9TH INTERNATIONAL CONFERENCE ON APPLIED OPERATIONAL RESEARCH
18-20 DECEMBER 2017 TAOYUAN, TAIWAN
FUZZY GREEDY SEARCH FOR COMBINATORIAL OPTIMIZATION PROBLEMS
ORLab Analytics, Vancouver BC Canada
Abstract. This presentation gives an overview of the fuzzy greedy search methodology for the integration of approaches for hard combinatorial optimisation problems. The use of a search technique on a solution space is central to the design a heuristic. Indeed, applying a robust search technique significantly improves the overall performance. The proposed methodology evaluates objects in a way that combines fuzzy reasoning with a greedy mechanism, thereby exploiting a fuzzy solution space using greedy methods. More recent developments and applications will be also discussed during the presentation.
DYNAMIC APPOINTMENT SCHEDULING IN PRIORITY QUEUEING SYSTEMS WITH ACCESS TIME TARGETS
Carrie Ka Yuk Lin
Department of Management Sciences, College of Business, City University of Hong Kong, Hong Kong (SAR), China
Abstract. Multi-class priority queuing systems with access time targets and dynamic arrivals are common in many environments, including emergency, healthcare, logistics and hospitality services. This study focuses on the waiting line management of new cases in the specialist outpatient clinics of the Hospital Authority of Hong Kong. The objectives are to assign appointment dates in a way to fulfill the access time targets for urgent (highest priority) and semi-urgent (second highest priority) new cases while maintaining acceptable access time for non-urgent (routine) new cases. A dynamic appointment scheduling algorithm is developed and tested through simulated instances created from the waiting time data and hospital attendance data for a one-year horizon. The experimental results are compared with the reported performance statistics. Performances are also compared with results of the deterministic problem assuming the arrival data were given. For certain ordered set of objective weights relevant to the current problem, the deterministic problem can be formulated as a transportation model and solved optimally by the classical transportation algorithm with an additional checking procedure.
TRANSIT NETWORK DESIGN AND SCHEDULING PROBLEM FOR MULTI-DEPOT ROUND-TRIP FIXED-INTERVAL ROUTES
James C. Chu
Associate Professor, Department of Civil Engineering, National Taiwan University, Taiwan
Abstract. The study formulates a mixed integer programming (MIP) model for the transit network design and scheduling problem (TNDSP) considering multi-depot round-trip fixed-interval routes and static travel time. TNDSP is particularly cumbersome to formulate and solve because it combines three major planning activities in transit network problems (TNP): network design, frequency setting, and timetabling. Therefore, the problem is rarely considered in the literature and the few models developed in the related studies are often size-limited or simplified (Guihaire and Hao, 2008). For example, Yan and Chen (2002) considered a rural bus system in which waiting time is ignored. Quak (2003) solved the problems of network design and scheduling in a sequential manner rather than as a single problem.
JOB INSERTION FOR THE PICKUP AND DELIVERY PROBLEM WITH TIME WINDOWS
Yi Qu 1, Timothy Curtois
1 Newcastle Business School, Northumbria University, Newcastle, UK
Abstract. Two heuristic algorithms are proposed for a practical but relatively under-studied vehicle routing scenario. It requires the insertion of jobs into already planned routes. It occurs when new jobs arrive throughout a day but the current plans are already being performed. The benefit of solving such a problem is providing a better service for collection and delivery jobs whilst also providing better vehicle fill rates and increased revenue for delivery companies. Solutions must be generated quickly because of the dynamic nature of the problem. Two iterative heuristic algorithms are presented and tested on a well-known benchmark set. The algorithms are able to insert new jobs quickly and efficiently and even found some new best known solutions for the benchmark instances.
