Dissertations / Theses on the topic 'Hybrid Evolution Algorithms'

La prise de décision est une partie intégrante de notre vie quotidienne où le décideur est confronté à des problèmes composés de plusieurs objectifs habituellement contradictoires. Dans ce travail, nous traitons des problèmes d'optimisation multiobjectif dans des espaces de recherche continus ou discrets. Nous avons développé plusieurs nouveaux algorithmes basés sur les métaheuristiques hybrides évolutionnaires, en particulier sur l'algorithme MOEA/D. Nous avons proposé l'algorithme HEMH qui utilise l'algorithme DM-GRASP pour construire une population initiale de solutions de bonne qualité dispersées le long de l'ensemble des solutions Pareto optimales. Les résultats expérimentaux montrent la supériorité de toutes les variantes hybrides proposées sur les algorithmes originaux MOEA/D et SPEA2. Malgré ces bons résultats, notre approche possède quelques limitations, levées dans une version améliorée de HEMH : HEMH2 et deux autres variantes HEMHde et HEMHpr. Le Adaptive Binary DE inclus dans les HEMH2 et HEMHde a de meilleures capacités d'exploration qui pallient aux capacités de recherche locale contenues dans la HEMH, HEMH2 et HEMHde. Motivés par ces résultats, nous avons proposé un nouvel algorithme baptisé HESSA pour explorer un espace continu de recherche où le processus de recherche est réalisé par différentes stratégies de recherche. Les résultats expérimentaux montrent la supériorité de HESSA à la fois sur MOEA/D et dMOPSO. Tous les algorithmes proposés ont été vérifiés, testé et comparés à certaines méthodes MOEAs. Les résultats expérimentaux montrent que toutes les propositions sont très compétitives et peuvent être considérés comme une alternative fiable
Many real-world decision making problems consist of several conflicting objectives, the solutions of which is called the Pareto-optimal set. Hybrid metaheuristics proved their efficiency in solving these problems. They tend to enhance search capabilities by incorporating different metaheuristics. Thus, we are concerned with developing new hybrid schemes by incorporating different strategies with exploiting the pros and avoiding the drawback of the original ones. First, HEMH is proposed in which the search process includes two phases DMGRASP obtains an initial set of efficient solutions in the 1st phase. Then, greedy randomized path-relinking with local search or reproduction operators explore the non-visited regions. The efficient solutions explored over the search are collected. Second, a comparative study is developed to study the hybridization of different metaheuristics with MOEA/D. The 1st proposal combines adaptive discrete differential Evolution with MOEA/D. The 2nd combines greedy path-relinking with MOEA/D. The 3rd and the 4th proposals combine both of them in MOEA/D. Third, an improved version of HEMH is presented. HEMH2 uses inverse greedy to build its initial population. Then, differential evolution and path-relink improves these solutions by investigating the non-visited regions in the search space. Also, Pareto adaptive epsilon concept controls the archiving process. Motivated by the obtained results, HESSA is proposed to solve continuous problems. It adopts a pool of search strategies, each of which has a specified success ratio. A new offspring is generated using a randomly selected one. Then, the success ratios are adapted according to the success of the generated offspring. The efficient solutions are collected to act as global guides. The proposed algorithms are verified against the state of the art MOEAs using a set of instances from literature. Results indicate that all proposals are competitive and represent viable alternatives

