one real case study about a furniture manufacturing company is presented to illustrate the effectiveness and feasibility of the proposed modeling idea and designed algorithm. [Copyright &y& Elsevier]

Abstract: In this paper, a promising heat engine technology capable of utilizing low grade heat is examined. Based on a two-phase thermofluidic oscillator concept, the novelty and advantage of this particular system lie in its use of phase change and its lack of reliance on inertia to sustain oscillations, though it is recognized that inertia will always be present in any physical manifestation of the engine. The system is analysed using lumped linearized one-dimensional network models, both with and without inertia, based on thermoacoustic principles and extending these to account for phase change. The gain (temperature difference between source and sink heat exchangers) and frequency at which marginal stability (desirable continuous oscillations) can be achieved is calculated. The effects of the load resistance (fluid drag) and fluid inertia, as well as of the flow resistance due the feedback valve on the marginal stability gain, frequency and exergetic efficiency of the system are investigated. It is found that an increase in feedback resistance leads to a need for a higher gain for oscillatory behaviour to be achieved. In addition, even though an increase in either the resistance or inertia in the load, or the feedback resistance at low values of these variables has almost no effect on the required gain and the oscillation frequency of the system, an increase in these variables can lead at higher values to increased gains and reduced frequencies. A reduced feedback resistance and greater load inertia can also lead to considerably higher efficiencies, while increasing the load resistance allows for an increase in efficiency until a maximum is reached, after which the efficiency decreases again. The validity of certain approximations made previously is considered, and it is shown that these must be made with care. The results from this study can be used for the improved design and optimization of such oscillators, and similar systems. [Copyright &y& Elsevier]

Abstract: Expensive optimization aims to find the global minimum of a given function within a very limited number of function evaluations. It has drawn much attention in recent years. The present expensive optimization algorithms focus their attention on metamodeling techniques, and call existing global optimization algorithms as subroutines. So it is difficult for them to keep a good balance between model approximation and global search due to their two-part property. To overcome this difficulty, we try to embed a metamodel mechanism into an efficient evolutionary algorithm, low dimensional simplex evolution (LDSE), in this paper. The proposed algorithm is referred to as the low dimensional simplex evolution extension (LDSEE). It is inherently parallel and self-contained. This renders it very easy to use. Numerical results show that our proposed algorithm is a competitive alternative for expensive optimization problems. [Copyright &y& Elsevier]

Abstract: Coupled optimization methods based on multi-objective genetic algorithms and approximation models are widely used in engineering optimizations. In the present paper, a similar framework is proposed for the aerodynamic optimization of turbomachinery by coupling the well known multi-objective genetic algorithm—NSGA-II and back propagation neural network. The verification results of mathematical problems show that the coupled method with the origin NSGA-II cannot get the real Pareto front due to the prediction error of BPNN. A modified crowding distance is proposed in cooperation with a coarse-to-fine approaching strategy based on the iterations between NSGA-II and BPNN. The results of mathematical model problems show the effect of these improving strategies. An industrial application case is implemented on a transonic axial compressor. The optimization objectives are to maximize efficiencies of two working points and to minimize the variation of the choked mass flow. CFD simulation is employed to provide the performance evaluation of initial training samples for BPNN. The optimized results are compared with optimization results of a single objective optimization based on weighting function. The comparison shows that the present framework can provide not only better solutions than the single objective optimization, but also various alternative solutions. The increase of computational costs is acceptable especially when approximation models are used. [Copyright &y& Elsevier]

Abstract: A methodology is proposed for the efficient determination of gradient information, when performing gradient based optimisation of an off-road vehicle’s suspension system. The methodology is applied to a computationally expensive, non-linear vehicle model, that exhibits severe numerical noise. A recreational off-road vehicle is modelled in MSC.ADAMS, and coupled to MATLAB for the execution of the optimisation. The successive approximation method, Dynamic-Q, is used for the optimisation of the spring and damper characteristics. Optimisation is performed for both ride comfort and handling. The determination of the objective function value is performed using computationally expensive numerical simulations. This paper proposes a non-linear pitch-plane model, to be used for the gradient information, when optimising ride comfort. When optimising for handling, a non-linear four wheel model, that includes roll, is used. The gradients of the objective function and constraint functions are obtained through the use of central finite differences, within Dynamic-Q, via numerical simulation using the proposed simplified models. The importance of correctly scaling these simplified models is emphasised. The models are validated against experimental results. The simplified vehicle models exhibit significantly less numerical noise than the full vehicle simulation model, and solve in significantly less computational time. [Copyright &y& Elsevier]

Abstract: In this paper, we analyse the computational complexity of an optimization version of the Simplified Partial Digest Problem (SPDP), which is a mathematical model for DNA mapping based on the results of a simplified partial digest experiment. We prove that recognizing 46.16% of the elements of the DNA map in the error-free simplified partial digest experiment is NP-hard in the strong sense. This implies that the problem of maximizing the number of correct elements of the DNA map in the error-free simplified partial digest experiment is pseudopolynomially non-approximable with the approximation ratio . [Copyright &y& Elsevier]