實驗設計中，由於實驗人員不知道正確的模型是為何，因此如何建構一個實驗設計(experimental design)讓實驗人員能夠分辨各種不同的模型是個重要的議題，但是如何建構一個最佳模型分辨實驗設計是個困難的問題，而近幾年，粒子群聚演算法(Particle Swarm Optimization)被廣泛應用在各領域並且有許多優點。由於粒子群聚演算法適用於解決連續型最佳化的問題，因此，此篇文章主要是利用粒子群聚演算法的概念建構出離散型粒子群聚演算法(Discrete Particle Swarm Optimization)，利用模型分辨設計準則來找出其最適實驗設計。數值結果發現使用離散型粒子群聚演算法能夠比其他演算法所找到的最適實驗設計更好。

Since the experimenters might not have prior knowledge on which main effects or interactions were likely to be significant, it is important to construct a experimental design that have the capability of screening main effects and two-factor interactions. Agboto et al. (2010) proposed model discrimination criteria. But how to construct an optimal model discrimination design based on these criteria is a difficult question. In recent years, Particle Swarm Optimization has been wildly used in many aspects because of the advantages of the PSO algorithm. In our study, since the PSO algorithm is designed to solve the continuous optimization problems, we need to modify the PSO algorithm due to particular design structure. The purpose of this paper is to present the Discrete Particle Swarm Optimization algorithm to construct an optimal model discrimination design. We implement our algorithm to optimize model discrimination design under the model discrimination criterion and compare results with Agboto et al. (2010) and the coordinate-exchange algorithm. The results show that the DPSO algorithm performs well and is compatible with other algorithms.

Agboto, V., Li,W., and Nachtsheim, C. (2010). Screening designs for model discrimination. Journal of Statistical Planning and Inference, 140(3):766–780.

Atkinson, A. and Donev, A. (1989). The construction of exact d-optimum experimental designs with application to blocking response surface designs. Biometrika, 76(3):515–
526.

Atkinson, A. and Fedorov, V. (1975). Optimal design: experiments for discriminating between several models. Biometrika, 62(2):289–303.

Bingham, D. and Li, W. (2002). A class of optimal robust parameter designs. Journal of Quality Technology, 34:244–259.

Cook, R. and Nachtsheim, C. (1980). A comparison of algorithms for constructing exact d-optimal designs. Technometrics, 22(3):315–324.
Fedorov, V. (1972). Theory of optimal experiments. Elsevier.

Johnson, M. andNachtsheim, C. (1983). Some guidelines for constructing exact d-optimal designs on convex design spaces. Technometrics, 25(3):271–277.

Jones, B., Li, W., Nachtsheim, C., and Ye, K. (2007). Model discrimination—another perspective on model-robust designs. Journal of Statistical Planning and Inference,
137(5):1576–1583.

Kennedy, J. and Eberhart, R. (1995). Particle swarm optimization. Proceedings of IEEE International Conferenec on Neural Networks.

Li,W. and Nachtsheim, C. (2000). Model-robust factorial designs. Technometrics, 42:345–352.

Li, W. and Wu, C. (1997). Columnwise pairwise algorithms with applications to the construction of supersaturated designs. Technometrics, 39(2):171–179.

Loeppky, J., Sitter, R., and Tang, B. (2007). Non-regular designs with desirable projection properties. Technometrics, 49(4):454–467.

Meyer, R. and Nachtsheim, C. (1995). The coordinate-exchange algorithm for constructing exact optimal designs. Technometrics, 37(1):60–69.

Phoa, F., Chen, R., Wang, W., and Wong, W. (2015). Optimizing two-level supersaturated designs using swarm intelligence techniques. Technometrics.

Smucker, B., Castillo, E., and Rosenberger, J. (2012). Model-robust two-level designs using coordinate exchange algorithms and a maximin criterion. Technometrics, 54(4):367–375.

Srivastava, J. (1975). Designs for searching nonnegligible effects. In A Survey of Statistical
Designs and LinearModels. North-Holland, Amsterdam.

Sun, D. (1993). Estimation capacity and related topics in experimental designs. Unpublished Ph.D. Dissertation, Department of Statistics and Actuarial Science, University of Waterloo.