Dissertations / Theses on the topic 'Trajectory search'

Thesis (M.S.)--Massachusetts Institute of Technology, Dept. of Aeronautics and Astronautics, 1990.
Title as it appears in the M.I.T. Graduate List, June, 1990: An application of the A* search technique to trajectory optimization.
Includes bibliographical references (leaves 87-88).
by Craig K. Niiya.
M.S.

Les propriétés de premier passage en général, et parmi elles le temps moyen de premier passage (MFPT), sont fréquemment utilisées dans les processus limités par la diffusion. Les processus réels de diffusion ne sont pas toujours Browniens : durant les dernières années, les comportements non-Browniens ont été observés dans un nombre toujours croissant de systèmes. Les milieux biologiques sont un exemple frappant où ce genre ce comportement a été observé de façon répétée. Nous présentons dans ce manuscrit une méthode basée sur les propriétés de premier passage permettant d'obtenir des informations sur le processus réel de diffusion, ainsi que sur l'environnement où évolue le marcheur aléatoire. Cette méthode permet de distinguer trois causes possibles de sous-diffusion : les marches aléatoires en temps continu, la diffusion en milieu fractal et le mouvement brownien fractionnaire. Nous étudions également l'efficacité des processus de recherche sur des réseaux discrets. Nous montrons comment obtenir les propriétés de premier passage sur réseau afin d'optimiser ensuite le processus de recherche, et obtenons un encadrement général du temps moyen de premier passage global (GMFPT). Grâce à ces résultats, nous estimons l'impact sur l'efficacité de recherche de plusieurs paramtres, notamment la connectivité de la cible, la mobilité de la cible ou la topologie du réseau.