Técnicas de Agrupamento vêm obtendo bons resultados quando utilizados em diversos problemas de análise de dados, como, por exemplo, a análise de dados de expressão gênica. Porém, uma mesma técnica de agrupamento utilizada em um mesmo conjunto de dados pode resultar em diferentes formas de agrupar esses dados, devido aos possíveis agrupamentos iniciais ou à utilização de diferentes valores para seus parâmetros livres. Assim, a obtenção de um bom agrupamento pode ser visto como um processo de otimização. Esse processo procura escolher bons agrupamentos iniciais e encontrar o melhor conjunto de valores para os parâmetros livres. Por serem métodos de busca global, Algoritmos Genéticos podem ser utilizados durante esse processo de otimização. O objetivo desse projeto de pesquisa é investigar a utilização de Técnicas de Agrupamento em conjunto com Algoritmos Genéticos para aprimorar a qualidade dos grupos encontrados por algoritmos de agrupamento, principalmente o k-médias. Esta investigação será realizada utilizando como aplicação a análise de dados de expressão gênica. Essa dissertação de mestrado apresenta uma revisão bibliográfica sobre os temas abordados no projeto, a descrição da metodologia utilizada, seu desenvolvimento e uma análise dos resultados obtidos.
Clustering techniques have been obtaining good results when used in several data analysis problems, like, for example, gene expression data analysis. However, the same clustering technique used for the same data set can result in different ways of clustering the data, due to the possible initial clustering or the use of different values for the free parameters. Thus, the obtainment of a good clustering can be seen as an optimization process. This process tries to obtain good clustering by selecting the best values for the free parameters. For being global search methods, Genetic Algorithms have been successfully used during the optimization process. The goal of this research project is to investigate the use of clustering techniques together with Genetic Algorithms to improve the quality of the clusters found by clustering algorithms, mainly the k-means. This investigation was carried out using as application the analysis of gene expression data, a Bioinformatics problem. This dissertation presents a bibliographic review of the issues covered in the project, the description of the methodology followed, its development and an analysis of the results obtained.

The main objective of this research is to develop an innovative multi-fidelity multi-disciplinary design, analysis and optimization suite that integrates certain solution generation codes and newly developed innovative tools to improve the overall optimization process. The research performed herein is divided into two parts: (1) the development of an MDAO framework by integration of variable fidelity physics-based computational codes, and (2) enhancements to such a framework by incorporating innovative features extending its robustness.

The first part of this dissertation describes the development of a conceptual Multi-Fidelity Multi-Strategy and Multi-Disciplinary Design Optimization Environment (M3 DOE), in context of aircraft wing optimization. M3 DOE provides the user a capability to optimize configurations with a choice of (i) the level of fidelity desired, (ii) the use of a single-step or multi-step optimization strategy, and (iii) combination of a series of structural and aerodynamic analyses. The modularity of M3 DOE allows it to be a part of other inclusive optimization frameworks. The M3 DOE is demonstrated within the context of shape and sizing optimization of the wing of a Generic Business Jet aircraft. Two different optimization objectives, viz. dry weight minimization, and cruise range maximization are studied by conducting one low-fidelity and two high-fidelity optimization runs to demonstrate the application scope of M3 DOE.

The second part of this dissertation describes the development of an innovative hybrid optimization framework that extends the robustness of M3 DOE by employing a proper orthogonal decomposition-based design-space order reduction scheme combined with the evolutionary algorithm technique. The POD method of extracting dominant modes from an ensemble of candidate configurations is used for the design-space order reduction. The snapshot of candidate population is updated iteratively using evolutionary algorithm technique of fitness-driven retention. This strategy capitalizes on the advantages of evolutionary algorithm as well as POD-based reduced order modeling, while overcoming the shortcomings inherent with these techniques. When linked with M3 DOE, this strategy offers a computationally efficient methodology for problems with high level of complexity and a challenging design-space. This newly developed framework is demonstrated for its robustness on a non-conventional supersonic tailless air vehicle wing shape optimization problem.
Ph. D.

