| DC 欄位 |
值 |
語言 |
| DC.contributor | 電機工程學系 | zh_TW |
| DC.creator | 黎乃仁 | zh_TW |
| DC.creator | Nai-Jen Li | en_US |
| dc.date.accessioned | 2015-7-24T07:39:07Z | |
| dc.date.available | 2015-7-24T07:39:07Z | |
| dc.date.issued | 2015 | |
| dc.identifier.uri | http://ir.lib.ncu.edu.tw:88/thesis/view_etd.asp?URN=985401029 | |
| dc.contributor.department | 電機工程學系 | zh_TW |
| DC.description | 國立中央大學 | zh_TW |
| DC.description | National Central University | en_US |
| dc.description.abstract | 本論文提出了以權重粒子(Weighted Particle)為基礎的粒子群最佳化法(Particle Swarm Optimization, PSO)之改良。權重粒子是由粒子群中，各粒子的個人最佳解位置與各粒子所紀錄的最佳目標函數值(最小化)所組成，而權重粒子可以提供更靠近最佳解的方向讓粒子群有更多機會找到最佳解。藉由加入權重粒子所改良的粒子群最佳化法，本論文提出三種改良演算法，第一種為嵌入權重粒子的增強式粒子群最佳化法(Enhanced Particle Swarm Optimization with Weighted Particle, EPSOWP)，EPSOWP藉由吸引值的設定將粒子群的解搜尋行為分為區域搜尋與全域搜尋，使得粒子能有彈性的搜尋策略。第二種為結合模糊推理與權重粒子的粒子群最佳化法(Hybrid Particle Swarm Optimization Incorporating Fuzzy Reasoning and Weighted Particle, HPSOFW)，HPSOFW藉由模糊推理調整吸引因子與慣性權重兩種參數，其目的是為了在搜尋解的過程中，各粒子都能根據紀錄最佳目標函數值與自己目前的目標函數值調整解的搜尋行為。第三種為結合權重粒子之改良粒子群最佳化法為基礎的模糊規則產生演算法(Fuzzy Rules Generation Based on Modified Particle Swarm Optimization with Weighted Particle, MPSOWP)，此模糊規則產生演算法(FRG)可視為一控制器，透過輸入控制器的資料產生模糊規則庫，接著利用MPSOWP搜尋最佳模糊規則庫的設計參數。其中MPSOWP將傳統PSO的搜尋行為中的全域最佳解粒子替換成權重粒子，並結合了另一個新的搜尋行為來改良搜尋解的策略。最後，透過模擬結果與應用於各種神經網路的系統建模、PID控制器和模糊規則庫設計來呈現改良演算法的效能。 | zh_TW |
| dc.description.abstract | This dissertation proposes weighted particle based modified particle swarm optimization (PSO) and its applications. The weighted particle is composed of personal best particles with their best objective value (minimal value), and the weighted particle can provide better opportunity getting closer to the optimal solution. Based on weighted particle, this dissertation proposes three modified algorithms. One is an enhanced particle swarm optimization with weighted particle (EPSOWP) which has two search behaviors of local search and global search so that the swarm in EPSOWP has more flexible search strategy via an attraction value. Another is a hybrid particle swarm optimization incorporating fuzzy reasoning and weighted particle (HPSOFW). The fuzzy reasoning adjusts attraction factor and inertia weight to change search behaviors of the particles by comparing the best objective value with their current objective value. The other is a fuzzy rule generation (FRG) based on modified particle swarm optimization with weighted particle (MPSOWP). The FRG is seen to be a controller to generate a fuzzy rule base which is composed of input data, and to optimize the parameters of fuzzy rule base by using proposed MPSOWP. In MPSOWP, the search behavior of conventional PSO is modified by changing from global best particle to weighted particle, and another proposed search behavior is incorporated into MPSOWP for improving search strategy. Finally, the simulation results of the neural network based system identification, the PID control and the fuzzy rule base design verify proposed algorithms which outperform various optimization algorithms. | en_US |
| DC.subject | 粒子群最佳化法 | zh_TW |
| DC.subject | 權重粒子 | zh_TW |
| DC.subject | 模糊系統 | zh_TW |
| DC.subject | 神經網路 | zh_TW |
| DC.subject | Particle Swarm Optimization | en_US |
| DC.subject | Weighted Particle | en_US |
| DC.subject | Fuzzy Rule Base | en_US |
| DC.subject | Neural Network | en_US |
| DC.title | 以權重粒子為基礎之改良式粒子群最佳化演算法與其應用 | zh_TW |
| dc.language.iso | zh-TW | zh-TW |
| DC.title | Weighted Particle Based Modified Particle Swarm Optimization and Its Applications | en_US |
| DC.type | 博碩士論文 | zh_TW |
| DC.type | thesis | en_US |
| DC.publisher | National Central University | en_US |