Made available in DSpace on 2019-03-30T00:01:20Z (GMT). No. of bitstreams: 0 Previous issue date: 2017-09-22
Some software systems must respond to thousands or millions of concurrent requests. These systems must be properly tested to ensure that they can function correctly under the expected load. Performance degradation and consequent system failures usually arise in stressed conditions. Stress testing subjects the program to heavy loads. Stress tests di¿er from other kinds of testing in that the system is executed on its breakpoints, forcing the application or the supporting infrastructure to fail. The search for the longest execution time is seen as a discontinuous, nonlinear, optimization problem, with the input domain of the system under test as a search space. In this context, search-based testing is viewed as a promising approach to verify timing constraints. Search-based software testing is the application of metaheuristic search techniques to generate software tests. The test adequacy criterion is transformed into a ¿tness function and a set of solutions in the search space is evaluated with respect to the ¿tness functionusingametaheuristic. Search-basedstresstestinginvolves¿ndingthebest-andworst-case executiontimestoascertainwhethertimingconstraintsareful¿lled. ServiceLevelAgreements (SLAs) are documents that specify realistic performance guarantees as well as penalties for non-compliance. SLAsaremadebetweenprovidersandcustomersthatincludeservicequality, resourcescapability,scalability,obligations,andconsequencesincaseofviolations. Satisfying SLAisofgreatimportanceandachallengingissue. Themainmotivationofthisthesisisto¿nd theadequateresponsetimeofSLAsusingStressTesting. Thisthesisaddressesthreeapproaches insearch-basedstresstests. First,HybridmetaheuristicusesTabuSearch,SimulatedAnnealing, andGeneticAlgorithmsinacollaborativemanner. Second,anapproachcalledHybridQusesa reinforcementlearningtechniquetooptimizethechoiceofneighboringsolutionstoexplore,reducingthetimeneededtoobtainthescenarioswiththelongestresponsetimeintheapplication. The best solutions found by HybridQ were on average 5.98% better that achieved by the Hybrid approach without Q-learning. Third, the thesis investigates the use of the multi-objective NSGA-II,SPEA2,PAESandMOEA/Dalgorithms. MOEA/Dmetaheuristicsobtainedthebest hypervolume value when compared with other approaches. The collaborative approach using MOEA/D and HybridQ improves the hypervolume values obtained and found more relevant workloadsthanthepreviousexperiments. AtoolnamedIAdapter,aJMeterpluginforperformingsearch-basedstresstests,wasdevelopedandusedtoconductalltheexperiments. Keywords: Search-based Testing, Stress Testing, Multi-objective metaheuristics, Hybrid metaheuristics,ReinforcementLearning.
Alguns sistemas de software devem responder a milhares ou milhões de requisições simultâneos. Tais sistemas devem ser devidamente testados para garantir que eles possam funcionar corretamente sob uma carga esperada. Normalmente, a degradação do desempenho e consequentes falhas do sistema geralmente ocorrem em condições de estresse. No teste de estresse o sistema é submetido a cargas de trabalho acima dos resquistos não funcionais estabelecidos. Os testes de estresse diferem de outros tipos de testes em que o sistema é executado em seus pontos de interrupção, forçando o aplicativo ou a infra-estrutura de suporte a falhar. Testes de estresse podem ser vistos como um problema de otimização descontínuo, não-linear, comodomínio de entrada do sistema em test ecomo espaço de busca. Neste contexto,ostestes baseados em busca (search-based tests) são vistos como uma abordagem promissora para veri¿car as restrições de tempo. O teste de software baseado em busca é a aplicação de técnicas de pesquisa metaheurística para gerar testes de software. O critério de adequação do teste é transformado em uma função objetivo e um conjunto de soluções no espaço de busca é avaliado em relação à função objetivo usando uma metaheurística. Otestedeestressebaseadoembusca envolve encontrar os tempos de execução melhores e piores para veri¿car se as restrições de tempo são cumpridas. Os acordos de nível de serviço (SLA) são documentos que especi¿cam garantias de desempenho realistas, bem como penalidades por incumprimento. Os SLAs são feitos entre provedores e clientes que incluem qualidade do serviço, capacidade de recursos, escalabilidade, obrigações e consequencias em caso de violação. Satisfazer o SLA é de grande importância e um problema desa¿ador. A principal motivação desta tese é encontrar o tempo de resposta adequado dos SLAs usando teste de estresse. Esta tese apresenta três abordagens em testes de estresse baseados em busca. Primeiro, a metaheurística híbrida usa Tabu Search, Simulated Annealing e Algoritmos Genéticos de forma colaborativa. Em segundo lugar, uma abordagem chamada HybridQ usa uma técnica de aprendizado de reforço para otimizar a escolha de soluções vizinhas para explorar, reduzindo o tempo necessário para obter os cenários com o tempo de resposta mais longo na aplicação. As melhores soluções encontradas pelo HybridQ foram em média 5,98 % melhores que alcançadas pela abordagem híbrida sem Qlearning. Em terceiro lugar, a tese investiga o uso dos algoritmos multi-objetivos NSGA-II, SPEA2, PAES e MOEA/D. A metaheurística MOEA/D obteve o melhor valor de hipervolume quando comparada com outras abordagens. A abordagem colaborativa usand oMOEA/DeHybridQ melhora os valores de hipervolume obtidos e encontrou workloads mais relevantes do que as experiências anteriores. Uma ferramenta chamada IAdapter, um plugin JMeter para realizar testes de esforço baseados em busca, foi desenvolvida e usada para realizar todas as experiências. Palavras-chave: Search-basedTesting,StressTesting,Multi-objective metaheuristics,Hybridmetaheuristics,ReinforcementLearning

Since the discovery of Ceres, asteroids have been of immense scientific interest and intrigue. They hold answers to many of the fundamental questionsabout the formation and evolution of the Solar System. Therefore, a missionsurveying the asteroid belt with close encounter of carefully chosen asteroidswould be of immense scientific benefit. The trajectory of such an asteroidtour mission needs to be designed such that asteroids of a wide range ofcompositions and sizes are encountered; all with an extremely limited ∆Vbudget.This thesis presents a novel heuristic algorithm to optimize trajectoriesfor an asteroid tour mission with close range flybys (≤ 1000 km). The coresearch algorithm efficiently decouples combinatorial (i.e. choosing the asteroids to flyby)and continuous optimization (i.e. optimizing critical maneuversand events) of what is essentially a mixed integer programming problem.Additionally, different methods to generate a healthy initial population forthe combinatorial optimization are presented.The algorithm is used to generate a set of 1800 feasible trajectories withina 2029+ launch frame. A statistical analysis of these set of trajectories isperformed and important metrics for the search are set based on the statistics.Trajectories allowing flybys to prominent families of asteroids like Flora andNysa with ∆V as low as 4.99 km/s are obtained.Two modified implementations of the algorithm are presented. In a firstiteration, a large sample of trajectories is generated with a limited numberof encounters to the most scientifically interesting targets. While, a posteriori, trajectories are filled in with as many small targets as possible. Thisis achieved in two different ways, namely single step extension and multiplestep extension. The former fills in the trajectories with small targets in onestep, while the latter optimizes the trajectory by filling in with one asteroid per step. The thesis also presents detection of asteroids for successfullyperforming flybys. A photometric filter is developed which prunes out badlyilluminated asteroids. The best trajectory is found to perform well againstthis filter such that nine out of the ten planned flybys are feasible.

