Aydin Sipahioglu - Academia.edu (original) (raw)

Papers by Aydin Sipahioglu

Eskişehir Osmangazi Üniversitesi Mühendislik ve Mimarlık Fakültesi Dergisi

Many optimization problems are complex, challenging and take a significant amount of computationa... more Many optimization problems are complex, challenging and take a significant amount of computational effort to solve. These problems have gained the attention of researchers and they have developed lots of metaheuristic algorithms to use for solving these problems. Most of the developed metaheuristic algorithms are based on some metaphors. For this reason, these algorithms have algorithm-specific parameters to reflect the nature of the inspired metaphor. This violates the algorithm's simplicity and brings extra workload to execute the algorithm. However, the optimization problems can also be solved with simple, useful, metaphor-less and algorithm-specific parameter-less metaheuristic algorithms. So, it is the essential motivation behind this study. We present a novel metaheuristic algorithm called Discrete Rao Algorithm (DRA) by updating some components of the generic Rao algorithm to solve the combinatorial optimization problems. To evaluate the performance of the DRA, we perform...

The milk-run is an in-plant lean logistics application where from a central warehouse; full-boxes... more The milk-run is an in-plant lean logistics application where from a central warehouse; full-boxes of components are supplied to the line-side buffer stock areas of the assembly stations, on a just-in time basis, and in a cyclic manner, so that the stations do not run out of stock [10]. The associated problem is the cyclic Inventory Routing Problem (IRP). In this study, a two stage approach is proposed where a mixed-integer mathematical model is solved to assign the stations to the routes and to decide the service periods of the routes. At the second stage, travelling salesman problem needs to be solved to find the sequence at each route. In addition, an alternative mixed-integer mathematical model is developed where routes are constructed such that the sequence of stations and the service periods are determined for each route, simultaneously. Both of the models are assignment-based that considerably reduces the solution times of the IRP. A medium-size hypothetical data set was solve...

In this study, a vehicle routing problem-based approach is presented to construct non-intersectin... more In this study, a vehicle routing problem-based approach is presented to construct non-intersecting routes for the members of a mobile robot team. It is assumed that each robot starts from a central location such as the charging point, completes its route and returns to the starting location. The proposed method consists of three algorithms: a sweep algorithm determines the position of each node in clockwise (or counter clockwise) manner with respect to the starting location; savings algorithm calculates the saving obtained by adding a node to the route of a robot; Dijkstra's shortest path algorithm is used to calculate the shortest distance from any node to another one when the network is sparse. Simulations are performed using some benchmark VRP problems and results are compared with the optimal solution of the same problems. It is shown that our approach constructs routes significantly fast with near optimal energy consumption.

Pamukkale University Journal of Engineering Sciences, 2019

Job Shop Scheduling Problem (JSP) is one of the hardest NP-hard class combinatorial optimization ... more Job Shop Scheduling Problem (JSP) is one of the hardest NP-hard class combinatorial optimization problems. Flexible Job Scheduling Problem (FJSP) occurs with the use of parallel machines in job shop environment and it is more complex than JSP because it contains two sub problems: operation sequencing and operation assignment to machines. There are two main approaches to solve FJSP: Hierarchical approach and integrated approach. In hierarchical approach, machine assignment and operation sequencing are independent from each other whereas in integrated approach they occur simultaneously. There are many heuristic methods to solve FJSP in the literature. Artificial Immune System inspired by the vertebrate immune system (AIS) is one of these methods. In this study, an artificial immune system approach based on hierarchical approach is developed to solve FJSP. To demonstrate the effectiveness of the algorithm, numerical experiments by using three benchmark problem sets are conducted. In pr...

Production Research, 2021

Milk-run, a cyclic material delivering system, aims to increase the efficiency of transportation ... more Milk-run, a cyclic material delivering system, aims to increase the efficiency of transportation and supply chain based on lean logistics perspective. There are two kinds of milk-run systems as supplier and in-plant milk-run system in the literature. In-plant milk-run system that has growing appeals with Industry 4.0 concept, is applied to manage the process of delivering materials from warehouse to assembly stations in plants. This system can be implemented using Autonomous Vehicles (AV), which provide automated materials handling. However, a challenging problem arises in determining milk-run routes and periods for each AV, simultaneously. Besides, this problem becomes even harder in the presence of assembly stations with buffer stock constraint and requiring more than one commodity (multi-commodity). Since this problem is quite difficult to handle with exact solution methods, this problem is tackled here by using Simulated Annealing algorithm. To evaluate the performance of the proposed algorithm, we carry out experiments using a range of test problems. The computational results indicate that the suggested algorithm is efficient to obtain both milk-run routes and periods for each AV in reasonable computation times.

