A surrogate based multistage-multilevel optimization procedure for multidisciplinary design optimization

Optimization procedure is one of the key techniques to address the computational and organizational complexities of multidisciplinary design optimization (MDO). Motivated by the idea of synthetically exploiting the advantage of multiple existing optimization procedures and meanwhile complying with the general process of satellite system design optimization in conceptual design phase, a multistage-multilevel MDO procedure is proposed in this paper by integrating multiple-discipline-feasible (MDF) and concurrent subspace optimization (CSSO), termed as MDF-CSSO. In the first stage, the approximation surrogates of high-fidelity disciplinary models are built by disciplinary specialists independently, based on which the single level optimization procedure MDF is used to quickly identify the promising region and roughly locate the optimum of the MDO problem. In the second stage, the disciplinary specialists are employed to further investigate and improve the baseline design obtained in the first stage with high-fidelity disciplinary models. CSSO is used to organize the concurrent disciplinary optimization and system coordination so as to allow disciplinary autonomy. To enhance the reliability and robustness of the design under uncertainties, the probabilistic version of MDF-CSSO (PMDF-CSSO) is developed to solve uncertainty-based optimization problems. The effectiveness of the proposed methods is verified with one MDO benchmark test and one practical satellite conceptual design optimization problem, followed by conclusion remarks and future research prospects.     

An efficient class of direct search surrogate methods for solving expensive optimization problems with CPU-time-related functions

In this paper, we characterize a new class of computationally expensive optimization problems and introduce an approach for solving them. In this class of problems, objective function values may be directly related to the computational time required to obtain them, so that, as the optimal solution is approached, the computational time required to evaluate the objective is significantly less than at points farther away from the solution. This is motivated by an application in which each objective function evaluation requires both a numerical fluid dynamics simulation and an image registration process, and the goal is to find the parameter values of a predetermined reference image by comparing the flow dynamics from the numerical simulation and the reference image through the image comparison process. In designing an approach to numerically solve the more general class of problems in an efficient way, we make use of surrogates based on CPU times of previously evaluated points, rather than their function values, all within the search step framework of mesh adaptive direct search algorithms. Because of the expected positive correlation between function values and their CPU times, a time cutoff parameter is added to the objective function evaluation to allow its termination during the comparison process if the computational time exceeds a specified threshold. The approach was tested using the NOMADm and DACE MATLAB? software packages, and results are presented.     

A survey of multidisciplinary design optimization methods in launch vehicle design

Optimal design of launch vehicles is a complex problem which requires the use of specific techniques called Multidisciplinary Design Optimization (MDO) methods. MDO methodologies are applied in various domains and are an interesting strategy to solve such an optimization problem. This paper surveys the different MDO methods and their applications to launch vehicle design. The paper is focused on the analysis of the launch vehicle design problem and brings out the advantages and the drawbacks of the main MDO methods in this specific problem. Some characteristics such as the robustness, the calculation costs, the flexibility, the convergence speed or the implementation difficulty are considered in order to determine the methods which are the most appropriate in the launch vehicle design framework. From this analysis, several ways of improvement of the MDO methods are proposed to take into account the specificities of the launch vehicle design problem in order to improve the efficiency of the optimization process.     

求解昂贵区间多目标优化问题的高斯代理模型

本文将数据挖掘(高斯过程回归建模)和智能进化算法(GA,NSGA-Ⅱ)进行结合,用于解决优化函数未知的昂贵区间多目标优化问题.首先利用高斯过程对采用中点和不确定度表示的未知目标函数和约束函数进行建模,由于相关性和准确性是区间函数模型的两个必备条件,故提出一种融合多属性决策的双层种群筛选策略,并将其嵌入到遗传算法求解高斯模型参数的过程中,第1层根据相关性属性排除候选解集中部分劣解,第2层根据准确性属性排除候选解集中其余超出种群规模的劣解,两属性的权重系数决定两层排除劣解的比例.然后将所建模型作为优化对象的代理模型引导区间NSGA-II算法优化求解,从而获得所需的Pareto前沿.     

