A Sequential Approximation Method Using Neural Networks for Nonlinear Discrete-Variable Optimization with Implicit Constraints.

Hsu, Yeh‐Liang; Dong, Yu-Hsin; Hsu, Ming-Sho

doi:10.1299/jsmec.44.103

Cited by 11 publications

(7 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In this paper, the 'sequential neural network approximation (SNA) method' [8,9] is used to solve the problem. In this method, first a back-propagation neural network is trained to simulate the feasible domain formed by the implicit constraints using a few representative training data.…”

Section: The Sequential Neural Network Approximation Methodsmentioning

confidence: 99%

One-pass milling machining parameter optimization to achieve mirror surface roughness

Wang

Hsu

2005

Proceedings of the Institution of Mechanical Engineers, Part B:

View full text Add to dashboard Cite

In this paper, the possibility of just using general machine centres in one-pass milling process to finish an aluminium plate with the mirror surface roughness is studied. In particular, how to find the optimal setting of machining parameters is presented. In this optimization problem, to evaluate whether the surface roughness meets the average criterion of a mirror surface requires real cutting experiments, and it is desirable to find the optimal machining parameters using as few experiments as possible. The 'sequential neural network approximation method' was used to find the optimal machining parameters, including the spindle speed, feed rate, depth of cut, and number of inserted blades in the cutter to maximize the metal removal rate while the surface roughness meets the average criterion of a mirror surface.

show abstract

Section: The Sequential Neural Network Approximation Methodsmentioning

confidence: 99%

One-pass milling machining parameter optimization to achieve mirror surface roughness

Wang

Hsu

2005

Proceedings of the Institution of Mechanical Engineers, Part B:

View full text Add to dashboard Cite

show abstract

“…The design variables are radius R = x 1 , length L = x 2 , thickness Ts = x 3 , and thickness Th = x 4 . The objective is to minimize the cost given as Table 2 Comparison of the results Sandgren [4] Qian [15] Kannan [10] Hsu [16] He [18] This paper Through 20 trials, the objective function has attained at 6060 6066 f ≤ ≤…”

Section: Discrete Design Variables Problem 2 (Feasible Domain Is Sepamentioning

confidence: 99%

Global Optimization by Generalized Random Tunneling Algorithm (4th Report Application to the Nonlinear Optimum Design Problem of the Mixed Design Variables)

Kitayama¹,

Yamazaki²

2008

JCST

View full text Add to dashboard Cite

This paper presents a method to obtain the global or quasi-optimum for the discrete and continuous design variables, based on the Modified Generalized Random Tunneling Algorithm (MGRTA). By handling the discrete design variables as penalty function, the augmented objective function is constructed. As a result, all design variables can be treated as the continuous design variables. The augmented objective function becomes non-convex, and has many local minima. That is, finding optimum of discrete design variables is transformed into finding global optimum of this augmented objective function. Then the MGRTA is applied to this augmented objective function, subject to the behavior and side constraints. We also propose the new update scheme of penalty parameter for the penalty function of discrete design variables in this paper. The proposed update scheme of penalty parameter utilizes the information of the penalty function value of discrete design variables. By utilizing the characteristics of MGRTA, some optima are obtained. The validity of the proposed method is examined through typical benchmark problems.

show abstract

“…(16) should be updated so as to satisfy Eq. (20). In order to investigate the influence of the penalty function, let us consider the following simple problem with one discrete variable.…”

Section: Properties Of Penalty Function For Discrete Variablesmentioning

confidence: 99%

A method for mixed integer programming problems by particle swarm optimization

Kitayama

Yasuda

2006

Electrical Engineering Japan

View full text Add to dashboard Cite

SUMMARYParticle Swarm Optimization (PSO) for mixed integer programming problems is proposed. PSO is mainly a method to find a global or quasi-minimum for a nonlinear and nonconvex optimization problem, and there have been few studies into optimization problems with discrete decision variables. In this paper, we present the treatment of discrete variables. To treat discrete decision variables as a penalty function, it is possible to treat all decision variables as a continuous decision variable. As a result, the penalty parameter for the penalty function is needed. In this paper, we also present how to determine the penalty parameter for the penalty function. Through mathematical and structural optimization problems, we examine the validity of PSO for the mixed decision variables.

show abstract

A Sequential Approximation Method Using Neural Networks for Nonlinear Discrete-Variable Optimization with Implicit Constraints.

Cited by 11 publications

References 8 publications

One-pass milling machining parameter optimization to achieve mirror surface roughness

One-pass milling machining parameter optimization to achieve mirror surface roughness

Global Optimization by Generalized Random Tunneling Algorithm (4th Report Application to the Nonlinear Optimum Design Problem of the Mixed Design Variables)

A method for mixed integer programming problems by particle swarm optimization

Contact Info

Product

Resources

About