STOCHASTIC SIMULATION BASED GENETIC ALGORITHM FOR A PRODUCTION REPAIR MODEL
R. K. Jana 1, Yuvraj Gajpal 2, and B. Chakraborty 3
1 Indian Institute of Management Raipur, GEC Campus, Sejbahar, CG 492015, India
2 Supply Chain Management, Asper School of Business, University of Manitoba, Canada
3 Department of Mathematics Institute of Engineering & Management, Kolkata, WB 700091, India
firstname.lastname@example.org; email@example.com; firstname.lastname@example.org
Abstract. This paper deals with a stochastic production repair model with chance constraints. A solution to such problems has great importance to the managers because such a solution serves as an invaluable tool for them in the process of developing a production plan to their companies. The common way of tackling such a problem is to derive the crisp equivalent of the original problem. This is possible if the parameters involved in the chance constraint follow some specific distribution or if we can estimate or approximate the chance constraints using some estimation procedure or numerical technique. In this paper, we first formulate a multi-product, single period and single plant chance constraint programming model having production and repair work simultaneously. The uncertainty about the cost, resources and man hour required to repair a product, arrival of products for repair, resources and manhour availabilities bring randomness to the formulated model. To solve the mathematical model, we propose a stochastic simulation based genetic algorithm. This algorithm can deal the stochastic parameters having any possible distribution. Finally, a numerical example is presented to prove the efficiency of the proposed algorithm.
MODELING EMERGENCY MEDICAL RESPONSE TO A MASS CASUALTY INCIDENT IN MULTIPLE LOCATIONS
Yun-Zhu Lin 1, Pei-Jung Liao 2
1 Associate Professor, Taoyuan City, Taiwan
2 Project assistant, Taoyuan City, Taiwan
Abstract. During the emergency response to mass casualty incidents decisions relating to the extrication, transporting and treatment of casualties are made in a real-time, sequential manner. In order to increase survival rates of casualties, the total response time of casualties, including waiting times at emergency sites, transportation times, waiting times at hospitals, and treatment times, needs to be minimized. The stochastic nature of injury degree and treatment time of casualties is considered in this study. We developed a multi-stage heuristic on a rolling time horizon to help decisions-makers make rescue decisions of emergency response to mass casualty incidents. Simulation of a multi-server queueing system with heterogeneous service times is embedded in the algorithm. The heuristic is evaluated over several potential problems, with results confirming its effectiveness.
SIMULATED ANNEALING OPTIMIZATION OF TUNED MASS DAMPERS FOR VIBRATION CONTROL OF SEISMIC-EXCITED BUILDINGS
Ming-Yi Liu 1, Wen-Che Liang 2, and Yun-Zhu Lin 1
1 Department of Civil Engineering, Chung Yuan Christian University, Taoyuan City, Taiwan
2 Department of Civil Engineering, National Cheng Kung University, Tainan City, Taiwan
Abstract. The objective of this paper is to investigate the simulated annealing optimization of tuned mass dampers (TMDs) for vibration control of high-rise buildings under seismic excitations. The computational procedure for an analytical model, including the frequency domain analysis, optimization analysis and time domain analysis, is presented for this purpose. Numerical examples, including the model validation and effectiveness assessment, are also provided to illustrate the analytical model. The numerical results determined by the simulated annealing are consistent with the exact solutions obtained from the gradient-based algorithm, suggesting that the simulated annealing provides the sufficient accuracy for such problems. Furthermore, the dynamic displacements of the main structure can be successfully controlled by the TMD, indicating that the vibration energy transferred from the main structure to the TMD is attributed to the frequency loci veering and mode localization when the natural frequency of the main structure and that of the TMD approach one another.
AN APPLICATION OF MATRIX GAMES WITH TRAPEZOIDAL INTUITIONISTIC FUZZY PAY OFFS TO TRANSPORTATION PROBLEM
Ratnesh R. Saxena 1, Ritika Chopra 2, Sanjiv Kumar 2
1 Department of Mathematics, Deen Dayal UpadhyayA College, Delhi University, Delhi, India
2 Department of Mathematics, Delhi University, Delhi, India
Abstract. In this paper, a two person zero sum game, with payoffs expressed as trapezoidal intuitionistic fuzzy numbers is considered. Trapezoidal intuitionistic fuzzy numbers (TrIFNs) are defined in a more general way by relaxing the normality condition in the definition. The generalized trapezoidal intuitionistic fuzzy numbers (GTrIFNs) are converted to interval valued fuzzy numbers by taking (a,b)-cuts. Then a new ranking index is introduced which gives a saddle point (if it exists) in an effective way. Numerical examples are provided to illustrate the proposed methodology. Further, the proposed ranking technique is applied to a transportation problem to get a fuzzy optimal solution using MODI method.
