基于混合粒子群算法的微波酯化反應模型的優化

化工論文編號:HG095  論文字數:9509,頁數:28

目  錄
中文摘要 I
英文摘要 II
目  錄 III
1. 緒論 1
2. 支持向量機(Support Vector Machines) 3
3. 粒子群算法 7
3.1  基本粒子群算法 7
3.2  混合粒子群算法 9
3.2.1  混沌粒子群優化算法 9
3.2.2  基于模擬退火的粒子群優化算法 11
3.3  混沌粒子群算法性能測試 11
4. 模型的建立和優化 12
4.1 數據預處理 12
4.2 交叉驗證 14
4.3 參數的選擇 15
4.4 建立模型 15
5. 總結與展望 19
致  謝 20
參考文獻 21
附錄 MATLAB代碼 22

 摘 要：粒子群優化(Particle Swarm Optimization，PSO)是一種高效的優化搜索算法，在函數優化、神經網絡訓練、模式分類、模糊系統控制等領域得到了廣泛的應用。標準粒子群優化（Standard Particle Swarm Optimization）算法容易使模型陷入局部極值，而采用混沌粒子群算法 (CPSO)，即對全局極值采用自適應混沌優化策略，當出現早熟收斂時，對部分較優粒子采用混沌優化策略，擺脫了局部極值，得到全局最優。本文針對微波酯化反應過程表現出較強的非線性以及眾多的影響因素，利用偏最小二乘支持向量機（partial Least Square, LS-SVM）基于實驗數據對水楊酸乙酯的微波催化合成反應進行建模。針對該反應，利用混沌粒子群算法優化偏最小二乘支持向量機模型，模型的擬合誤差平方和為0.066%，得到最優條件為：酸醇比0.14，功率402W，催化劑用量3.00mL，反應時間42min，對應的產率為80.11%。
 
 關鍵詞：酯化；建模；支持向量機；混合粒子群

Hybrid Particle Swarm Optimization of Microwave esterification reaction model
 
Abstract: PSO is an efficient search algorithm,it has been widely used In function optimization, neural network training, pattern classification, fuzzy system control and other fields. Standard PSO algorithm is easy to make the model into a local minimum. Chaotic particle swarm optimization (CPSO), the global extreme adaptive chaos optimization strategy, when there is premature convergence, the optimum particle on the part of the chaos optimization strategy used, out of a local minimum, obtain the global optimum. Chaotic particle swarm optimization using partial least squares support vector machine model. This article papers for the esterification process of microwave showed a strong non-linear as well as numerous factors，Using partial least squares support vector machine based on experimental data on the Microwave Synthesis of Ethyl Salicylate reaction modeling。Model fitting error square is 0.066%, the optimal conditions: acid alcohol ratio of 0.14,the power of 402W, catalyst of 3.00 mL and reaction time of 42 min. The corresponding yield is 80.11%.

Keywords：Esterification; modeling; support vector machine; hybrid particle swarm