Journal of Industrial & Management Optimization, 2017

One of the most common plastic manufacturing methods is injection molding. In injection molding p... more One of the most common plastic manufacturing methods is injection molding. In injection molding process, scheduling of plastic injection machines is very difficult because of the complex nature of the problem. For example, similar plastic parts should be produced sequentially to prevent long setup times. On the other hand, to produce a plastic part, its mold should be fixed on an injection machine. Machine eligibility restrictions should be considered because a mold can be usually fixed on a subset of the injection machines. Some plastic parts which have same shapes but different colors are used same mold so these parts can only be scheduled simultaneously if their mold has copies, otherwise resource constraints should be considered. In this study, a multi-objective mathematical model is proposed for parallel machine scheduling problem to minimize makespan, total tardiness, and total waiting time. Since NP-hard nature of problem, this paper presents a two-stage mathematical model and a two-stage solution approach. In the first stage of mathematical model, jobs are assigned to the machines and each machine is scheduled separately in the second stage. The integrated model and two-stage mathematical model are scalarized by using goal programming, compromise programming and Lexicographic Weighted Tchebycheff programming methods. To solve large-scale problems in a short time, a two-stage solution approach is also proposed. In the first stage of this approach, jobs are assigned to machines and scheduled by using proposed simulated annealing algorithm. In the second stage of the approach, starting time, completion time and waiting time of the jobs are calculated by using a mathematical model. The performance of the methods is demonstrated on randomly generated test problems.

The quadratic multiple knapsack problem (QMKP) is a generalization of the quadratic knapsack prob... more The quadratic multiple knapsack problem (QMKP) is a generalization of the quadratic knapsack problem from a single knapsack to k knapsacks. In this study, the mathematical model of the QMKP is generalized as covering the constraints that we may face in real life problems. A new hybrid algorithm which combines F-MSG algorithm and genetic algorithm is proposed to solve the Generalized QMKP (GQMKP). The performance of the F-MSG is analyzed and compared by using randomly generated test instances. Keywords: generalized quadratic multiple knapsack problem (GQMKP), F-MSG (modified subgradient algorithm based on feasible values), genetic algorithm (GA).

IEEE International Symposium on Intelligent Control, 2006

Abstract One of the main concerns in multi-robot applications is the effects of interactions amon... more Abstract One of the main concerns in multi-robot applications is the effects of interactions among the robots on the total performance of the team. If the robots are assigned to spatially separate tasks, the negative impact of these interactions may be decreased. In this paper a VRP-based method is proposed to create non-intersecting routes for a team of robots. In the method, a heuristic approach composed of sweep, savings, and Dijkstra's shortest path algorithm is used to find routes for each robot. Simulations are performed on some VRP ...

Lecture Notes in Computer Science

The Quadratic Multiple Knapsack Problem (QMKP) is a generaliz- ation of the quadratic knapsack pr... more The Quadratic Multiple Knapsack Problem (QMKP) is a generaliz- ation of the quadratic knapsack problem, which is one of the well-known combinatorial optimization problems, from a single knapsack to k knapsacks with (possibly) different capacities. The objective is to assign each item to at most one of the knapsacks such that none of the capacity constraints are violated and the

IEEE Transactions on Cybernetics, 2014

In this study, a novel dynamic path planning approach is proposed for multi-robot sensor-based co... more In this study, a novel dynamic path planning approach is proposed for multi-robot sensor-based coverage considering energy capacities of the mobile robots. The environment is assumed to be narrow and partially unknown. A Generalized Voronoi diagram-based network is used for the sensor-based coverage planning due to narrow nature of the environment. On the other hand, partially unknown nature is handled with proposed dynamic re-planning approach. Initially, the robots are assumed to be at the same depot with equal initial energy capacities. In this case, an initial complete coverage route is constructed considering robot energy capacities using classical capacitated arc routing problem (CARP) approach with some minor modifications related to coverage problem. But, due to partially unknown nature, the robots may face with blockage on routes, and a fast re-planning is required which considers remaining energy capacities and current positions of the robots. So, new plan is obtained by a modifying Ulusoy's algorithm that was developed for classical CARP. The developed algorithm is coded in C++ and implemented on P3-DX mobile robots in MobileSim simulation environment.

2009 IEEE International Conference on Control Applications, 2009

In this study, a novel sensor-based coverage algorithm is proposed for multi-robots considering e... more In this study, a novel sensor-based coverage algorithm is proposed for multi-robots considering energy capacities of the mobile robots. Firstly, the environment is modeled by a Generalized Voronoi diagram-based graph to guarantee complete sensor based coverage. Secondly, depending on required arc set, an initial complete coverage route is created by using Chinese postman problem (CPP) and/or rural postman problem (RPP).

Robotics and Autonomous Systems, 2008

Optimization, 2013

ABSTRACT

European Journal of Operational Research, 2007

In this paper we study a 1.5-dimensional cutting stock and assortment problem which includes dete... more In this paper we study a 1.5-dimensional cutting stock and assortment problem which includes determination of the number of different widths of roll stocks to be maintained as inventory and determination of how these roll stocks should be cut by choosing the ...

Computers & Operations Research, 2014

Advanced Robotics, 2009

Multi-robot systems have recently received a great deal of attention due to the ability to perfor... more Multi-robot systems have recently received a great deal of attention due to the ability to perform an assigned task in a more reliable, faster and cheaper way beyond what is possible with a single robot. However, they may have some drawbacks, such as obstruction among robots during a task. Non-intersecting tours are preferable for the robots to prevent obstruction. Moreover,