Ching-Jung Ting

Dynamic diversity control in genetic algorithm for mining unsearched solution space in TSP problems

The applications of genetic algorithms (GAs) in solving combinatorial problems are frequently faced with a problem of early convergence and the evolutionary processes are often trapped in a local optimum. This premature convergence occurs when the population of a genetic algorithm reaches a suboptimal state that the genetic operators can no longer produce offspring with a better performance than their parents. In the literature, plenty of work has been investigated to introduce new methods and operators in order to overcome this essential problem of genetic algorithms. As these methods and the belonging operators are rather problem specific in general. In this research, we take a different approach by constantly observing the progress of the evolutionary process and when the diversity of the population dropping below a threshold level then artificial chromosomes with high diversity will be introduced to increase the average diversity level thus to ensure the process can jump out the local optimum and to revolve again. A dynamic threshold control mechanism is built up during the evolutionary process to further improve the system performance. The proposed method is implemented independently of the problem characteristics and can be applied to improve the global convergence behavior of genetic algorithms. The experimental results using TSP instances show that the proposed approach is very effective in preventing the premature convergence when compared with other approaches.

AN IMPROVED ANT COLONY SYSTEM ALGORITHM FOR THE VEHICLE ROUTING PROBLEM

ABSTRACT The vehicle routing problem (VRP), a well-known combinatorial optimization problem, holds a central place in logistics management. Many meta-heuristic approaches like Simulated Annealing (SA), Genetic Algorithms (GA), Tabu Search (TS), and Ant Colony Optimization (ACO) have been proposed to solve VRP. Ant Algorithm is a distributed meta-heuristic approach that has been applied to various combinatorial optimization problems, including traveling salesman problem, quadratic assignment problem. In this research, we proposed an improved ant colony system (IACS) algorithm that possesses a new state transition rule, a new pheromone updating rule and diverse local search approaches. The computational results on 14 VRP benchmark problems show that our IACS yields better solutions than those of other ant algorithms in the literature and is competitive with other meta-heuristic approaches in terms of solution quality.

Optimization Through Simulation of Waterway Transportation Investments

The cost of tow delays is a serious problem in a waterway network. One way to reduce the delay cost is to increase capacity at waterway locks. Planners must determine how much additional capacity to provide at particular lock sites and when to implement the capacity expansion projects. Answers for such project sizing and timing problems are difficult to obtain analytically. The use of a new approach for optimizing through simulation, called simultaneous perturbation stochastic approximation (SPSA), is investigated. This approach, which seeks optimal values for all decision variables after each pair of simulation runs, is quite promising for optimizing large problems relatively fast. A small numerical example tests how this simulation and optimization algorithm may be used to optimize lock capacities and implementation times.

A hybrid genetic-immune algorithm with improved lifespan and elite antigen for flow-shop scheduling problems

In this paper, a hybrid genetic-immune algorithm (HGIA) is proposed to reduce the premature convergence problem in a genetic algorithm (GA) in solving permutation flow-shop scheduling problems. A co-evolutionary strategy is proposed for efficient combination of GA and an artificial immune system (AIS). First, the GA is adopted to generate antigens with better fitness, and then the population in the last generation is transformed into antibodies in AIS. A new formula for calculating the lifespan of each antibody is employed during the evolution processes. In addition, a new mechanism including T-cell and B-cell generation procedures is applied to produce different types of antibodies which will be merged together. The antibodies with longer lifespan will survive and enter the next generation. This co-evolutionary strategy is very effective since chromosomes and antibodies will be transformed and evolved dynamically. The intensive experimental results show the effectiveness of the HGIA approach. The hybrid algorithm can be further extended to solve different combinatorial problems.

Efficient tool path planning for 5-axis flank milling of ruled surfaces using ant colony system algorithms

Five-axis CNC flank milling has recently received much attention in industry. The flank milling operation is efficient in shaping ruled geometry, but introduces a challenging task in machining error control. Previous work proposed a dynamic-programming based scheme for generating optimal tool path by minimising the machining error. However, to compute the tool path takes a considerable amount of time. This paper presents a new scheme using meta-heuristics, ant colony system (ACS), for tool path planning in 5-axis flank milling, with a focus on improving the computation efficiency. The path planning problem is first formulated as mapping two boundary curves of a ruled surface. An ACS-based optimisation algorithm is then applied to search for the mapping that minimises the machining error. The solution is nearly as good as using dynamic programming but takes only half of the computational time. The results from machining experiment and 3D measurement validate the effectiveness of the proposed scheme.