Submitted by Jaqueline Silva (jtas29@gmail.com) on 2014-09-26T21:10:33Z No. of bitstreams: 2 Caetano, Samuel Sabino-2013-dissertação.pdf: 1031464 bytes, checksum: 39f6a5947aed90b8f7c8e56b76d93e5a (MD5) license_rdf: 23148 bytes, checksum: 9da0b6dfac957114c6a7714714b86306 (MD5)
Approved for entry into archive by Jaqueline Silva (jtas29@gmail.com) on 2014-09-26T21:12:31Z (GMT) No. of bitstreams: 2 Caetano, Samuel Sabino-2013-dissertação.pdf: 1031464 bytes, checksum: 39f6a5947aed90b8f7c8e56b76d93e5a (MD5) license_rdf: 23148 bytes, checksum: 9da0b6dfac957114c6a7714714b86306 (MD5)
Made available in DSpace on 2014-09-26T21:12:31Z (GMT). No. of bitstreams: 2 Caetano, Samuel Sabino-2013-dissertação.pdf: 1031464 bytes, checksum: 39f6a5947aed90b8f7c8e56b76d93e5a (MD5) license_rdf: 23148 bytes, checksum: 9da0b6dfac957114c6a7714714b86306 (MD5) Previous issue date: 2013-09-09
Increasingly, learning in groups has become present in school environments. This fact is also part of the organizations, when considers learning in the workplace. Conscious of the importance of group learning at the workplace (CSCL@Work) emerges as an application area. In Computer Supported Collaborative Learning(CSCL), researchers have been struggling to maximize the performance of groups by techniques for forming groups. Is that why this study developed three (3) algorithmic approaches to formation of intraheterogeneous and inter-homogeneous groups, as well as a model proposed in this work in which integrates dichotomous functional characteristics and preferred roles. We made an algorithm that generates random groups, a Canonical Genetic Algorithm and Hybrid Genetic Algorithm. We obtained the input data of the algorithm by a survey conducted at the Court of the State of Goiás to identify dichotomous functional characteristics, and after we categorize these characteristics, based on the data found and the model proposed group formation. Starting at real data provided of employees whom participated in a course by Distance Education (EaD), we apply the model and we obtained the input data related to functional features. As regards the favorite roles, we assigned randomly values to the employees aforementioned, from a statistical statement made by Belbin into companies in the United Kingdom. Then, we executed the algorithms in three test cases, one considering the preferred papers and functional characteristics, while the other two separately considering each of these perspectives. Based on the results obtained, we found that the hybrid genetic algorithm outperforms the canonical genetic algorithm and random generator.
A aprendizagem em grupos tem se tornado realidade cada vez mais presente nos ambientes de ensino. Esta realidade também faz parte das organizações quando considera-se a aprendizagem no contexto do trabalho. Cientes da importância da aprendizagem em grupo no ambiente de trabalho, uma nova abordagem, denominada CSCL@Work, surge como uma aplicação da área Aprendizagem Colaborativa Apoiada pelo Computador, no inglês, Computer Supported Collaborative Learning (CSCL), no ambiente de trabalho. Em CSCL, pesquisadores tem se esforçado cada vez mais para maximizar o desempenho dos grupos através de técnicas para formação de grupos. Por isso neste trabalho desenvolvemos 3 (três) abordagens algorítmicas para formação de grupos intra-heterogêneos e inter-homogêneos, a partir de um modelo proposto nesta pesquisa, que integra características funcionais dicotômicas e papéis preferidos. Confeccionamos um algoritmo que gera grupos aleatoriamente, um algoritmo genético canônico e um algoritmo genético híbrido. Para obter os dados de entrada do algoritmo, realizamos uma pesquisa no Tribunal de Justiça do Estado de Goiás para identificar características funcionais dicotômicas, categorizamos estas características, com base nos dados encontrados e no modelo de formação de grupos proposto. A partir de dados reais fornecidos de funcionários que participaram de um curso por Educação a Distância (EaD), aplicamos o modelo e obtivemos os dados de entrada relativos às características funcionais. Quanto aos papéis preferidos, atribuímos os valores aleatoriamente aos funcionários mencionados, partindo de um levantamento estatístico feito por Belbin em empresas no Reino Unido. Em seguida, executamos os algoritmos em três casos de testes, um considerando as características funcionais e papéis preferidos, e os outros dois considerando separadamente cada uma destas perspectivas. A partir dos resultados obtidos, constatamos que o algoritmo genético híbrido obtém resultados superiores ao algoritmo genético canônico e método aleatório.

The effectiveness of an optimization algorithm can be reduced to its ability to navigate an objective function’s topology. Hybrid optimization algorithms combine various optimization algorithms using a single meta-heuristic so that the hybrid algorithm is more robust, computationally efficient, and/or accurate than the individual algorithms it is made of. This thesis proposes a novel meta-heuristic that uses search vectors to select the constituent algorithm that is appropriate for a given objective function. The hybrid is shown to perform competitively against several existing hybrid and non-hybrid optimization algorithms over a set of three hundred test cases. This thesis also proposes a general framework for evaluating the effectiveness of hybrid optimization algorithms. Finally, this thesis presents an improved Method of Characteristics Code with novel boundary conditions, which better characterizes pipelines than previous codes. This code is coupled with the hybrid optimization algorithm in order to optimize the operation of real-world piston pumps.

