This thesis investigates the performance of pulse charging, which is believed to be superior to constant current charging in some respects, such as charging efficiency and charging speed. The investigation is focused upon the extensively used secondary batteries, lead-acid batteries. The consecutive orthogonal arrays method is applied to search for the optimum operating variables of pulse charging, including pulse amplitude, duty ratio and frequency of the charging current. Unfortunately, the experimental results of consecutive orthogonal arrays reveal that charging efficiency is not obviously affected by pulse amplitude, duty ratio or frequency. Instead, charging rate is dominantly influenced by average charging current. These results indicate that pulse charging scheme is not superior to constant current charging. To compare these two charging schemes further, a series of experiments are carried out to discuss the effects of each operating variables. Unfortunately, no evidence from the experimental results can prove the superiority of pulse charging to constant current charging as formerly documented.

Due to the improvement of luminous efficiency, LED has been applied in general lighting nowadays. However, the efficacy and life of high power LED lamps are degraded for the problem of overheat temperature of the chips. In accordance with the above reason, the thermal management has been important in LED application. The thesis aims to analyze the thermal problem of high power LED lamp by FEM. In the beginning, we analyzed the temperature curves and thermal stress curves of chips by changing the thickness of AuSn solder in 4 structures. According to the results, the better thickness range and structures were chosen. In the second step, the effect of different sizes of a vapor chamber on the different thermal conductivities of lamp system was discussed. Finally, we orthogonalized all the designed parameters by Taguchi orthogonal array method, and then found the optimal design by comparing the results with the initial analyses. In this study, we proposed a solution to improve the quality characteristics of LED lamp by Taguchi method. So the effect of each control factors on the performance was able to be determined. For different features of demand, the present study is helpful to achieve the ideal design in manufacturing. Key words: FEM, high power LED lamps, Taguchi orthogonal array method

To realize efficient parallel processing, which is one of effective methods that deal with computing intensive applications, the technology of solving the problems of task matching and scheduling becomes extremely important. In this thesis, an Ant Colony Optimization (ACO) approach is employed for allocating task graphs onto a heterogeneous computing system. The approach uses a new state transition rule to reduce the time needed for finding a satisfactory solution. And a local search procedure is designed to improve the obtained solution. Furthermore, by applying the Taguchi Method in the technology of Quality Engineering, and further utilizing the Orthogonal Array (OA) to reduce the number of experiments and find the optimal combination of parameters, which allows the Ant Colony Algorithm to find solutions more efficient. The proposed algorithm is compared with the genetic-algorithm-based approach and the dynamic priority scheduling (DPS) heuristic. Experimental results show that the ACO approach outperforms two computing approaches in solving the task matching and scheduling problem.

A robust design utilizing consecutive orthogonal arrays algorithm is proposed for designing electronic ballasts of fluorescent lamps. By this design method, the variation in the lamp power can be less than 10% under different operating conditions. In the manipulation of the consecutive orthogonal arrays, component values of the ballast circuit and DC-link voltage are used as controllable variables for inner orthogonal arrays; while manufacturers, ambient temperature, used hours, and variation in DC-link voltage are treated as uncontrollable variables for outer orthogonal arrays. The average effects of the output power for each controllable variable are calculated from simulation results, which are served as indexes to find the combination of circuit parameters with a better solution. With consecutive orthogonal arrays, the target values of the circuit parameters are approached step by step. In addition, the effect of the DC-link voltage on the lamp power can be understood from the uncontrollable variable of outer orthogonal arrays. The proposed design tool is implemented on the design of an electronic ballast for a 40W fluorescent lamp. The test results show that the designed electronic ballast can be adopted for the lamps from different manufacturers, with different used hours, and under variation in a wide range of ambient temperature.

