【作者】 丁丽娅；

【导师】 侯春萍；

【作者基本信息】 天津大学 ， 通信与信息系统， 2010， 博士

【摘要】 信息系统为复杂系统。复杂性科学被称为是科学史上继相对论和量子力学之后的又一次革命,它的出现极大地促进了科学向纵深发展,标志着人类的认识水平步入了一个崭新的阶段。非线性是复杂系统的一个重要特征,因此采用非线性动力学理论解决复杂系统问题成为一条主要而又有效的途径。本文将从非线性动力学角度出发,研究信息系统中,包括神经信息系统和通信信息系统,一些难以用传统方法解决的控制问题。首先,对神经信息系统的研究表明,生理参数变化引发生物系统分岔可能是某些神经系统方面疾病的诱因。因此对分岔的稳定性控制研究在神经科学中占有十分重要的地位。本文以Hodgkin-Huxley神经元模型为研究对象,提出了一种利用washout滤波器控制函数的线性项来稳定Hopf分岔的新方法。通过应用Routh-Hurwitz稳定性判据,推导出使Hopf分岔稳定的控制系数范围,并且经过仿真证实了该方法的有效性。其次,从非线性动力学角度来看,神经信息系统中的短期记忆工作在双稳态动力学区域,该区域的起点为双重极限环分岔。所以对双重极限环分岔点的控制研究将为与记忆有关疾病的治疗提供有效方法和理论依据。本文提出一种利用washout滤波器控制函数的立方项来改变FitzHugh-Nagumo神经元模型双重极限环分岔点的新方法。该方法应用中心流形定理和规范形理论,推导出加入控制后闭环系统控制系数的取值范围。改变控制系数的大小就能够将该分岔点前移或置后。最后,基于对前两个具有明确数学表达形式系统的研究,提出了一种用于通信信息系统中抑制Turbo迭代译码算法的非线性动力学行为的新控制方法。研究表明,Turbo迭代译码中存在分岔、混沌等动力学行为,严重影响了译码性能。本文提出了一种改进的Mexican hat小波函数控制方法。该方法利用小波函数具有收敛性的特点,使迭代译码算法避开了低信噪比区域出现的动力学行为,快速收敛到一个不动点上。仿真结果表明所提出的方法明显降低了译码复杂度,使Turbo迭代译码算法在动力学区域内的译码性能平均改善了0.3dB,其结果优于已有报道的延迟反馈方法。更多还原

【Abstract】 The information system is a complex system. Science of complexity is called the third scientific revolution following on the heels of relativity and quantum physics. The emergence of science of complexity greatly promotes the development of science in depth and indicates the new phase of awareness of human. Since nonlinearity is an important characteristic of complex system, it is the main and effective method to solve questions of complex system with theory of nonlinear dynamics. In this dissertation, control questions unsolvable with traditional method for information system, including neural information system and communicational information system, are studied from the point of view of dynamicsFirst, research on neural information system show, bifurcations of biological systems caused by variation of physiological parameters may be the reason for some neural system diseases. Hence study on stability control of bifurcation holds on an important position in neural science. In this dissertation, a new control method via washout filter controller with linear term is given to realize stability control of Hopf bifurcation in Hodgkin-Huxley neural model. By use of Routh-Hurwitz stability criterion, the range of control coefficient is derived. Further, simulation results demonstrate the effectiveness of the proposed method.Second,from the point of view of dynamics, short-term memory of neural information system works during bistable dynamic regime. And double cycle bifurcation is the onset of bistable dynamic regime. Therefore research on control of location of double cycle bifurcation may provide therapeutic approaches and theory basis for memory problems. A new control method via washout filter controller with cubic term is presented to change location of double cycle bifurcation in this dissertation. By applying of center manifold and normal form theory, we can derive the range of control coefficient of closed system. Consequently, double cycle bifurcation poiont is advanced or delayed along with control coefficient.Finally, based on the former study of two systems with definite mathematic expressions, a new control method is proposed to suppresse nonlinear dynamic behavior in Turbo iterative decoding algorithm of communicational information system. Research show there exist dynamic behaviors in Turbo iterative decoding algorithm such as bifurcation and chaos. Thus decoding performance is seriously affected. In this dissertation, control method with modified Mexican hat wavelet function is developed. By exploiting convergence property of wavelet function, decoding algorithm can avoid the dynamic behavior under low signal-to-noise condition and converges quickly to a fixed point. Simulation results show that the proposed method makes decoding complexity decreased greatly. Moreover, on average, turbo decoding algorithm presents a gain of about 0.3 dB in dynamic region, which is superior to that of the delay feedback method reported.更多还原