Trajectory planning is an important component of autonomous driving. It takes the result of route-level navigation plan and generates the motion-level commands that steer an autonomous passenger vehicle (APV). Prior work on solving this problem uses either a sampling-based or optimization-based trajectory planner, accompanied by some high-level rule generation components.

Thesis: M. Eng., Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science, 2019
Cataloged from PDF version of thesis.
Includes bibliographical references (pages 75-76).
Trajectory optimization is a powerful tool for determining good control sequences for actuating dynamical systems. In the past decade, trajectory optimization has been successfully used to train and guide policy search within deep neural networks via optimizing over many trajectories simultaneously, subject to a shared neural network policy constraint. This thesis seeks to understand how this specific formulation converges in comparison to known globally optimal policies for simple classical control systems. To do so, results from three lines of experimentation are presented. First, trajectory optimization control solutions are compared against globally optimal policies determined via value iteration on simple control tasks. Second, three systems built for parallelized, non-convex optimization across trajectories with a shared neural network constraint are described and analyzed. Finally, techniques from deep learning known to improve convergence speed and quality in non-convex optimization are studied when applied to both the shared neural networks and the trajectories used to train them.
by Robert Verkuil.
M. Eng.
M.Eng. Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science

This thesis develops a systematic framework to identify sub-paths using trajectories, which includes data preparation, raw sub-path generation, progression ratio evaluation and commonality factor-based final sub-path extraction. The results of the empirical study based on real traffic data in Brisbane Australia demonstrate the validity of the methodology, which has significant value for the identification of key corridors.

Applying reinforcement learning to control systems enables the use of machine learning to develop elegant and efficient control laws. Coupled with the representational power of neural networks, reinforcement learning algorithms can learn complex policies that can be difficult to emulate using traditional control system design approaches. In this thesis, three different model-free reinforcement learning algorithms, including Monte Carlo Control, REINFORCE with baseline, and Guided Policy Search are compared in simulated, continuous action-space environments. The results show that the Guided Policy Search algorithm is able to learn a desired control policy much faster than the other algorithms. In the inverted pendulum system, it learns an effective policy up to three times faster than the other algorithms. In the cartpole system, it learns an effective policy up to nearly fifteen times faster than the other algorithms.

With devastating natural disasters on the rise, technological improvements are needed in the field of search and rescue (SAR). Unmanned aerial vehicles (UAVs) would be ideal for the search function such that manned vehicles can be prioritized to distributing first-aid and ultimately saving lives. One of the major reasons that UAVs are under utilized in SAR is that they lack a long flight endurance which compromises their effectiveness. Dirigibles are well suited for SAR missions since they can hover and maintain lift without consuming energy and can be easily deflated for packaging and transportation. This research focuses on extending flight endurance of small-scale airship UAVs through improvements to the infrastructure design and flight trajectory planning. In the first area, airship design methodologies are reviewed leading to the development and experimental testing two hybrid fuel-electric power plants. The prevailing hybrid power plant design consists of a 4-stroke 14cc gasoline engine in-line with a brushless DC motor/generator and variable pitch propeller. The results show that this design can produce enough mechanical and electrical power to support 72 hours of flight compared to 1-4 hours typical of purely electric designs. A power plant configuration comparison method was also developed to compare its performance and endurance to other power plant configurations that could be used in dirigible UAVs. Overall, the proposed hybrid power plant has a 600% increase in energy density over that of a purely electric configuration. In the second area, a comprehensive multi-objective cost function is developed using spatially variable wind vector fields generated from computational fluid dynamic analysis on digital elevations maps. The cost function is optimized for time, energy and collision avoidance using a wavefront expansion approach to produce feasible trajectories that obey the differential constraints of the airship platform. The simulated trajectories including 1) variable vehicle velocity, 2) variable wind vector field (WVF) data, and 3) high grid resolutions were found to consume 50% less energy on average compared to planned trajectories not considering one of these three characteristics. In its entirety, this research addresses current UAV flight endurance limitations and provides a novel UAV solution to SAR surveillance.

