Research On New Mechanism And Application Of Artificial Endocrine System

Nature Inspired Computing seeks to discover and extract useful organizing and processing mechanisms of information from nature and to apply them to computing problems which traditional methods cannot solve. In the achievements of Nature Inspired Computing, algorithms inspired by human body are the most successful, such as Artificial Neural Networks, Genetic Algorithm. Because human is the only intelligent creature in the world, it is straightforward to learn from human body to achieve truly intelligence. Therefore, taking inspiration from human body is the most important research direction of Nature Inspired Computing. The information processing organisms of human body mainly include neural system, genetic system, immune system and endocrine system. These systems cooperate to modulate human body and to maintain its normal working. The systems also exhibit high intelligence and self-adaptability in their modulation process, which is source of imitation for Nature Inspired Computing.Among the control systems of human body, endocrine system is a crucial one. It plays an essential role in the modulation of growth and metabolism. It cooperates closely with neural system and immune system to maintain homeostasis of human body. Endocrine system modulates human body by a substance called hormone, which spreads everywhere in the body and modules the body in a distributed way. Endocrine system and neural system is closely tied together; they cooperate to module human body. Neural system modulates endocrine system by neuroendocrine cells; while endocrine system influences neural system by hormone. Neural system and endocrine system cooperate to form a negative feedback loop to ensure homeostasis of human body. In endocrine system itself, there are complex negative feedback modulation mechanisms too, which ensure that its modulation of human body is controllable. At present, research on endocrine system of human body is still premature. Summing all of those up, endocrine system must bear some new inspiration to computing theory. Therefore, endocrine system is worthy a lot of research efforts, and is an important research direction of Nature Inspired Computing. Computing models inspired by endocrine system are called Artificial Endocrine Models. Research on Artificial Endocrine Models is still very preliminary and applications of Artificial Endocrine Models are still very limited. Aiming as a supplement for previous work, this thesis will focus on endocrine system and strive to extract meaningful information processing mechanisms from it.This thesis is mainly composed of four parts:1. By studying the ties between neural system and endocrine system, this thesis analyzes and formalizes their correlations and applys their interaction mechanisms to improve traditional Artificial Neural Networks.When controlling human body, neural system cooperates with endocrine system closely. Neural system modulates endocrine system by neuroendocrine cells. On the other hand, hormones can affect excitement of neurons. Moreover, endocrine system can influence emotion, which could modulte neural system from high level. After thorough survey on relations between neural system and endocrine system, this thesis proposes TAES, in which Artificial Endocrine System modulates Artificial Neural Networks. Compared with traditional Artificial Neural Networks, TAES enables Artificial Neural Networks to adapt to dynamical environment more quickly.2. By studying the way endocrine system maintains homeostasis of human body, this thesis propose a model in which endocrine system help maintain stability of control system.One of the most important functions of endocrine system and neural system is to maintain homeostasis of human body. Generally, endocrine system cooperates with neural system to maintain homeostasis of human body. Neural system and endocrine system form a negative feedback loop together, which guarantees homeostasis of human body. Such processes are self-adaptive and self-organized, contains a lot of interesting and complex information processing mechanisms. Inspired by such mechanism, this thesis proposes a dynamic control system called EMNCS. Compared with traditional control systems, EMNCS can help maintain stability of the control system.3. Inspired by biological hormone, this thesis proposes a distributed control model applied in multi-robot system.To study relation of endocrine system and neural system is to study endocrine system in macroscopic perspective. However, from microscopic perspective, endocrine system modulates human body by hormones. Hormones are secreted by endocrinal glands and spread everywhere in human body via many different transmission channels. The ways hormone spread are correlated with density, type, and many other factors. Due to the distributive characteristics of hormone, research on mechanisms of hormones may give distinct insight to distributed systems like multi-robot system and others. Inspired by hormone, we propose a model called DIIMRCS which is used to control motion of robots in multi-robot system. Experimental results show that DHMRCS can reduce computation substantially.4. This thesis scrutinizes the ways hormone spreads and functions and formalizes many distinct characteristics of it. Inspired by such characteristics, this thesis proposes an algorithm which is applied in formation problem in multi-robot system.The mechanisms by which hormones spread are different from common communication principles of modern distributed systems. There hide interesting distributed communication and control ideas in endocrine system. Inspired by hormone’s way of communication and control, we propose DHFS, which is applied in formation problem in multi-robot system. Experimental results show that DHFS can improve precision of formation for multi-robot system compared with traditional formation algorithms.Based on thorough study on biological endocrine system, this thesis analyzes and scrutinizes many aspects of the system. By extracting meaningful mechanisms of the systems and apply them in many computing problems, this thesis either improve traditional Nature Inspired Computing models or propose new computing models. This thesis has made many experiments to verify proposed models, and the experimental results confirm rationality and effectiveness of those models.