Neste trabalho foi realizada uma proposta de utilização de Sistemas de Inferência Nebulosos para controlar, em tempo de execução, parâmetros de Algoritmos Genéticos. Esta utilização busca melhorar o desempenho de Algoritmos Genéticos diminuindo, ao mesmo tempo: a média de iterações necessárias para que um Algoritmo Genético encontre o valor ótimo global procurado; bem como diminuindo o número de execuções do mesmo que não são capazes de encontrar o valor ótimo global procurado, nem mesmo para quantidades elevadas de iterações. Para isso, foram analisados os resultados de diversos experimentos com Algoritmos Genéticos, resolvendo instâncias dos problemas de Minimização de Funções e do Caixeiro Viajante, sob diferentes configurações de parâmetros. Com base nos resultados obtidos a partir destes experimentos, foi proposto um modelo com a troca de valores de parâmetros de Algoritmos Genéticos, em tempo de execução, pela utilização de Sistemas de Inferência Nebulosos, de forma a melhorar o desempenho do sistema, minimizando ambas as medidas citadas anteriormente.
This work addressed a proposal of the application of Fuzzy Systems to adjust parameters of Genetic Algorithms, during execution time. This application attempts to improve the performance of Genetic Algorithms by diminishing, at the same time: the average number of necessary generations for a Genetic Algorithm to find the desired global optimum value, as well as diminishing the number of executions of a Genetic Algorithm that are not capable of finding the desired global optimum value even for high numbers of generations. For that purpose, the results of many experiments with Genetic Algorithms were analyzed; addressing instances of the Function Minimization and the Travelling Salesman problems, under different parameter configurations. With the results obtained from these experiments, a model was proposed, for the exchange of parameter values of Genetic Algorithms, in execution time, by using Fuzzy Systems, in order to improve the performance of the system, minimizing both of the measures previously cited.

博士
國立臺灣科技大學
營建工程系
103
Construction management everywhere faces problems and challenges. Resource scheduling is a crucial part of project planning of any management companies. Successful tradeoff optimization resource scheduling problems within the project scope is necessary to maximize overall company benefits. This study investigated the potential use of various advanced techniques to improve multiple objective Differential Evolution. Three hybrid multiple objective Differential Evolution (MODE) algorithms that integrate chaotic maps, opposition-based learning technique and Artificial Bee Colony are introduced to solve the resource scheduling problems. Firstly, chaotic initialized adaptive multiple objective Differential Evolution (CAMODE) model is presented. CAMODE utilizes the advantages of chaotic sequences for generating an initial population and an external elitist archive to store non-dominated solutions found during the evolutionary process. In order to maintain the exploration and exploitation capabilities during various phases of optimization process, an adaptive mutation operation is introduced. Secondly, opposition-based Multiple Objective Differential Evolution (OMODE) model is presented. OMODE employs an opposition-based learning technique for population initialization and for generation jumping. Opposition numbers are used to improve the exploration and convergence performance of the optimization process. Finally, a new hybrid multiple-objective artificial bee colony with differential evolution (MOABCDE) model is proposed. The proposed algorithm integrates crossover operations from differential evolution (DE) with the original artificial bee colony (ABC) in order to balance the exploration and exploitation phases of the optimization process. Numerous real construction case studies including time-cost-quality tradeoff, time-cost tradeoffs in resource-constrained, time-cost-labor utilization tradeoff and time-cost-environment impact tradeoff problems are used to demonstrate the proposed models. The proposed models are validated by comparing with current widely used multiple objective algorithms, including the non-dominated sorting genetic algorithm (NSGA-II), the multiple objective particle swarm optimization (MOPSO), the multiple objective differential evolution (MODE), and the multiple objective artificial bee colony (MOABC) and previous works via comparison indicators and hypothesis test. Experimental results obtained from the proposed models confirm that using the newly established models can be a highly beneficial for decision-makers when solving various problems in the field of construction management.