SO-MI: A surrogate model algorithm for computationally expensive nonlinear mixed-integer black-box global optimization problems

This paper introduces a surrogate model based algorithm for computationally expensive mixed-integer black-box global optimization problems with both binary and non-binary integer variables that may have computationally expensive constraints. The goal is to find accurate solutions with relatively few function evaluations. A radial basis function surrogate model (response surface) is used to select candidates for integer and continuous decision variable points at which the computationally expensive objective and constraint functions are to be evaluated. In every iteration multiple new points are selected based on different methods, and the function evaluations are done in parallel. The algorithm converges to the global optimum almost surely. The performance of this new algorithm, SO-MI, is compared to a branch and bound algorithm for nonlinear problems, a genetic algorithm, and the NOMAD (Nonsmooth Optimization by Mesh Adaptive Direct Search) algorithm for mixed-integer problems on 16 test problems from the literature (constrained, unconstrained, unimodal and multimodal problems), as well as on two application problems arising from structural optimization, and three application problems from optimal reliability design. The numerical experiments show that SO-MI reaches significantly better results than the other algorithms when the number of function evaluations is very restricted (200–300 evaluations).     

Development of a multidisciplinary design optimization test simulator

Many of the method development efforts in the field of multidisciplinary design optimization (MDO) attempt to simplify the design of a large, complex system by dividing the system into a series of smaller, simpler, and coupled subsystems. A representative and efficient means of determining the feasibility and robustness of MDO methods is crucial. This paper describes the construct and applications of a test simulator, CASCADE (Complex Application Simulator for the Creation of Analytical Design Equations), that is capable of randomly generating and then converging a system of coupled analytical equations, of user-specified size (Hulme and bloebaum 1996). CASCADE-generated systems can be used for test sequencing and system reduction strategies, convergence strategies, optimization techniques, MDO methods, and distributed computing techniques (via Parallel Virtual Machine), among others.     

Global kriging surrogate modeling for general time-variant reliability-based design optimization problems

While design optimization under uncertainty has been widely studied in the last decades, time-variant reliability-based design optimization (t-RBDO) is still an ongoing research field. The sequential and mono-level approaches show a high numerical efficiency. However, this might be to the detriment of accuracy especially in case of nonlinear performance functions and non-unique time-variant most probable failure point (MPP). A better accuracy can be obtained with the coupled approach, but this is in general computationally prohibitive. This work proposes a new t-RBDO method that overcomes the aforementioned limitations. The main idea consists in performing the time-variant reliability analysis on global kriging models that approximate the time-dependent limit state functions. These surrogates are built in an artificial augmented reliability space and an efficient adaptive enrichment strategy is developed that allows calibrating the models simultaneously. The kriging models are consequently only refined in regions that may potentially be visited by the optimizer. It is also proposed to use the same surrogates to find the deterministic design point with no extra computational cost. Using this point to launch the t-RBDO guarantees a fast convergence of the optimization algorithm. The proposed method is demonstrated on problems involving nonlinear limit state functions and non-stationary stochastic processes.     

Gaussian process as complement to test functions for surrogate modeling

It is common for papers on surrogate fitting to select test functions for testing algorithms. This raises the issue of how well the algorithms generalize t     

Sequential approximation optimization assisted particle swarm optimization for expensive problems

In this paper, an efficient sequential approximation optimization assisted particle swarm optimization algorithm is proposed for optimization of expensive problems. This algorithm makes a good balance between the search ability of particle swarm optimization and sequential approximation optimization. Specifically, the proposed algorithm uses the optima obtained by sequential approximation optimization in local regions to replace the personal historical best particles and then runs the basic particle swarm optimization procedures. Compared with particle swarm optimization, the proposed algorithm is more efficient because the optima provided by sequential approximation optimization can direct swarm particles to search in a more accurate way. In addition, a space partition strategy is proposed to constraint sequential approximation optimization in local regions. This strategy can enhance the swarm diversity and prevent the preconvergence of the proposed algorithm. In order to validate the proposed algorithm, a lot of numerical benchmark problems are tested. An overall comparison between the proposed algorithm and several other optimization algorithms has been made. Finally, the proposed algorithm is applied to an optimal design of bearings in an all-direction propeller. The results show that the proposed algorithm is efficient and promising for optimization of the expensive problems.     

