酱油种曲培养过程的控制、建模与优化
文献类型：学位论文
| 作者 | 张如意 |
| 学位类别 | 工程硕士 |
| 答辩日期 | 2012-05-28 |
| 授予单位 | 中国科学院研究生院 |
| 授予地点 | 中国科学院自动化研究所 |
| 导师 | 王学雷 |
| 关键词 | 酱油种曲 控制系统 广义回归神经网络 粒子群优化 Seed Koji of Soy Sauce Control System General Regression Neural Network Particle Swarm Optimization |
| 其他题名 | Control, Modeling and Optimization in Soy sauce’s Seed Koji Cultivation Process |
| 学位专业 | 计算机技术 |
| 中文摘要 | 酱油是我国传统调味品，历史悠久。酱油种曲是酱油生产的种子，是由酱油生产所需的菌种（米曲霉、酱油曲霉、黑曲霉等）经纯种培养而得含有大量孢子的曲种。种曲的优劣直接影响到酱油的质量、发酵速度以及蛋白质和淀粉的水解程度等，所以研究提高种曲产量和质量对酱油产业具有重要的经济意义和实用价值。 对于种曲培养过程的研究，传统方法都是在实验室环境下，工艺人员设计单因素实验分析影响种曲质量的因素，并设计正交试验优化各个因素。随着科技水平的日益提高和大规模生产需要，种曲培养方式逐渐由人工培养转向计算机控制自动培养。计算机运行中会记录大量种曲培养过程的数据，提供了基于数据分析和优化种曲过程的可能性。 本文在企业横向委托项目——“酱油种曲系统”的基础上完成。首先使用SIEMENS过程控制系统软硬件设计完成种曲控制系统，提出培养过程压力、温度和湿度的控制方法，并在工程应用中取得了优良效果；其次根据种曲车间中采集的数据研究酱油种曲培养过程的分析方法。针对复杂非线性的种曲培养过程，建立广义回归神经网络模型预测种曲孢子数，对比K近邻算法和BP神经网络，广义回归神经网络更为准确和稳定；最后以广义回归神经网络为基础研究种曲培养过程的优化方法，采用粒子群算法优化种曲培养条件，优化的温度、压力等培养条件对改善种曲工艺具有重要的指导意义。 |
| 英文摘要 | Soy sauce is a traditional Chinese seasoning which has a long story. Seed koji is koji that contains a large number of spores, pure cultured from moulds (Aspergillus oryzae, Aspergillus sojae, Aspergillus niger, etc). As the seed of soy sauce production, Seed koji’s quality is directly related to the quality of soy sauce, fermentation speed, and, protein and starch hydrolysis degree. Therefore, the study of improvement of seed koji in quantity and quality has great economic significance and practical value. Most of research about koji cultivation process is conducted in laboratory. First, Technologists carry out experiments based on single parameter that affects the quality of seed koji; then orthogonal experiments are implemented to determine the best value of these parameters. As the computer technology progresses and the need of large scale production, traditional way of koji cultivation (artificial cultivation) is being replaced by computer control system. Lots of data is recorded when the computer is running, so it’s possible to analysis and optimize seed koji cultivation process based on data. The background of this paper is the project: “Seed koji system of soy sauce”. First, we implemented the control system by the use of SIEMENS PLC, redesigned control strategy of pressure, temperature and humidity in cultivation process, and gained considerable result in practice. Second, after data acquisition we fully discussed methods for analyzing and optimizing the process. Against complex nonlinear of seed koji cultivation process, General Regression Neural Network (GRNN) was constructed to predict spores number, compared with K-nearest neighbor method and BP neural network, GRNN shows better performance both in accuracy and stability. Last, base on the GRNN model, Particle Swarm Optimization algorithm was adapted to optimize cultivation conditions. The optimized result had important guiding significance to improve Seed koji process parameters. |
| 语种 | 中文 |
| 其他标识符 | 2009M8014629012 |
| 源URL | [http://ir.ia.ac.cn/handle/173211/7616]  |
| 专题 | 毕业生_硕士学位论文 |
| 推荐引用方式 GB/T 7714 | 张如意. 酱油种曲培养过程的控制、建模与优化[D]. 中国科学院自动化研究所. 中国科学院研究生院. 2012. |
入库方式： OAI收割
来源：自动化研究所
浏览0
下载0
收藏0
其他版本
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。