My original contribution to knowledge is the successful application of two modelling paradigms 1) SALMO-PLUS process based model and 2) HEA data driven model to tropical lakes of different morphometry and trophic status. The application of SALMO-PLUS to tropical lakes involves utilising the SALMO-OO model structure for optimisation. This was followed by multi objective parameter optimisation on selected parameters to seek the optimum parameter values that can model the algal dynamics and state variables fluctuations in the tropical lakes to an acceptable magnitude and peaks. SALMO-PLUS is another version SALMO-OO with capability to run optimisation by means of particle swarm optimisation (PSO) method. SALMO-OO has been used as a research tool over a number of lakes with different trophic states and mixing conditions to simulate algal succession and respond to ecosystem dynamic. SALMO-OO is driven by process-based differential equations and works by utilizing a library of three phytoplankton growth and three grazing process models. Evolutionary algorithms (EA) are bio-inspired adaptive methods which mimic processes of biological evolution, natural selection and genetic variation such as cross-over and mutation to develop solutions to complex computational problems (Recknagel et al, 2006). HEA is designed for rule discovery in water quality timeseries (Cao et al., 2006b) and is capable of forecasting potential algal population dynamics and outbreaks in water bodies. The SALMO-PLUS model was applied for simulating the state variables of selected lakes (Lake Kenyir, Lake Penang, Saidenbach Reservoir, Roodeplaat Dam and South Para Reservoir). Measured data from the year 2005 and 1992 were used for Lake Penang and Lake Kenyir respectively. The HEA was applied for predicting the Chl-a and algal biovolume abundance on tropical lakes (Lake Putrajaya, Lake Penang and Lake Kenyir) in Malaysia. This study discusses the application of SALMO-PLUS and HEA towards tropical lakes eutrophication management. The results of application of SALMO-PLUS on tropical lakes are presented, simulating response of the phytoplankton community to fluctuation in nutrient loading, light availability and hydrological aspect in the water bodies. Results of applying HEA on tropical lakes are also interpreted in the context of empirical and causal knowledge on Chl-a and algal biovolumes dynamics under tropical lake water quality conditions by means of rule-based model. Results for both Lake Kenyir and Penang showed that SALMO-PLUS were able to predict annual average trends not only for chlorophyll-a but also other state variables and algal functional groups. Simulated state variables namely Chl-a, N and P showed good agreement with field observations data for both lakes. Despite the fact that this is the first time SALMO-PLUS been used for tropical lakes and the limited data availability from this region, the simulated values of biological and nutrient state variables match reasonably with measured data. Outcomes from SALMO-PLUS simulation show consistent compliance with algal community assembly obtained from other researchers. The HEA achieved reasonable accuracy in predicting timing and magnitudes of algal blooms up to 7-days-ahead. The HEA proved to be most efficient in modelling and predicting seasonal dynamics of chlorophyll-a and algal biovolumes. A sensitivity analysis conducted for Lake Penang revealed that algal abundance is not only driven by physical and chemicals characteristics of the water body but also by impact of inorganic substances in the water that contributes to high level of chemical oxygen demand in the water as well. In addition, this study has successfully implemented a new process model from Law et al. (2009) consisting algal growth, algal grazing, zooplankton growth and zooplankton mortality functions into the SALMO-OO simulation library. Combination of this new process models were tested on dataset from Lake Kenyir, Lake Penang, Saidenbach Reservoir and Roodeplaat Dam within the simulation library to discover the optimal model structures for respective water bodies. Even though the new process model was not selected in complete totality as the optimal model structure for any of the test lakes, the addition has added another alternative for water body simulation in SALMO-OO process library. Based on these forecasting results, both SALMO-PLUS and HEA have showed potential for utilisation in early warning and strategic control of algal blooms in tropical freshwater lakes. The generic nature of HEA forecast model was also observed when tested for forecasting algal biovolume for merged data of similar lake ecosystem category. Results from merged Lake Kenyir and Lake Penang data showed reasonable accuracy in predicting the timing and magnitudes of algal blooms up to 7-days-ahead. Addition of the new process model from Law et al. (2009) into the SALMO-PLUS simulation library has also expanded the alternative for lake category simulation to give a more comprehensive decision support tool for lake and reservoir management. This study has also affirmed the generality and flexibility of SALMO-PLUS for usage in tropical lakes modelling. SALMO-PLUS was observed to be capable of simulating simultaneous seasonal fluctuations in algal growth and nutrients (phosphate and nitrate) making it valuable for forecasting the impacts of various simulated scenarios for various lake management regimes.
Thesis (Ph.D.) -- University of Adelaide, School of Earth and Environmental Sciences, 2012