In this thesis, an alternative solution for designing power line conductive electromagnetic interference (EMI) filter by using the consecutive orthogonal array method is proposed. The circuit parameters of EMI filters to be determined are assigned as the control variables in the orthogonal arrays, and the average effects corresponding to each control variable are calculated from the measured results. In accordance with the inferential rules, the average effects are used as the observational indices to adjust the levels of the control variables of the subsequent orthogonal array. Through manipulating consecutive orthogonal arrays step by step, the applicable ranges of circuit parameters are approached with desired output performances. Finally, the component values of EMI filters with minimum size can be found. The design procedure and the inferential rules are described by illustrative examples for a single-stage high-power-factor converter.

No description available.

An approach with ¡§consecutive orthogonal arrays (COA)¡¨ is proposed for solving the problems in designing power electronic circuits. This approach is conceptually based on the orthogonal array method, which has been successfully implemented in quality engineering. The circuit parameters to be determined are assigned as the controlled variables of the orthogonal arrays. Incorporating with the inferential rules, the average effects of each control variable levels are used as the indices to determine the control variable levels of the subsequent orthogonal array. By manipulating on COA, circuit parameters with the desired circuit performances can be found from an effectively reduced number of numerical calculations or experimental tests. In this dissertation, the method with COA is implemented on solving four problems often encountered in the design of power electronic circuits. The first problem one has to deal with is to find a combination with the best performance from a great number of analyzed results. The illustrative example is the design of LC passive filters. Using COA method, the desired component values of the filter can be effectively and efficiently found with far fewer calculations. The second design problem arises from the non-linearity of circuit. An experienced engineer may be able to figure out circuit parameters with satisfactory performance based on their pre-knowledge on the circuit. Nevertheless, they are always questioned whether a better choice can be made. The typical case is the self-excited resonant electronic ballast with the non-linear characteristics of the saturated transformer and the power transistor storage-time. In this case, the average effects of COA obtained from experimental tests are used as the observational indexes to search a combination of circuit parameters for the desired lamp power. The third problem is that circuit functions are mutually exclusive. The designers are greatly perplexed to decide the circuit parameters, with which all functions should be met at the same time. The method with COA is applied to design a filter circuit to achieve the goals of low EMI noise and high power factor simultaneously. Finally, one has to cope with the effects of the uncontrolled variables, such as: ambient temperature, divergence among different manufacturers, and used hours. By applying COA with inferential rules, electronic ballasts can be robustly designed to operate fluorescent lamps at satisfied performance under the influence of these uncontrolled variables.

A method for performing robust optimal design that combines the efficiency of experimental designs and the accuracy of nonlinear programming (NLP) has been developed, called Search-and-Zoom. Two case studies from the RF and communications industry, a high-frequency micro-strip band-pass filter (BPF) and a rectangular, directional patch antenna, were used to show that sensitivity optimization could be effectively performed in this industry and to compare the computational efficiency of traditional NLP methods (using fmincon solver in MATLAB R2013a) and they hybrid method Search-and-Zoom. The sensitivity of the BPF's S11 response was reduced from 0.06666 at the (non-robust) nominal optimum to 0.01862 at the sensitivity optimum. Feasibility in the design was improved by reducing the likelihood of violating constraints from 20% to nearly 0%, assuming RSS (i.e., normally-distributed) input tolerances and from 40% to nearly 0%, assuming WC (i.e., uniformly-distributed) input tolerances. The sensitivity of the patch antenna's S11 function was also improved from 0.02068 at the nominal optimum to 0.0116 at the sensitivity optimum. Feasibility at the sensitivity optimum was estimated to be 100%, and thus did not need to be improved. In both cases, the computation effort to reach the sensitivity optima, as well as the sensitivity optima with RSS and WC feasibility robustness, was reduced by more than 80% (average) by using Search-and-Zoom, compared to the NLP solver.