L'objet de cette thèse est l'étude et la mise en oeuvre d'un système de gestion automatique de la configuration d'une formation d'avions sans pilote, ou drones. Les objectifs sont, d'une part, d'améliorer la sécurité et l'efficacité d'un groupe de drones de combat, et, d'autre part, de faire le lien entre les niveaux de planification de missions et les niveaux fonctionnels de contrôle de la formation. Le vol en formation est particulièrement bien adapté pour des applications militaires en milieux hostiles, qui requièrent des synchronisations pour l'arrivée sur les cibles ou du support mutuel pour le brouillage. L'une des difficultés soulevées est le choix autonome de la configuration. Notre approche est de mettre en oeuvre, entre les niveaux décisionnels et les niveaux fonctionnels, une couche intermédiaire dédiée à la formation et à la gestion autonome de sa configuration. La configuration ainsi déterminée doit être affectée aux avions de la formation en tenant compte des contraintes tactiques et des ressources de chacun. Enfin, la sécurité du vol est un élément primordial. Il faut donc pouvoir planifier des manoeuvres de reconfiguration pour passer d'une configuration à une autre, en respectant les distances minimales entres avions. Des solutions ont été développées à partir de l'algorithme Branch & Bound pour résoudre les problèmes d'allocations, et de l'algorithme A* pour la planification de trajectoires dans la formation. De plus, un contrôle de vol de la formation a été implémenté. Ceci a permis de valider notre approche par des simulations et des expérimentations réelles.

碩士
國立交通大學
多媒體工程研究所
97
In this thesis, we propose a tracking method that only employs bounding boxes of detected objects without using color, texture and contour information. With a simple overlap analysis of its bounding boxes in successive images, we can track an object and construct its trajectory very quickly. Using these trajectories along with their associated attributes obtained from object tracking, the trajectory-based indexes of a surveillance video database can be constructed. During the object tracking, undesirable size change of bounding boxes may occur. A novel size analysis method is developed to resize such bounding boxes to improve the results of overlap analysis. Besides, we can also employ the proposed size analysis of bounding boxes to help the separation of occluded objects. Finally, a search interface is provided, so that a user can search a target by specifying a direction or sketching a trajectory. By using the trajectory-based indexes, the search interface will only return video segments with proper trajectory attributes to users, making the search effective and very efficient.

Trajectory sharing and searching have received significant attention in recent years. In this thesis, we propose and investigate the methods to find and recommend the best trajectory to the traveler, and mainly focus on a novel technique named User Oriented Trajectory Search (UOTS) query processing. In contrast to conventional trajectory search by locations (spatial domain only), we consider both spatial and textual domains in the new UOTS query. Given a trajectory data set, the query input contains a set of intended places given by the traveler and a set of textual attributes describing the traveler’s preference. If a trajectory is connecting/close to the specified query locations, and the textual attributes of the trajectory are similar to the traveler’s preference, it will be recommended to the traveler. This type of queries can enable many popular applications such as trip planning and recommendation. There are two challenges in UOTS query processing, (i) how to constrain the searching range in two domains and (ii) how to schedule multiple query sources effectively. To overcome the challenges and answer the UOTS query efficiently, a novel collaborative searching approach is developed. Conceptually, the UOTS query processing is conducted in the spatial and textual domains alternately. A pair of upper and lower bounds are devised to constrain the searching range in two domains. In the meantime, a heuristic searching strategy based on priority ranking is adopted for scheduling the multiple query sources, which can further reduce the searching range and enhance the query efficiency notably. Furthermore, the devised collaborative searching approach can be extended to situations where the query locations are ordered. Extensive experiments are conducted on both real and synthetic trajectory data in road networks. Our approach is verified to be effective in reducing both CPU time and disk I/O time.

