人工神经网络BP算法的改进和结构的自调整

本文解决了BP神经网络结构参数和学习速率的选取问题,并对传统的BP算法进行了改进,提出了BP神经网络动态全参数自调整学习算法,又将其编制成计算机程序,使得隐层节点和学习速率的选取全部动态实现,减少了人为因素的干预,改善了学习速率和网络的适应能力。计算结果表明：BP神经网络动态全参数自调整算法较传统的方法优越。训练后的神经网络模型不仅能准确地拟合训练值,而且能较精确地预测未来趋势。     

基于PCA与偏好强度可调的DEA方法的多响应稳健参数设计

稳健参数设计旨在选择可控因子的水平组合来降低产品或过程对噪声因子的敏感性,能够有效地提高和改善产品的质量.针对多响应参数优化设计中响应间存在相关性的问题,提出了基于主成分分析(Principal Component Analysis,PCA)与偏好强度可调的数据包络分析(Data Envelopment Analysis,DEA)的稳健参数设计;并结合实例验证了方法的有效性.研究结果表明,结合PCA与偏好强度可调的DEA方法能够有效地实现相关多响应的参数优化设计.     

基于改进CS优化BP神经网络的降水量预测模型

针对传统BP神经网络易陷入局部极值和连接权值难以确定的问题,提出了一种基于融合PSO(Particle Swarm Optimization)和CS(Cuckoo Search)的混合算法优化设计BP神经网络(PCS-BP)的预测模型.该优化方法主要利用混合算法优秀的全局搜索能力和收敛速度设计优化BP神经网络的连接权值和网络结构,解决了BP神经网络由于参数随机取值引起的网络震荡和过拟合的问题,提高了预测模型的准确性.结合具体实例,分别采用BP神经网络、CS-BP模型和PCS-BP模型对汉中地区的月降水量进行预测,实验结果表明,PCS-BP的平均绝对误差(MAE)为0.3966,均方根误差(RMSE)为2.3793,平均绝对百分比误差(MAPE)为0.46%,均优于其他模型,具有较好的预测能力.     

基于灰关联分析的GA-BP神经网络在汽车油耗估算中的应用

从设计参数特征入手分析影响汽车油耗的因素,利用灰关联分析方法,解析了各设计参数对汽车油耗的影响程度,选择其中灰关联度较大的设计参数作为输入数据,综合工况油耗作为输出数据,构建6-5-1层结构的BP神经网络预测模型,并利用遗传算法获得优化后的BP神经网络的权值和阈值,然后训练BP神经网络得到最优值,最后以国内市场340款汽车作为研究样本,进行有效性验证.研究结果表明,模型利用灰关联分析获得影响汽车油耗的主要因素,简化了网络结构;与优化前的BP神经网络相比,具有更高的预测精度和可靠性.     

基于NFA-BP算法的企业经营状况评价模型

针对萤火虫算法(FA)易出现过早收敛,陷入局部最优的缺点,引入小生境技术,提出一种小生境萤火虫优化算法(NFA),通过测试后,利用其搜索BP神经网络的参数.最后建立基于小生境萤火虫优化BP算法的企业经营状况评价模型,并与传统的BP神经网络模型进行对比,仿真结果表明,基于NFA-BP算法的经营状况评价模型的正确识别率高于传统的BP模型,是一种有效的评价模型.     

融合鲸鱼算法的混合灰狼优化算法

针对灰狼优化(GWO)算法存在容易陷入局部最优、收敛速度慢、求解精度不高等问题,提出一种融合鲸鱼算法的混合灰狼优化(HWGWO)算法.首先在鲸鱼算法的螺旋泡网狩猎行为中融入Levy飞行并将其整体引入灰狼优化算法;然后将动态权重和差分进化思想引入灰狼优化算法;最后利用贪婪选择策略来保留较好的灰狼位置.选取23个测试函数进行数值试验,结果表明,HWGWO算法在收敛速度和求解精度上都有所提升.此外,利用HWGWO算法求解拉伸/压缩弹簧设计问题得到的设计方案更有效.     

田口方法和双重曲面响应(DRSM)法

田口参数设计在理论界遭到较多批评 ,也提出一些替代方法。双曲面响应 ( DRSM)法是结合田口参数设计的思想和统计回归模型发展起来的。本文通过正交旋转设计和计算机模拟噪声因子变化来提高双曲面响应法的精度并对两种方法应用结果进行比较 ,说明田口参数设计仍具有实践和方法论的启发意义     

逐步区间删失Weibull分布的Bayesian稳健设计

逐步区间删失是获取高可靠性产品相关信息的一种重要方法.文章研究了产品寿命服从Weibull分布,带有随机移除逐步区间删失寿命试验的最优设计问题.采用极大似然方法获取模型参数的估计及其信息矩阵.利用Bayesian方法处理模型参数未知情况下设计准则对模型参数的依赖问题,获得了模型参数估计的Bayesian稳健设计准则.在考虑试验费用有限制的条件下,给出了获得最优稳健设计非线性混合整数算法.同时对先验选取及约束参数设定的敏感性做了分析.数值结果表明文章提出的方法是有效可行的.     

