A Study on Construction of Fundamental Theory for Biological-Oriented Manufacturing Systems

面向生物的制造系统概念开发是一个人工制品的伪生态系统。特别是从分子生物学解释蛋白质合成的角度，详细讨论了从原材料到产品的多阶段发展过程。面向生物的制造系统的进化与学习研究了自主装配所需的装配模型及其表示方法，生物面向制造系统的进化与学习建立了一个简单的生物面向制造系统的数学模型。将制造系统视为一个伪生态系统，其中不同的主体和生产要求分别是物种和全球环境。针对细菌的定向进化现象，提出了一种改进的遗传算法--中性突变遗传算法。通过求解包括遗传算法难度的问题，如非平稳性和欺骗性，检验了优化性能。此外，为制造系统开发了两种所需的学习模式。其中一种是复合随机学习自动机，以获得群体之间的合作行为，另一种是一种新的有效的强化学习算法在动态环境中。

Conceptual Development of Biological-Oriented Manufacturing System was purposed as a pseudo ecosystem of artifacts. Especially, the multi-developmental process from original materials to products in detail was discussed, from the point of view of the explanation of protein synthesis in molecular biology.Evolution and Learning in a Biological-Oriented Manufacturing SystemThe formulation of assembly models required for autonomous assembly and its notation method is developed, named the graph of Connection Precedence Relationship (CPR-graph).Evolution and Learning in a Biological-Oriented Manufacturing SystemA mathematical model of a simple Biological-Oriented Manufacturing System was purposed. The manufacturing system was considered as a pseudo ecosystem, in which the different kinds of agents and the production requests were species and the global environment, respectively. An extended genetic algorithm was developed, named GA with neutral mutations, motivated by the phenomenon of the directed evolution in bacteria. The optimization performance was examined by solving problems including GA-difficulty, such as nonstationarity and deception.In addition, two learning schema required was developed for the manufacturing system. One of them was the composite stochastic learning automata in order to obtain the cooperative behavior among the group of agents, the other was a new efficient reinforcement learning algorithm in a dynamic environment.