碩士
國立中興大學
資訊科學與工程學系所
97
The Multiple Trajectory Search (MTS) for solving the maximum clique problem is proposed in this thesis. The maximum clique problem is a famous NP-hard problem in the graph theory and has attracted much attention of researches for a very long period. Since this problem is NP-hard, exact algorithms in general can solve only small-sized instances. In recent years, metaheuristics were proposed to find optimal or near-optimal solutions of this problem. The multiple trajectory search is also a metaheuristic method. It is constituted by an upper level search and a lower level search. The upper level search tries to search a large neighborhood of a solution and this search of large neighborhood is based on the lower level search that searches a small neighborhood. The experimental results of applying the multiple trajectory search to the eighty problems presented by DIMACS were reported. The performance of the multiple trajectory search for the maximum clique problem is good except the brock and C benchmarks.

碩士
國立交通大學
資訊科學與工程研究所
97
In recent years, GPS-equipped devices are widely used in daily life. People can record their trajectories and share them on the Internet. As such data increases dramatically, several issues on trajectories have arisen. However, exploring meaningful trajectories from a trajectory database is important. In the traditional trajectory search, users can query a range to derive all trajectories in a trajectory database passing the region, or query a trajectory to derive trajectories in a trajectories database with higher similar- ities. In this paper, we intent to rank trajectories in a trajectory database and search top-k meaningful trajectories from the database with a given query range. We introduce a framework of pattern-aware trajectory search. In this framework, we propose a PageRank-like approach to discover implicit patterns from a trajectory database, and a pruning approach to improve the eciency for searching pattern-aware trajectories from a large database. We collect shared trajectories in Taiwan from several travel websites to evaluate the performance of our proposed approaches. Experimental results show effectiveness and eciency of our proposed approaches.

A spacecraft tour trajectory is a trajectory in which a spacecraft visits a number of objects in sequence. The target objects may consist of satellites, moons, planets or any other body in orbit, and the spacecraft may visit these in a variety of ways, for example flying by or rendezvousing with them. The key characteristic is the target object sequence which can be represented as a discrete set of decisions that must be made along the trajectory. When this sequence is free to be chosen, the result is a hybrid discrete-continuous optimization problem that combines the challenges of discrete and combinatorial optimization with continuous optimization. The problem can be viewed as a generalization of the traveling salesman problem; such problems are NP-hard and their computational complexity grows exponentially with the problem size. The focus of this dissertation is the development of a novel methodology for the solution of spacecraft tour trajectory optimization problems. A general model for spacecraft tour trajectories is first developed which defines the parameterization and decision variables for use in the rest of the work. A global search methodology based on the tabu search metaheuristic is then developed. The tabu search approach is extended to operate on a tree-based solution representation and neighborhood structure, which is shown to be especially efficient for problems with expensive solution evaluations. Concepts of tabu search including recency-based tabu memory and strategic intensification and diversification are then applied to ensure a diverse exploration of the search space. The result is an automated, adaptive and efficient search algorithm for spacecraft tour trajectory optimization problems. The algorithm is deterministic, and results in a diverse population of feasible solutions upon termination. A novel numerical search space pruning approach is then developed, based on computing upper bounds to the reachable domain of the spacecraft, to accelerate the search. Finally, the overall methodology is applied to the fourth annual Global Trajectory Optimization Competition (GTOC4), resulting in previously unknown solutions to the problem, including one exceeding the best known in the literature.
text

碩士
國立中興大學
資訊科學與工程學系所
99
The satisfiability problem is one of famous NP-complete problem in computer science field, with applications in artificial intelligence, machine learning and VLSI design and test。 The satisfiability problem is a problem in discrete domain. In this thesis, we proposed a method to convert this problem into real domain (continuous domain), and applied the multiple trajectory search to solve this problem. We compared the performance of the proposed algorithm with that of other algorithms published in the literature on the benchmarks presented by Gottlieb, Marchiori and Rossi in 2002.

碩士
國立中興大學
資訊科學與工程學系所
100
The permutation flowshop scheduling problem is one of the core NP-hard problems with many applications filed such as, the industry production line and the software development. In this thesis, we propose a hybrid genetic local search algorithm to solve the multi-objective flowshop scheduling problem. The objectives are considered, namely, the makespan and the total flow time. Observing the weak points with some algorithms published in the literature, we developed our algorithm with two characteristics:1.Adding the orthogonal-array crossover operator to enhance the intensification. 2.We use two-layer genetic algorithm to search the solution space more thoroughly. 3.Maintaining diversification. We compare the proposed algorithm with other algorithms in the literature published on the benchmarks offered by Taillard’s 1993. The quality of solutions obtained by the proposed algorithm is competitive with other algorithms, whereas the computation time needed is more than that of other algorithms.

