A DIRECTED MUTATION OPERATION FOR THE DIFFERENTIAL EVOLUTION ALGORITHM Hui-Yuan Fan 1, 2 , and Jouni Lampinen 1,* 1 Department of Information Technology Lappeenranta University of Technology P. O. Box 20, FIN-53851 Lappeenranta, Finland E-mail: {fan, jlampine}@lut.fi 2 School of Energy and Power Engineering Xi'an Jiaotong University Xi'an, 710049, P R China Email: huiyfan@xjtu.edu.cn * Corresponding author A modification is proposed for the differential evolution algorithm which aims to improve its convergence performance. The modification embeds an additional operation, directed mutation, into an original version of the differential evolution. The aim of this operation is to increase the convergence velocity of the differential evolution and thereby to obtain an acceptable solution with a lower number of objective function evaluations. Such an improvement can be useful in many real-world problems where the evaluation of a candidate solution is a computationally expensive operation and consequently finding the global optimum or a good sub-optimal solution with the algorithm is too time-consuming, or even impossible within the time available. The modified version of the differential evolution was empirically examined with a suite of six well-known test problems and compared with the original version of the differential evolution algorithm. The obtained numerical simulation results suggested drawing a preliminary conclusion that the modified version statistically outperforms the original one. Keywords: Evolutionary algorithm, Differential evolution, Nonlinear optimization. A SIMULATION BASED GENETIC ALGORITHM FOR RISK-BASED PARTNER SELECTION IN NEW PRODUCT DEVELOPMENT Hongyi Cao and Dingwei Wang Collage of Information Science and Engineering Northeastern University Shenyang, Liaoning, People’s Republic of China In this paper, we investigate the problem of partner selection in new product development. First, we give a formal description of the problem and model it as a 0-1integer programming problem with non-linear objective function and stochastic constraints. Because of the complexity of the constraints, a Monte Carlo method is used to measure the probabilities of stochastic constraints satisfaction. Then, we develop a simulation based genetic algorithm approach to find the optimal solution for partner selection. The approach is demonstrated by some numerical examples. The results show that the suggested approach has high efficiency and the model has potential to practical applications. Keywords: development. Genetic algorithm, Monte Carlo simulation, Partner selection, Risk analysis, New product ADVANCED PROCESS PLANNING AND SCHEDULING WITH PRECEDENCE CONSTRAINTS AND MACHINE SELECTION USING A GENETIC ALGORITHM Chiung Moon and Young Hae Lee Department of Industrial Engineering Hanyang University, Ansan 425-791, Korea Tel:+82-31-400-5268 Fax:+82-31-400-3843 cumoon@hanyang.ac.kr or cumoon@hanmail.net ABSTRACT This paper deals with integrated process planning and scheduling problems with minimizing makespan for a flexible flow manufacturing where alternative operations sequences with precedence constraints and alternative machines. The problem is formulated as a mathematical model which includes operation sequencing, machine selection, and operation scheduling. The integrated planning of having more than one machine to perform the same operation and precedence constraints for sequences increases the size of the solution space, and consequently, makes the problem even more complex. We develop a new genetic algorithm approach using topological sort to solve the model efficiently. Schedules with operations sequences and machine selections are currently decided by the proposed approach. Some experimental results are presented for various problem sizes and parameter settings to describe the performance of the proposed approach. KEYWORDS Integrated planning and scheduling, topological sort, traveling salesman problem, genetic algorithms. OPTIMIZING THE PRODUCTION SCHEDULING OF A PETROLEUM REFINERY THROUGH GENETIC ALGORITHMS Mayron Rodrigues de Almeida Sílvio Hamacher Marco Aurélio Cavalcanti Pacheco Marley Maria B.R. Vellasco Department of Industrial Engineering Pontifícia Universidade Católica do Rio de Janeiro Rua Marquês de São Vicente, 225, Gávea, Rio de Janeiro, RJ, Brasil, 22453-900 {Mayron@ons.org.br ; Hamacher@rdc.puc-rio.br} ICA – Laboratory of Applied Computational Intelligence Department of Electrical Engineering Pontifícia Universidade Católica do Rio de Janeiro Rua Marquês de São Vicente, 225, Gávea, Rio de Janeiro, RJ, Brasil, 22453-900 {Marco@ele.puc-rio.br ; Marley@ele.puc-rio.br} This paper presents a method, based on Genetic Algorithms, to optimize the production scheduling of the fuel oil and asphalt area in a petroleum refinery. The fuel oil and asphalt area is a multi-product plant, with two machine stages – one mixer and a set of tanks – with no setup time and with resource constrains in continuous operation. Two genetic algorithm models were developed to establish the sequence and size of all production shares. A special mutation operator – Neighborhood Mutation - is proposed to minimize the number of changes in the production. A multiobjective fitness evaluation technique, based on an energy minimization method, was also incorporated to the Genetic Algorithm models. The results obtained confirm that the proposed Genetic Algorithm models, associated with the multi-objective energy minimization method, are able to solve the scheduling problem, optimizing the refinery operational objectives. Keywords: Production Scheduling, Petroleum Refinery, Genetic Algorithms. A Genetic Algorithm Approach to Solving a Multiple-Inventory Loading Problem J. Cole Smith* Department of Systems and Industrial Engineering University of Arizona P.O. Box 210020 Tucson, AZ 85721 Email: cole@sie.arizona.edu In this paper we consider a multiple-inventory loading problem involving a set of commodities that must be transported from a distributor to a retailer. The vehicle carrying out this distribution is divided into several compartments, in which only one type of commodity may be