A ESTIMATING THE OPTIMAL INTRA-COMPANY WAGE GAPS FOR IMPROVING PRODUCTIVITY -EVIDENCE FROM JAPANESE LISTED COMPANY-
School of Commerce, Meiji University, Tokyo, Japan
Abstract. This study examined the contribution on the productivity of the companies by the intra-company wage gaps as a measure of the performance-based wage systems and got some significant results and implications for management. In order to remove reverse effect by productivity on wage gap, instrumental variable methods are applied. Analyses on all industries implied that larger wage gaps do not motivate employees of Japanese companies. Analyses by nonlinear models to confirm the existence of the optimal wage gaps implied the existence of the optimal wage gaps in a few industries, but aggregately worst wage gaps are recognized. Concretely, effects by wage gaps on the productivity decrease, as wage gaps expand. These findings are contrary to those by Cirillo et al. (2017) in Europe, Dai et al. (2017) in China and Park et al. (2017) in Korea as most recent previous researches.
BENCHMARK FUNCTIONS BASED PERFORMANCE EVALUATION BY INFERENCE MODEL PYRAMID TREE (PT) AND OPERATION TREE (OT)
Li-Chuan Lien 1,4, Shou-Bin Chen 2, Jing-Yi Xu 3, Yan-Ni Liu 4 and Qi-Sheng Wang 1
1 Chung Yuan Christian University, Taoyuan, Taiwan
2 Fuzhou CityConstruction Design and Research Institute Co., Ltd., Fuzhou, China
3 China Construction Fifth Engineering Branch Co., Ltd, Fuzhou, China
4 Fujian University of Technology, Fuzhou, China
Abstract. Structure equation model (SEM) approaches based on the historical data and perform its formula. Thus, the researcher can observed internal knowledge on problems and ensure sensitivity analysis or decision making via formula. This study used an SEM inference model approach, named pyramid tree (PT), model structure idea was from pyramid and operation trees (OTs). Four triangle layers organize a pyramid object. OT models formula based on the numerical historical data. Particle bee algorithm (PBA) optimize PT model’s structure based on the structure behavior that designed by this study. In order to measure PT performance, this study evaluated PT performance based on proposed benchmark functions. This study evaluated performance of PT model’s structure with fourteen 2- or 4-dimensional benchmark functions by 100, 400, 700 and 1000 of PBA iterations. Results showed PT performance was satisfied to fit those benchmark functions formula and better than single OT. PT is a self-modeling formula inference model method and more accurate than single OT. It is suitable to solve the problem while human would to figure out the formula based on the historical data such as material composite or other practical problems.
BENCHMARKING STATE ROAD TRANSPORT UNDERTAKINGS OF INDIA: A DEA-BASED STEPWISE APPROACH
Associate Professor, Department of Mathematics, Shaheed Rajguru College of Applied Sciences for Women (University of Delhi), India
Abstract. Benchmarking is the practice of any decision making unit to compare key metrics of its operations with the peers. It has a twofold advantage. Firstly, it allows the units to see how efficiently they are performing in comparison to others. Secondly, it allows the units to identify and analyze the areas of potential improvements to become more competitive. Data Envelopment Analysis (DEA) is a non-parametric tool used for efficiency evaluations and to yield a reference benchmark target for each inefficient unit in the data set under study. However, the benchmarks provided are at times practically infeasible to attain. The inefficient units need to strategize the best practices to be adopted in a stepwise manner so that gradually they can raise themselves up to the level of the reference target. In the present paper, a stepwise benchmarking approach has been developed by selecting intermediate targets. The transitional target units have been selected on the basis of similarity between uses of inputs, yield of outputs and peer appraisal. Initially the units were classified as efficient, marginally efficient, below average and highly inefficient units based on their DEA scores. For a highly inefficient unit, an intermediate target from the next cluster was selected using the criterion of distance between the input and output values and maverick index. In this manner a feasible efficient path has been identified to be adopted by a highly inefficient unit to grow to an efficient unit.
© ORLab Analytics Inc.