碩士
國立中興大學
資訊科學與工程學系所
98
The quadratic assignment problem (QAP) is one of the NP-hard problems. It is a well-known combinatorial optimization problem. There are many practical problems that can be formulated as the QAP, for example, problems dealing with typewriter keyboard design, school and hospital layout design and scheduling. In this thesis, we propose a new method called the multiple trajectory search(MTS) algorithm to solve the quadratic assignment problem. Observing the weak points of some algorithms published in the literature, we developed our algorithm with following three characteristics: 1.We used a new architecture to search more thoroughly and efficiently the solution space; 2.We proposed medium-scale region search to improve the search; 3.We used a variety of local search methods to achieve a more detailed search. We compared the proposed algorithm with other algorithms published in the literature on the benchmarks offered by Tseng and Liang in 2006, and James, Rego and Glover in 2009.

碩士
國立中興大學
資訊科學與工程學系所
98
The Sequential Ordering Problem (SOP) is one of the NP-hard problems. It is a well-known combinatorial optimization problem. There are many practical problems that can be formulated as the SOP, for example, the transportation problem and the production planning problem. In this thesis, we use the Multiple Trajectory Search (MTS) Algorithm to solve the sequential ordering problem. Observing the weak points of some algorithms published in the literature, we developed our algorithm with following three characteristics: 1.We use genetic local search algorithm to generate initial solutions; 2. We proposed the region search as the second tier of the MTS to enhance the diversification; 3. We make search more effective by adding restart conditions to the local search scheme to abandon the solutions near by an already searched local optimum. We compared the proposed algorithm with other algorithms published in the literature on the benchmarks provided by the TSPLIB.

碩士
國立中興大學
資訊網路多媒體研究所
98
The open shop scheduling problem is an NP-hard combinatorial optimization problem and has long attracted much attention. In this thesis, we propose a new metaheuristic called the multiple trajectory search to solve the open shop scheduling problem. Observing the weak points of some algorithms published in the literature, we developed our algorithm with following three characteristics: 1.We use a new architecture to search more thoroughly and efficiently the solution space, 2.We propose the region search for searching a medium-scale neighborhood, and 3.We use different local search methods to achieve a more detailed search. We compared the proposed algorithm with other algorithms published in the literature on the benchmarks offered by Brucker、Gueret and Taillard. The performance of applying the multiple trajectory search is good except for the Taillard benchmark. On Gueret’s instance, the quality of solutions found is good and our method had improved the best known solution of one instance.

碩士
國立中興大學
資訊科學與工程學系所
100
The vehicle routing problem with time windows is an NP-hard problem. When the scale of the problem is large, it is very difficult to find the optimal solution within limited time. Therefore, meta-heuristic algorithms are good candidate for solving this problem. This problem has applications in logistics and transportation. Also it has the characteristic of the multiple objectives. In this thesis, we propose a framework that hybridizes genetic algorithm, local search algorithms and multiple trajectory search to solve the vehicle routing problem. We aim to minimize the number of cars and the distance of all routes. The proposed algorithm operates in two phases. On the first phase, it searches the solution space globally and tries to find solutions that have minimum number of cars. Then, in the second phase, it searches more thoroughly the neighborhoods of the good solutions obtained in the first phase by applying several local search methods. Finally, the proposed algorithm was applied to solve Solomon’s 56 VRPTW 100-customers instances. The experimental results were compared with the results of previously published studies. The comparison shows that our approach has very good performance in some of the instances.