基于非对称损失函数的参数设计

本文针对望目特性的正态指标,在非对称的二次质量损失函数下,讨论了参数设计的可行性, 证明田口方法的稳健性设计和灵敏度设计依然行之有效.定义了调整参数,求出了使质量损失最小的数值解,并给出了参数设计的具体步骤.     

混合优化RBF-BP网络的板形缺陷识别

针对传统板形模式识别方法存在精度低、鲁棒性弱的问题,提出了一种混合优化RBF-BP组合神经网络板形模式识别方法。首先利用自组织映射网络(SOM)对样本聚类,利用聚类后的网络拓扑结构确定RBF的中心,并计算RBF的宽度,克服了传统聚类算法随机选取中心导致聚类结果不稳定的问题。然后利用遗传算法(GA)良好的全局搜索能力优化整个网络的权值。RBF-BP组合神经网络是由一个RBF子网和一BP子网串联构成的,该网络同时具备BP神经网络能较好地预测未知样本的能力以及RBF神经网络的逼近速度快的优点。并以某900HC可逆冷轧机板形识别为应用背景,在MATLAB2010a环境下进行仿真实验,结果表明混合优化RBF-BP组合神经网络的板形模式识别方法能够识别出常见的板形缺陷,提高了板形缺陷识别精度并具有较好的鲁棒性,可以满足板带轧机高精度的板形控制要求。     

A new approach with orthogonal array for global optimization in design of experiments

In this paper, we develop an approach that determines the overall best parameter setting in design of experiments. The approach starts with successive orthogonal array experiments and ends with a full factorial experiment. The setup for the next orthogonal-array experiment is obtained from the previous one by either fixing a factor at a given level or by reducing the number of levels considered for all currently non-fixed factors. We illustrate this method using an industrial problem with seven parameters, each with three levels. In previous work, the full factorial of 3⁷ = 2,187 points was evaluated and the best point was found. With the new method, we found the same point using 3% of these evaluations. As a further comparison, we obtained the optimum using a traditional Taguchi approach, and found it corresponded to the 366th of the 2,187 possibilities when sorted by the objective function. We conclude the proposed approach would provide an accurate, fast, and economic tool for optimization using design of experiments.     

Applying neural network and scatter search to optimize parameter design with dynamic characteristics

Parameter design is critical to enhancing a system's robustness by identifying specific control factor set points (levels) that make the system least sensitive to noise. Engineers have conventionally applied Taguchi methods to optimize parameter design. However, Taguchi methods can only obtain the optimal solution among the specified control factor levels. They cannot identify the real optimum when the parameter values are continuous. This study proposes a hybrid procedure combining neural networks and scatter search to optimize the continuous parameter design problem. First, neural networks are used to simulate the relationship between the control factor values and corresponding responses. Second, scatter search is employed to obtain the optimal parameter settings. The desirability function is utilized to transform the multiple responses into a single response. A case with dynamic characteristics is carried out in blood glucose strip manufacturing in Taiwan to demonstrate the practicability of the proposed procedure.     

A new multi-objective discrete robust optimization algorithm for engineering design

This paper proposes a novel multi-objective discrete robust optimization (MODRO) algorithm for design of engineering structures involving uncertainties. In the present MODRO procedure, grey relational analysis (GRA), coupled with principal component analysis (PCA), was used as a multicriteria decision making model for converting multiple conflicting objectives into one unified cost function. The optimization process was iterated using the successive Taguchi approach to avoid the limitation that the conventional Taguchi method fails to deal with a large number of design variables and design levels. The proposed method was first verified by a mathematical benchmark example and a ten-bar truss design problem; and then it was applied to a more sophisticated design case of full scale vehicle structure for crashworthiness criteria. The results showed that the algorithm is able to achieve an optimal design in a fairly efficient manner attributable to its integration with the multicriteria decision making model. Note that the optimal design can be directly used in practical applications without further design selection. In addition, it was found that the optimum is close to the corresponding Pareto frontier generated from the other approaches, such as the non-dominated sorting genetic algorithm II (NSGA-II), but can be more robust as a result of introduction of the Taguchi method. Due to its independence on metamodeling techniques, the proposed algorithm could be fairly promising for engineering design problems of high dimensionality.     

An efficient algorithm for optimal design of area traffic control with network flows

An equilibrium network design (EQND) is a problem of finding the optimal design parameters while taking into account the route choice of users. This problem can be formulated as an optimization by taking the user equilibrium traffic assignment as a constraint. In this paper, the methods solving the EQND problem with signal settings are investigated via numerical calculations on two example road networks. An efficient algorithm is proposed in which improvement on a locally optimal search by combining the technique of parallel tangents with the gradient projection method is presented. As it shows, the method combines the locally optimal search and globally search heuristic achieved substantially better performance than did those other approaches.     