近年來，我國對於田口式品質工程方法的運用，已漸漸普及於各個生產事 業。然而，在以往田口方法的應用實例中大多只使用一種數據分析法，即 傳統實驗設計的變異數分析法，或田口方法的信號雜音比( S/N ratio )，對於二者之間的比較及異同處，則較少提及﹔且絕大部分是只探討單 一品質特性最佳化的問題，對於同一製程中同時具有多個品質特性需最佳 化的問題，亦較少觸及。本研究首先分別引用二種數據分析法進行提昇自 動光學檢驗(AOI)偵測能力」}的數據解析、比較二種分析法在理論上及應 用上的差別，並依保守原則，初步決定各品質特性(本研究有三個品質特 性)的最佳因子水準組合。其次，針對各品質特性間因子水準互相矛盾的 情形，本文將引用下述3種分析方法於研究案例，並綜合各種方法的優缺 點及適用狀況，藉以找出多品質特性同時最佳化的因子水準組合。此三種 分析方法為: (1) 畫出各因子對個別品質特性之影響效果總調查表，再經 由人為的比較判斷後選取最佳的條件組合 (2) 以個別品質特性之信號雜 音比的加權和作分析 (3) 使用品質特性值標準化的方法做分析由分析結 果，吾人可知現有各種方法均有其優缺點，唯有熟悉各種方法並配合豐富 經驗與專業知識，視實驗狀況而選擇性的加以應用，方能獲致良好實驗效 果。

L’objectif de ces travaux de proposer aux autorités locales un outil d’aide à la décision pour formaliser une démarche participative tout au long de la conception d’un projet de transport urbain durable, dans un contexte multicritères et multi acteurs. La méthodologie développée repose sur l’évaluation a priori des impacts d’un projet local, en impliquant les acteurs dès la phase de diagnostic. Cette phase permet d’identifier des groupes d’acteurs selon leur perception des phénomènes urbains, des interactions et des préférences déclarées d’évolution. Elle permet d’établir une liste d’indicateurs communs qui seront évalués. Le choix des alternatives à étudier est le résultat de l’application conjointe de techniques de transférabilité - fondées sur l’étude de projets développés dans d’autres villes - et de l’étude des préférences déclarées des usagers locaux. La probabilité d’utilisation du service s’évalue à l’aide d’un modèle comportemental agrégé qui intègre la perception floue et l’indécision des usagers devant une situation nouvelle. Les changements de comportement des usagers sont pris en compte à l’aide d’un indicateur de robustesse qui teste les effets des paramètres exogènes sur l’évolution de la probabilité d’utilisation du service. Un micro-simulateur de trafic permet d’évaluer les impacts des différents scénarii sur le trafic, l’environnement et le bien-être des citoyens sous la forme monétarisée, afin d’illustrer les coûts et les bénéfices indirects attendus de l’implantation du projet. Une solution de compromis est proposée : elle vise à fournir la solution qui contenterait au mieux l’ensemble des représentants des groupes d’acteurs - et non la solution optimale en termes d’impacts. / The objective of this research is to provide to the local authorities a decision aid tool in order to formalize a participatory approach during the conception of a sustainable urban transport project, in a multi-criteria and multi-actors context. The methodology is based on the a priori evaluation of the impacts of a local project, involving stakeholders as soon as the diagnosis phase. This phase aims at identifying groups of actors according to their perception of urban phenomena, their interactions and their stated preferences of evolution. The diagnosis phase leads to the setting-up of a list of common indicators to be evaluated. The choice of alternatives to be studied is the result of the transferability techniques - based on the projects developed in other cities - and the stated preferences of local users. The probability of using the service is evaluated using an aggregated behavioral model that takes into account the fuzzy perception and the indecision of users in a new situation. Changes in the behaviour of the users are taken into account thanks to an indicator of robustness that tests the impact of exogenous parameters on the evolution of the probability of using a service. A traffic micro--simulator aims at assessing the impacts of the various scenarios on traffic, environment and the welfare of citizens– which is monetized. It illustrates the costs and indirect benefits expected with the implementation of project. A compromise solution is proposed: it aims at identifying an alternative that would best satisfy the representatives of the stakeholder groups – and not necessarily the optimal solution in terms of impacts.