Queuing network analysis for waterways with artificial neural networks

An Artificial Neural Network (ANN) model has been developed for analyzing traffic in an inland waterway network. The main purpose of this paper is to determine how well such a relatively fast model for analyzing a queuing network could substitute for far more expensive simulation. Its substitutability for simulation is judged by relative discrepancies in predicting tow delays between the ANN and simulation models. This model is developed by integrating five distinct ANN submodels that predict tow headway variances at (1) merge points, (2) branching (i.e., diverging) points, (3) lock exits, and (4) link outflow points (e.g., at ports, junctions, or lock entrances), plus (5) tow queuing delays at locks. Preliminary results are shown for those five submodels and for the integrated network analysis model. Eventually, such a network analyzer should be useful for designing, selecting, sequencing, and scheduling lock improvement projects, for controlling lock operations, for system maintenance planning, and for other applications where many combinations of network characteristics must be evaluated. More generally, this method of decomposing complex queuing networks into elements that can be analyzed with ANNs and then recombined provides a promising approach for analyzing other queuing networks (e.g., in transportation, communication, computing, and production systems).

Combination of Multiple Ant Colony System and Simulated Annealing for the Multidepot Vehicle-Routing Problem with Time Windows

The vehicle-routing problem (VRP) is an important management problem in the field of physical distribution and logistics. In practice, the logistics system usually includes more than one depot, and the start of the service at each customer must be within a given time window. Hence, the multidepot vehicle-routing problem with time windows (MDVRPTW) is an important variant of the VRP. The MDVRPTW is a difficult combinatorial optimization problem due to the many complex constraints involved. The research presented in this paper proposes a multiple ant colony system (MACS) to solve the problem. In addition, two hybrid algorithms, which combine the strengths of MACS and simulated annealing, are developed to improve solution quality. The performance of the proposed algorithms is tested on several benchmark instances and compared with that of other algorithms in the literature. The results indicate that the proposed algorithms are effective in solving the MDVRPTW, and six new best solutions are found.

A fuzzy neural network for the flow time estimation in a semiconductor manufacturing factory

Flow time assignment problem creates a great challenge to semiconductor manufacturing managers especially when a company is facing the competitive pressure from customers requirements of quick response, on-time delivery and low cost. This paper presents a Neural-Fuzzy model for the flow time estimation using simulated data generated from a Foundry Service company located in Hsin-Chu science-based park of Taiwan. This Neural-Fuzzy model applies influential factors identified from the shop floor, i.e. order processing time, total work-in-process, and total jobs in system and utilization of bottleneck machines, to estimate the flow time of a new order. The fuzzy neural network is trained using a back-propagation algorithm to adjust the weight coefficients of the network and the parameters of the fuzzy membership functions. The trained network is then adopted to predict the flow time of each order generated from the simulated data. Model evaluation results for a simulated factory indicate that the Neural-Fuzzy model performs better than the Case-Based Reasoning method and a Multi-Layer Perceptrons Neural Network.

Convexity and semicontinuity of fuzzy sets

In this paper we study a modified concept called fuzzy convexity which was proposed by Ammar and Metz (Fuzzy Sets and Systems 49 (1992) 135). A criteria for convex fuzzy sets under lower semicontinuity is given. We prove in the upper semicontinuous case, that the class of semistrictly quasiconvex fuzzy sets lies between the quasiconvex and strictly quasiconvex classes. We also prove for both families of semistrictly quasiconvex and strictly quasiconvex fuzzy sets, that every local maximizer is also a global one. In addition, a characterization of quasimonotonic fuzzy sets in terms of level sets is given.

Developing a varietal GA with ESMA strategy for solving the pick and place problem in printed circuit board assembly line

The main issue for enhancing the productivity in Printed Circuit Board (PCB) is to reduce the cycle time for pick and place (PAP) operations; i.e., to minimize the time for the PAP operations. According to the characteristics of the PAP problems, the sequence for the placement of components can be mostly treated as the Travelling Salesman Problem (TSP). In this paper, a Genetic Algorithm (GA) with External Self-evolving Multiple Archives (ESMA) is developed for minimizing the PAP operations in PCB assembly line. ESMA focuses on the issue of improving the premature convergence time in GA by adopting efficient measures for population diversity, effective diversity control and mutation strategies to enhance the global searching ability. Three mechanisms for varietal GA such as Clustering Strategy, Switchable Mutation and Elitist Propagation have been designed based on the concept of increasing the dynamic diversity of the population. The experimental results in PCB and TSP instances show that the proposed approach is very promising and it contains the ability of local and global searching. The experimental results show ESMA can further improve the performance of GA by searching the solution space with more promising results.