碩士
國立中正大學
電機工程研究所
92
With the trend of deregulation of electric powe industries, the demand of quality of electricity supply rises.The electric power utility always pursues stable and reliable electricity supply to satisfy the requirements of consumers in a way of economic and secure operation for the power system.This thesis is an attempt to explore the security-constrained economic operation of power systems. The problem of security-constrained economic operation of power systems is an optimization problem, which looks for both the economic power generation and the appropriate reactive power compensation. Real power outputs of generation units are economically dispatched to reach a minimum generation cost, whereas reactive power outputs of generation units and capacitor banks are appropriately dispatched to compensate the reactive power requirement of the system and control the bus voltage as well as line flows. The problem under study is an optimization problem and is to be solved using the hybrid differential evolution (HDE) method. HDE is computationally simple, which provides robust-search capability in a huge solution space. It employs two additional operations than the previous differential evolution (DE), these two operations are the acceleration technique and the migration technique. The acceleration technique can increase the convergence speed and the migration technique can avoid falling into a local solution and achieve the optimal solution. Application of the proposed approach is demonstrated and verified using two application systems including a 9-bus and a 26-bus systems.

碩士
國立虎尾科技大學
資訊工程系碩士班
104
In this thesis, the Cloud based Hybrid evolution algorithm for NP-Complete Pattern in Nurse Scheduling Problem (NSP) is proposed as the Software Computing as a Service (SCaaS). Due to the low birth rate, the human resource becomes the limited resource for job assignment. To find the optimal solution for staff scheduling becomes an important issue. The proposed system follows the definition of NSP and recognizes the possible problem of NP-Complete Pattern. Only the pattern is recognized as the NSP optimal problem, the proposed system can find the optimal solution. Then, the different types of evolutionary algorithm in evolution steps are integrated. Based on the proposed Feedback Assistance method, the suitable evolution steps of the evolutionary algorithm can be dynamically decided and executed. Similar to the Tasktracker and Jobtracker in cloud, all the computing load can be divided and distributed. The simulation results show that the proposed hybrid evolution algorithm can find the optimal solution with about 50% less evolution generations.

碩士
淡江大學
航空太空工程學系碩士班
103
An Hybrid Artificial Bee Colony (ABC) and Differential Evolution Algorithm (DE) was applied to the optimum design of structures in this study. The ABC algorithm is swarm intelligence based optimization technique inspired by the intelligent foraging behavior of honeybees. The advantages of ABC algorithm are quick convergence, less settings of parameter, easy to escape from the local optimal solution, and extensive searching range. Differential Evolution algorithm is an evolutional technique that has advantages of easy to implement, little parameter tuning requirement, using the difference calculation to enhance regional search capabilities, and also exhibits reliable, accurate and fast convergence. The advantage of hybrid ABC and DE algorithm is that the merit of both methods can be enhanced to overcome the disadvantage of each method. By using the characteristics of artificial bee colony algorithm which is easy to escape from the local optimal solution and the characteristics of differential evolution algorithm which is using the difference calculation to enhance regional search capabilities, the purpose of optimum design can be reached. The FORTRAN and APDL of ANSYS software are integrated into asystematic ABC-DE optimization program.The optimization problem can be transformed into a mathematical function. Then the optimum deign of structures can be obtained by ABC-DE algorithm. Minimum weight design will be demonstrated in six numerical examples. The results of ABC-DE algorithm are better than other references in the examples.

碩士
元智大學
電機工程學系
98
This thesis proposes two kinds of novel hybrid differential evolution algorithms for the design of Functional-link based Petri fuzzy neural system (FLPFNS). One is the differential evolution with back-propagation local search and self-adaptive scaling factor (DELSBP) and the other one is the differential evolution with chemotaxis and cooperative strategy (CCDE). The DELSBP enhances the performance of traditional DE algorithm by using a dynamic update strategy and back-propagation-based local search for updating the optimal vector instantaneously and enhancing the neighborhood searching capability, respectively. In addition, a self-adaptive scaling factor strategy, developed using a fuzzy logic system, is introduced to accelerate the convergence velocity. Subsequently, we propose the CCDE algorithm to enhance the performance and convergent speed by using the chemotaxis and cooperative strategy. The CCDE consists of the modified mutation strategy which includes the mutation strategy of traditional, best information, the momentum term, and the self-adaptive scaling factors, chemotaxis back-propagation strategy with optimal learning rates, and the simple cooperative strategy. The cooperative strategy provides the capability of searching optima solution by replacing the parameters of one rule of the trial vector with the parameters of the same rule of the parent vector. For the dynamic neural-fuzzy systems, the FLPFNS is proposed to enhance the ability of function approximation. We use the CCDE to learn the parameters of the FLPFNS. Finally, several simulations including nonlinear system control, impulse noise removal and function approximation and image noise removal are shown to illustrate the effectiveness of DELSBP and CCDE.