An algorithm for computationally expensive engineering optimization problems

Abstract

Modern engineering design often relies on computer simulations to evaluate candidate designs, a scenario which results in an optimization of a computationally expensive black-box function. In these settings, there will often exist candidate designs which cause the simulation to fail, and can therefore degrade the search effectiveness. To address this issue, this paper proposes a new metamodel-assisted computational intelligence optimization algorithm which incorporates classifiers into the optimization search. The classifiers predict which candidate designs are expected to cause the simulation to fail, and this prediction is used to bias the search towards designs predicted to be valid. To enhance the search effectiveness, the proposed algorithm uses an ensemble approach which concurrently employs several metamodels and classifiers. A rigorous performance analysis based on a set of simulation-driven design optimization problems shows the effectiveness of the proposed algorithm.     

Multiparametric optimization for multidisciplinary engineering design

The area of Multiparametric Optimization (MPO) solves problems that contain unknown problem data represented by parameters. The solutions map parameter values to optimal design and objective function values. In this paper, for the first time, MPO techniques are applied to improve and advance Multidisciplinary Design Optimization (MDO) to solve engineering problems with parameters. A multiparametric subgradient algorithm is proposed and applied to two MDO methods: Analytical Target Cascading (ATC) and Network Target Coordination (NTC). Numerical results on test problems show the proposed parametric ATC and NTC methods effectively solve parametric MDO problems and provide useful insights to designers. In addition, a novel Two-Stage ATC method is proposed to solve nonparametric MDO problems. In this new approach elements of the subproblems are treated as parameters and optimal design functions are constructed for each one. When the ATC loop is engaged, steps involving the lengthy optimization of subproblems are replaced with simple function evaluations.     

Multi-objective optimization for design under uncertainty problems through surrogate modeling in augmented input space

Multi-objective design under uncertainty problems that adopt probabilistic quantities as performance objectives and consider their estimation through stochastic simulation are examined in this paper, focusing on development of a surrogate modeling framework to reduce computational burden for the numerical optimization. The surrogate model is formulated to approximate the system response with respect to both the design variables and the uncertain model parameters, so that it can simultaneously support both the uncertainty propagation and the identification of the Pareto optimal solutions. Kriging is chosen as the metamodel, and its probabilistic nature (its ability to offer a local estimate of the prediction error) is leveraged within different aspects of the framework. To reduce the number of simulations for the expensive system model, an iterative approach is established with adaptive characteristics for controlling the metamodel accuracy. At each iteration, a new metamodel is developed utilizing all available training points. A new Pareto front is then identified utilizing this surrogate model and is compared, for assessing stopping criteria, to the front that was identified in the previous iteration. This comparison utilizes explicitly the potential error associated with the metamodel predictions. If stopping criteria are not achieved, a set of refinement experiments (new training points) is identified and process proceeds to the next iteration. A hybrid design of experiments is considered for this refinement, with a dual goal of global coverage and local exploitation of regions of interest, separately identified for the design variables and the uncertain model parameters.

     

A computational intelligence algorithm for expensive engineering optimization problems

The modern engineering design optimization process often replaces laboratory experiments with computer simulations, which leads to expensive black-box optimization problems. Such problems often contain candidate solutions which cause the simulation to fail, and therefore they will have no objective value assigned to them, a scenario which degrades the search effectiveness. To address this, this paper proposes a new computational intelligence optimization algorithm which incorporates a classifier into the optimization search. The classifier predicts which solutions are expected to cause a simulation failure, and its prediction is used to bias the search towards solutions for which the simulation is expected to succeed. To further enhance the search effectiveness, the proposed algorithm continuously adapts during the search the type of model and classifier being used. A rigorous performance analysis using a representative application of airfoil shape optimization shows that the proposed algorithm outperformed existing approaches in terms of the final result obtained, and performed a search with a competitively low number of failed evaluations. Analysis also highlights the contribution of incorporating the classifier into the search, and of the model and classifier selection steps.     