博士
國立中興大學
資訊科學與工程學系
102
Real-parameter optimization is often applied in science, electrical engineering, computer science and other fields. Real-parameter optimization can be divided into two categories: single objective real-parameter optimization and multi-objective real parameter optimization. Recently, the research on multi-objective real-parameter optimization is very active. There are many papers published on this topic. In this dissertation, a new metaheuristic method called the multiple trajectory search is proposed for real-parameter optimization. It can solve constrained single objective real-parameter optimization problems, unconstrained single objective real-parameter optimization problems, constrained multi-objective real-parameter optimization problems and unconstrained multi-objective real-parameter optimization problems. The benchmark problems proposed in the competitions held in the Congress on Evolutionary Computation 2008, 2009 and 2010 are used to test the performance of the proposed method. Most of the metaheuristic methods do not search the solution space systematically due to their randomization characteristic. Also, most of the metaheuristic methods do not use different region search methods that are adapted to the landscape of the neighborhood of a solution. The multiple trajectory search tries to overcome these two drawbacks. Initially, it will generate solutions uniformly distributed over the solution space. When it searches the neighborhood of a solution, it will choose one of the region search methods that best fits the landscape of the neighborhood. Furthermore, at the end of each search session, solutions will be evaluated and better solutions will be put in the foreground while other solutions put in the background. And only the solutions in the foreground will be searched in the next search session, so the searching can be focused on the promising area. Revealed by the experimental results, the proposed multiple trajectory search is very competitive.

碩士
國立中興大學
資訊網路多媒體研究所
98
In this paper, we use the Multiple Trajectory Search (MTS) Algorithm to solve the Resource-Constrained Project Scheduling Problem (RCPSP). Observing the weak points of some algorithms published in the literature, we developed our algorithm with following three characteristics: 1. We propose the concept of topN, topN can be combined with the genetic local search algorithm to generate initial solution. This approach can also be used in solving other problems. 2. We proposed the region search, as the second tier of the MTS, to enhance the diversification. 3. We find out the shortcoming of the Forward Backward Improvement (FBI), and we used an iterative approach to enhance the search capabilities, and we also designed a mutation operator to improve the diversification ability. We compared the proposed algorithm with other algorithms published in the literature on the benchmarks offered by Kolisch and Sprecher in 1997.

博士
國立中興大學
應用數學系所
98
Prediction of the protein structure is one of the most important problems in the computational biology, and it remains one of the biggest challenges in the structural biology. Disulfide bonds play an import structural role in stabilizing protein conformations. For the protein-folding prediction, a correct prediction of disulfide bridges can greatly reduce the search space. The prediction of disulfide bonding pattern helps, to a certain degree, predicts the 3D structure of a protein and hence its function since disulfide bonds imposes geometrical constraints on the protein backbones. In this dissertation, we first used the position-specific scoring matrix (PSSM), normalized bond lengths, the predicted secondary structure of protein, and the physicochemical properties index of the amino acid as the features for designing the classifier based on the support vector machine (SVM). The classifier was trained to compute the connectivity probabilities of cysteine pairs. In addition, an evolutionary algorithm called the multiple trajectory search (MTS) was integrated with the SVM model to tune the parameters of the SVM and the window sizes for the features. The maximum weighted perfect matching algorithm was then used to find the disulfide connectivity pattern. In this study, the experimental results show that the accuracies rate reaches 79.8% for the prediction of the overall disulfide connectivity pattern (QP) and that of disulfide bridge prediction (QC) is 80.9% for dataset SP39. Without the prior knowledge of the bonding states of cysteines, the results show that the accuracies rate reaches 54.5% (QP) and 60% (QC), respectively. Then, the protein 3D structure related features called normalized pair distance (NPD) vector were imposed. From experiments, we obtained the good performance for four problems in disulfide bond prediction. With the prior knowledge of the bonding states of cysteines, the results show that the accuracies rate reaches 92.2% (QP) and 94.2% (QC) respectively for dataset SP39. Without the prior knowledge of the bonding states of cysteines, the results show that the accuracies rate reaches 84.4% (QP) and 94.6% (QC) respectively for dataset SPX.

碩士
國立中興大學
資訊網路多媒體研究所
100
Both the multi-objective 0-1 knapsack problem and the multi-objective unconstrained facility location problem are NP-Hard problems. The solutions to both problems can be represented by binary sequence. The knapsack problem is defined as follows. Given a set of items, each item has its own weight and value, We want to select a collection of items such that the total weight of these items is less than or equal to a given weight limit and the total value is as large as possible. The knapsack problem is a popular research topic in the algorithm flied with many applications such as the hardware design, the stock cutting, and the budget problem. The unconstrained facility location problem is defined as follows. Given a set of potential facility locations and a set of customers, We want to choose a set of facility locations to open such that the cost is minimized. It is a popular research topic in the business flied. In this thesis, we propose a new Multiple Trajectory Search(MTS) algorithm which is based on the genetic algorithm and apply this new algorithm to both problems. The proposed algorithm tries to search the solution space more systematically and more thoroughly. The proposed Multiple Trajectory Search was applied to the knapsack benchmark instances proposed by Eckart Zitzler and Lothar Thiele in 1999 and the result were compared with other state-of-the-art methods. Revealed by the experimental results, the MTS is very competitive with other algorithms on the multi-objective 0-1 knapsack problem.