碩士
國立虎尾科技大學
資訊工程研究所
103
This thesis develops Modular Feedback Assistance Hybrid Evolution Algorithm Based on Cloud Environment to find the optimal solution of NP-complete problems such as job shop scheduling problems. In this research, the different steps and types of the evolution algorithm can be established via individual thread procedures based on various virtual machines in cloud. After the evolution steps, methods, or procedures of the genetic algorithm, the fitness evaluation result and survival ratio of different crossover methods in the current generation can be used for the proposed feedback assistance method. The feedback assistance method can be added into the evolution procedure and dynamically emphasize the corresponding methods or procedures with better performance in optimal solution searching. All the steps or methods in genetic algorithm are created independently (or modular). Furthermore, via using the feedback assistance, the convergence time of the optimal solution can be enhanced.

碩士
國立高雄應用科技大學
機械與精密工程研究所
103
Suitable for solving complex polynomial optimisation problems, the differential evolution (DE) algorithm is an evolutionary algorithm that is widely used in engineering applications. However, although the DE algorithm presents an effective global searching capability, it lacks local searching capability. The objective of this study is to develop an improved DE algorithm that overcomes the deficiency in the original DE algorithm. By referencing the notion of hybrid particle swarm optimisation (HPSO) proposed by the current research team, the partial evolution characteristics of the genetic algorithm (GA) and particle swarm optimisation (PSO) were applied to improve the DE algorithm. The resulting algorithm is referred to as the hybrid differential evolution (HDE) algorithm, which is used to optimize the dimensional synthesis of vehicle steering mechanisms without/with joint clearance. The steering mechanism of vehicles should satisfy the Ackermann steering principle to avoid sideslip of wheels from occurring. In theory, a steering mechanism that fully satisfies the Ackerman principle cannot be obtained; the only method is to minimize the steering error as possible during design processes. In addition, because of the manufacturing error of joints and wear from long-term use, joint clearance exists in physical linkage mechanisms. The joint clearance was considered a massless virtual link and the structural error of the steering mechanism was the minimized objective function. A two-stage method was used to perform the optimal dimensional synthesis of vehicle steering mechanisms with joint clearances. The GA, PSO, DE, and HDE algorithms were used to optimize the dimensional synthesis of 4-bar, 6-bar, and 8-bar vehicle steering mechanisms without joint clearance. The structural errors of optimized steering mechanisms were then compared. The comparison result showed that the HDE algorithm yielded the minimum error and optimal convergence efficiency. Subsequently, the GA was used to solve the Lagrange equation of motion for the vehicle steering mechanisms with joint clearances, and to determine the angle variations of the clearance links during the motion. Subsequently, the HDE algorithm was employed to optimize the dimensional synthesis of vehicle steering mechanisms with joint clearances. In addition, the transmission angles of the optimized steering mechanisms were analyzed. Assuming that 1-mm and 2-mm joint clearances existed, the structural errors of the optimized steering mechanism that considered joint clearances were obviously smaller than those of optimized steering mechanism that did not consider joint clearances.

碩士
國立虎尾科技大學
資訊工程研究所
102
This paper develops a modular cloud hybrid evolution algorithm based on feedback assistance for optimal data solution in traveling salesman problem. Different steps and types of the evolution algorithm can be established via individual thread procedures and various virtual machines in cloud. According to the proposed XML format, system users can upload only the coding of chromosomes without the evolution algorithm implementation. The proposed feedback assistance is based on the fitness evaluation result and survival ratio of different crossover methods. The Reduce the cost for establishing the evolution algorithm according to the proposed system can. The feedback assistance can interact with the different crossover methods and emphasize the method that can enhance more survival individuals for the next evolution generation. Furthermore, via using the feedback assistance, the convergence time of the optimal solution can be enhanced.