Optimization of process parameters using weighted convex loss functions

Setting the optimal process parameters under certain criterion is among the most important factor to determine a product’s profit. In this study, we generalize Taguchi quality model to more general process scenario, which consider multiple input and output quality variables simultaneously. Using the weighted convex loss function, we discuss the optimization of process parameters for Taguchi quality model and a trade-off problem as well. The working requirements for optimization are presented and the examples to illustrate our result are demonstrated.     

A Comparison of Global Optimization Methods for the Design of a High-speed Civil Transport

The conceptual design of aircraft often entails a large number of nonlinear constraints that result in a nonconvex feasible design space and multiple local optima. The design of the high-speed civil transport (HSCT) is used as an example of a highly complex conceptual design with 26 design variables and 68 constraints. This paper compares three global optimization techniques on the HSCT problem and two test problems containing thousands of local optima and noise: multistart local optimizations using either sequential quadratic programming (SQP) as implemented in the design optimization tools (DOT) program or Snyman's dynamic search method, and a modified form of Jones' DIRECT global optimization algorithm. SQP is a local optimizer, while Snyman's algorithm is capable of moving through shallow local minima. The modified DIRECT algorithm is a global search method based on Lipschitzian optimization that locates small promising regions of design space and then uses a local optimizer to converge to the optimum. DOT and the dynamic search algorithms proved to be superior for finding a single optimum masked by noise of trigonometric form. The modified DIRECT algorithm was found to be better for locating the global optimum of functions with many widely separated true local optima.     

An SR1/BFGS SQP algorithm for nonconvex nonlinear programs with block-diagonal Hessian matrix

We present a quasi-Newton sequential quadratic programming (SQP) algorithm for nonlinear programs in which the Hessian of the Lagrangian function is block-diagonal. Problems with this characteristic frequently arise in the context of optimal control; for example, when a direct multiple shooting parametrization is used. In this article, we describe an implementation of a filter line-search SQP method that computes search directions using an active-set quadratic programming (QP) solver. To take advantage of the block-diagonal structure of the Hessian matrix, each block is approximated separately by quasi-Newton updates. For nonconvex instances, that arise, for example, in optimum experimental design control problems, these blocks are often found to be indefinite. In that case, the block-BFGS quasi-Newton update can lead to poor convergence. The novel aspect in this work is the use of SR1 updates in place of BFGS approximations whenever possible. The resulting indefinite QPs necessitate an inertia control mechanism within the sparse Schur-complement factorization that is carried out by the active-set QP solver. This permits an adaptive selection of the Hessian approximation that guarantees sufficient progress towards a stationary point of the problem. Numerical results demonstrate that the proposed approach reduces the number of SQP iterations and CPU time required for the solution of a set of optimal control problems.     

D-optimal design of a monitoring network for parameter estimation of distributed systems

This paper addresses the design of a network of observation locations in a spatial domain that will be used to estimate unknown parameters of a distributed parameter system. We consider a setting where we are given a finite number of possible sites at which to locate a sensor, but cost constraints allow only some proper subset of them to be selected. We formulate this problem as the selection of the gauged sites so as to maximize the log-determinant of the Fisher information matrix associated with the estimated parameters. The search for the optimal solution is performed using the branch-and-bound method in which an extremely simple and efficient technique is employed to produce an upper bound to the maximum objective function. Its idea consists in solving a relaxed problem through the application of a simplicial decomposition algorithm in which the restricted master problem is solved using a multiplicative algorithm for optimal design. The use of the proposed approach is illustrated by a numerical example involving sensor selection for a two-dimensional convective diffusion process.     

压缩感知中信号重构的极大熵方法(英文)

本文针对压缩感知理论中BP算法的l1最优化问题,构造了一种新的信号重构的极大熵方法.极大熵方法克服了l1最优化问题的非光滑性,同时根据同伦方法构造极大熵函数的最优解序列来逼近全局最优稀疏解.数值实验表明极大熵方法是十分有效的信号重构方法.     

Optimization of the injection molding process for short-fiber-reinforced composites

A fast and effective methodology integrating the finite-element and Taguchi methods is presented to determine the optimal design conditions of the injection molding process for short-fiber-reinforced polycarbonate composites. The finite-element-based flow simulation software, M-flow, was employed to simulate the molding process to obtain the fiber orientation distributions required. The Taguchi optimization technique was used to identify the optimal settings of injection molding parameters to maximize the shear layer thickness. The effects of four main parameters — the filling time, melt temperature, mold temperature, and injection speed — on the fiber orientation or the shear layer thickness were investigated and discussed. It is found that the dominant parameter is the filling time. The best levels of the four parameters to acquire the thickest shear layer are also identified.