碩士
國立中正大學
勞工所
97
A study on the behavior of job search about Yo-Yo Trajectory for college student－Taichung City as an Example Abstract In 2006 as a result of the International Labor Organization (ILO) published the "Global Employment Trends for Youth" report that global youth unemployment rate in 10 years increased from 12.3 percent to 13.5 percent, much higher than the adult unemployment rate of 4.6%. The youth share 25% of the global population, they constitute the world''s population about 43 percent unemployment; Western advanced countries since the early 80s began to appear the theme for discussion about youth unemployment. When 90s, with the labor deregulation and the development of atypical employment, the boundary between employment and unemployment has become increasingly blurred. The form of youth employment is also becoming more diversified. The combined effect of the two, the transition of youth result has a process of back and forth, become more diverse and complex, scholars called that Yo-Yo. But Taiwan Yo-Yo has a special form and development. Read the relevant literature in recent years has led to a motive to carry out this study. This study would like to know the relationship between the type of employment and the result of employment, during the interfere of the free degree of job-seekers’ network. And the relationship between the type of unemployment and the duration of unemployment , during the interfere of the free degree of job-seekers’ network. Quantitative data of this study was to adopt the "Taichung City government－ the demand for youth employment research program of 97" used by the incumbent of the questionnaire and the unemployed questionnaire. According to the classification of information is restricted to a sample of young working people and unemployed young people in Taichung , aged above 15 under 35 years old, have graduated from college identity. At present, the overall employment environment, the youth employment situation of facing an unprecedented test, in particular a high unemployment rate of highly educated for years. Youth labor market is facing severe challenges. We’re looking forward to a clear understanding of the situation of the high unemployment rate of highly educated Youth and a further analysis of Yo -Yo. Keyword: Yo-Yo, youth transition, the free degree of job-seekers’ network, job search behavior, the high unemployment rate of highly educated

Multiple gravity assist (MGA) spacecraft trajectories can be difficult to find, an intractable problem to solve completely. However, these trajectories have enormous benefits for missions to challenging destinations such as outer planets and primitive bodies. Techniques are presented to aid in solving this problem with a global search tool and additional investigation into one particular proximity operations option is discussed. Explore is a global grid-search MGA trajectory pathsolving tool. An efficient sequential tree search eliminates v∞ discontinuities and prunes trajectories. Performance indices may be applied to further prune the search, with multiple objectives handled by allowing these indices to change between trajectory segments and by pruning with a Pareto-optimality ranking. The MGA search is extended to include deep space maneuvers (DSM), v∞ leveraging transfers (VILT) and low-thrust (LT) transfers. In addition, rendezvous or nπ sequences can patch the transfers together, enabling automatic augmentation of the MGA sequence. Details of VILT segments and nπ sequences are presented: A boundaryvalue problem (BVP) VILT formulation using a one-dimensional root-solve enables inclusion of an efficient class of maneuvers with runtime comparable to solving ballistic transfers. Importantly, the BVP VILT also allows the calculation of velocity-aligned apsidal maneuvers (VAM), including inter-body transfers and orbit insertion maneuvers. A method for automated inclusion of nπ transfers such as resonant returns and back-flip trajectories is introduced: a BVP is posed on the v∞ sphere and solved with one or more nπ transfers – which may additionally fulfill specified science objectives. The nπ sequence BVP is implemented within the broader search, combining nπ and other transfers in the same trajectory. To aid proximity operations around small bodies, analytical methods are used to investigate stability regions in the presence of significant solar radiation pressure (SRP) and body oblateness perturbations. The interactions of these perturbations allow for heliotropic orbits, a stable family of low-altitude orbits investigated in detail. A novel constrained double-averaging technique analytically determines inclined heliotropic orbits. This type of knowledge is uniquely valuable for small body missions where SRP and irregular body shape are very important and where target selection is often a part of the mission design.