Component-oriented decomposition for multidisciplinary design optimization in building design

The potential of Multidisciplinary Design Optimization (MDO) is not sufficiently exploited in current building design practice. I argue that this field of engineering requires a special setup of the optimization model that considers the uniqueness of buildings, and allows the designer to interact with the optimization in order to assess qualities of aesthetics, expression, and building function. For this reason, the approach applies a performance optimization based on resource consumption extended by preference criteria. Furthermore, building design-specific components serve for the decomposition and an interactive way of working. The component scheme follows the Industry Foundation Classes (IFC) as a common Building Information Model (BIM) standard in order to allow a seamless integration into an interactive CAD working process in the future. A representative case study dealing with a frame-based hall design serves to illustrate these considerations. An N-Square diagram or Design Structure Matrix (DSM) represents the system of components with the disciplinary dependencies and workflow of the analysis. The application of a Multiobjective Genetic Algorithm (MOGA) leads to demonstrable results.     

Memetic algorithm using multi-surrogates for computationally expensive optimization problems

In this paper, we present a multi-surrogates assisted memetic algorithm for solving optimization problems with computationally expensive fitness functions. The essential backbone of our framework is an evolutionary algorithm coupled with a local search solver that employs multi-surrogate in the spirit of Lamarckian learning. Inspired by the notion of ‘blessing and curse of uncertainty’ in approximation models, we combine regression and exact interpolating surrogate models in the evolutionary search. Empirical results are presented for a series of commonly used benchmark problems to demonstrate that the proposed framework converges to good solution quality more efficiently than the standard genetic algorithm, memetic algorithm and surrogate-assisted memetic algorithms.     

A rigorous framework for optimization of expensive functions by surrogates

The goal of the research reported here is to develop rigorous optimization algorithms to apply to some engineering design problems for which direct application of traditional optimization approaches is not practical. This paper presents and analyzes a framework for generating a sequence of approximations to the objective function and managing the use of these approximations as surrogates for optimization. The result is to obtain convergence to a minimizer of an expensive objective function subject to simple constraints. The approach is widely applicable because it does not require, or even explicitly approximate, derivatives of the objective. Numerical results are presented for a 31-variable helicopter rotor blade design example and for a standard optimization test example.     

A two-stage adaptive multi-fidelity surrogate model-assisted multi-objective genetic algorithm for computationally expensive problems

Surrogate model-assisted multi-objective genetic algorithms (MOGA) show great potential in solving engineering design problems since they can save computational cost by reducing the calls of expensive simulations. In this paper, a two-stage adaptive multi-fidelity surrogate (MFS) model-assisted MOGA (AMFS-MOGA) is developed to further relieve their computational burden. In the warm-up stage, a preliminary Pareto frontier is obtained relying only on the data from the low-fidelity (LF) model. In the second stage, an initial MFS model is constructed based on the data from both LF and high-fidelity (HF) models at the samples, which are selected from the preliminary Pareto set according to the crowding distance in the objective space. Then the fitness values of individuals are evaluated using the MFS model, which is adaptively updated according to two developed strategies, an individual-based updating strategy and a generation-based updating strategy. The former considers the prediction uncertainty from the MFS model, while the latter takes the discrete degree of the population into consideration. The effectiveness and merits of the proposed AMFS-MOGA approach are illustrated using three benchmark tests and the design optimization of a stiffened cylindrical shell. The comparisons between the proposed AMFS-MOGA approach and some existing approaches considering the quality of the obtained Pareto frontiers and computational efficiency are made. The results show that the proposed AMFS-MOGA method can obtain Pareto frontiers comparable to that obtained by the MOGA with HF model, while significantly reducing the number of evaluations of the expensive HF model.

     

Hybrid meta-model-based design space exploration method for expensive problems

Structural and Multidisciplinary Optimization - Multiple meta-models used together in the search process at least can offer an insurance against the poorly fitted meta-models and can improve...