基于记忆学习与免疫系统的仿生控制研究

项目名称： 基于记忆学习与免疫系统的仿生控制研究

项目编号： No.61503192

项目类型： 青年科学基金项目

立项/批准年度： 2016

项目学科： 自动化技术、计算机技术

项目作者： 翁理国

作者单位： 南京信息工程大学

项目金额： 20万元

中文摘要： 本项目旨在研究记忆学习过程与免疫系统行为以及相关生物理论基础，寻找启发，探索并建立一套新颖的仿生控制算法。记忆学习系统按照编码、储存以及提取对比三个主要步骤，通过改变神经元（Neuron）以及突触（Synapse）之间的电流强度来完成学习改进的过程；免疫系统则可以动态的调用先天免疫（Innate Immunity）与特异免疫（Acquired Immunity），根据入侵抗原（Antigen）的特征分泌与之相对应的抗体(Antibody)，保护生物体免受侵害。本项目的重点在于通过映射的方法将以上生物系统的特性“移植”到所要建立的仿生控制算法当中，使其具有免疫系统以及记忆学习系统的容错性、自我配置性、动态规划性以及自适应性，从而可以应对复杂非线性系统在复杂环境下的控制问题。在项目研究的最后阶段，将给出模拟实验结果来验证此仿生控制算法的有效性以及理论证明过程来保证此仿生控制算法的收敛性。

中文关键词： 仿生控制；记忆学习；免疫系统

英文摘要： This project focuses on mechanisms of Memory and Learning and Immune System. We are trying to explore an novel bio-inspired control algorithm based on the discoveries and inspirations from these biological behaviors. The Human Memory/Learning System is able to improve itself through adjusting electricity magnitudes between Neurons and Synapses, and during the process, the system will go through the following three steps: Encoding, Storing and Retrieving; The Immune System is able to dynamically invoke Innate Immune and Acquired Immune systems, secreting specific antibodies according to invading antigens, hence protecting our body from external viruses and bacteria. In this project, we are trying to design a bio-inspired control algorithm which possesses the properties of fault-tolerance, self-configuration, dynamic planning and self-adaptiveness through mathematical modeling and behavior mapping. This bio-inspired control algorithm is able to deal with highly nonlinear systems under very complex circumstances. Simulation results and theoretical proofs will also be given to prove its effectiveness and convergence.

英文关键词： Bio-inspired Control;Memory and Learning;Immune System