loaded. The problem becomes one of determining optimal assignments of vehicle compartments to commodities in order to minimize a mix of transportation and inventory costs. We first demonstrate the weakness of the underlying linear relaxation of a traditional mixed-integer programming approach that must be solved in a branch-and-bound framework. Instead of pursuing the development of an exact algorithm, we instead recommend the use of a genetic algorithm to quickly provide good quality solutions. Next, we introduce an additional strategy for defeating symmetry complications arising in certain specially structured problems. Finally, the effectiveness of each of the proposed techniques is demonstrated on a test bed of problems. Keywords: Genetic algorithms, Mathematical modeling, Inventory management, Logistics * Acknowledgement: This research was supported by University of Arizona Foundation and by the Office of the Vice President for Research and Graduate Studies. The author also gratefully acknowledges the comments provided by two anonymous referees, the Guest Editor, and by Mr. Ashwin Naik, which led to an improved presentation of this topic. A NEURAL NETWORKS APPROACH FOR DUE-DATE ASSIGNMENT IN A WAFER FABRICATION FACTORY Pei-Chann Chang and Jih-Chang Hsieh Department of Industrial Engineering Yuan Ze University Nei-Li, Taoyuan, Taiwan Email: iepchang@saturn.yzu.edu.tw The production processes in a wafer fabrication factory are very complicated and time-consuming. This presents a challenging problem to the production planning and scheduling department for the due-date assignment of each order. This research proposes a simulation model to mimic a real wafer fabrication factory and the flowtime of each order is collected for the purpose of due-date assignment. Various influential variables related to the flowtime of each order are identified through regression analysis. Accordingly, a neural network model is established to forecast the due-date of each order. The system is very applicable in the real world and the experimental results show that the proposed approach is very convincing when compared with the traditional approaches. Keywords: Wafer fabrication factory, Due-date assignment, Backpropagation neural networks JOINT MONITORING OF THE MEAN AND VARIANCE OF A PROCESS BY USING AN ARTIFICIAL NEURAL NETWORK APPROACH Chuen-Sheng Cheng and Shin-Jia Chen Department of Industrial Engineering and Management Yuan-Ze University 135 Yuan-Tung Road, Taoyuan,Taiwan, R.O.C. In this paper we consider the joint control of process mean and variance using artificial neural network technology. The performance of the neural network was evaluated by estimating the recognition accuracy. Extensive comparisons show that the neural network appears to be a good control procedure for joint monitoring of the process mean and variance. Keywords: Shewhart Control Chart, Neural Network. COMPARISON OF NEURAL AND STATISTICAL ALGORITHMS FOR SUPERVISED CLASSIFICATION OF MULTI-DIMENSIONAL DATA Te-Sheng Li1, Chang-You Chen2 and Chao-Ton Su2 1Department of Industrial Engineering and Management Minghsin Institute of Technology, Hsinchu, 300 Taiwan 2Department of Industrial Engineering and Management National Chiao Tung University, Hsinchu, 300 Taiwan Various algorithms for supervised classification of multi-dimensional data have been implemented in the past decades. Among these algorithms, neural and statistical classifiers are two major methodologies used in the literature. In this paper, a comparison of different neural networks and statistical algorithms used for classification is presented. Three types of neural classifiers are considered: Back-propagation (BP) network, Radial Basis Function (RBF), and Learning Vector Quantization (LVQ). Meanwhile, the k-nearest neighbor (KNN) statistical classifier and linear discriminant analysis (LDA) are also discussed in order to compare the accuracy of classification with those of using neural network models. This paper includes an introduction to the theoretical background of the classifiers, their implementation procedures, and two case studies to evaluate their performance in diagnosis of diseases and glass identification. Both neural networks and statistical models are demonstrated to be efficient and effective methods for multi-dimensional data classification. For neural classifiers, the type of neural network, the type of data, and the parameters of the neural network may have considerable impact on the classification accuracy. For statistical models, the type of data distribution and the criteria used to determine the thresholds are the two most important factors in classification. In the cases studied in this paper, the overall performance of neural networks is better than that of statistical models. Finally, the comparison and discussion of these approaches are presented in view of practical and theoretical considerations. Keywords: Back-propagation, Radial Basis Function, Learning Vector Quantization, k-nearest neighbor, Discriminant analysis THE PREDICTION OF AIRPLANE LANDING GRAVITY USING CASEBASED REASONING Chaochang Chiu and Nan-Hsing Chiu Department of Information Management Yuan Ze University Taoyuan, Taiwan, R. O. C. Email:{C. C. Chiu, imchiu@saturn.yzu.edu.tw; N. H. Chiu, s887720@mail.yzu.edu.tw} Most flight accidents occurring worldwide are due to the lack of an appropriate approach in the landing phase. Recent data mining developments have provided aviation insights into the landing phase. Among these methods, case-based reasoning is a potential approach that can be applied for predicting landing gravity. This research proposes a novel model construction method that consists of non-linear similarity functions and dynamic weighting mechanisms to optimize the prediction accuracy. We illustrate our approach with the data obtained directly from flight data recorders of Boeing 747-400 airplanes. This experiment also shows that non-linear similarity functions demonstrate better prediction accuracy over the results from other approaches. Keywords: Case-based reasoning, Genetic algorithms, Feature weights, Similarity functions, Landing gravity.