基于模型的癫痫状态预测控制研究

发布时间：2018-04-16 23:32

本文选题：预测控制 + 癫痫　；参考：《天津大学》2014年硕士论文

【摘要】：癫痫作为一种严重危害人类健康的常见神经系统疾病,是由大脑局部病变引起的。目前基于模型的研究是许多致力于癫痫疾病控制研究的科学家最青睐的研究方法。这些基于模型的理论研究的目的是找到一个控制律使特定的性能指标最优,目前最吸引人的是最优控制对神经系统疾病的控制研究。本文的目标是对癫痫疾病的模型实施预测控制,预测控制不仅保留了最优控制中性能指标最优的特点,而且利用在线滚动优化过程弥补了最优控制中全局优化的不足,并且预测控制为闭环控制,可以改善临床上神经系统疾病的开环刺激效果。本论文的研究是基于计算模型的,这是理解疾病的最快,最简便有效的研究阶段。通过对能够代表癫痫特性的多维电导间室模型,一维相模型和神经网络模型的预测控制,实现了神经元和神经网络的放电模式的控制,具体研究包括以下内容:首先,本文利用两层控制算法——输入输出广义预测控制实现癫痫神经元放电模式的控制。对神经元模型施加两种控制策略:控制加在胞体和控制加在树突的控制策略。两种控制策略对Pinsky-Rinzel(PR)模型模拟的正常和癫痫状态下的放电模式的预测控制均得到很好的效果。同时将此控制算法的控制性能指标与单独用输入输出线性化控制的两种控制策略进行比较。其次,利用预测控制实现一维简化相模型的癫痫神经元放电相位控制。利用PR模型的相响应曲线分别得到其正常状态下和癫痫状态下的相模型。基于输入输出广义预测控制实现对癫痫状态下相模型的相位的控制,使其跟踪正常状态是相模型的相位。最后,实现癫痫状态的预测控制。研究了癫痫疾病与小世界网络和平均场电位的关系,建立了Hindmarsh-Rose(HR)神经元小世界网络模型。将代表癫痫状态的小世界网络同步放电模式的平均场电位控制为代表正常状态的小世界网络去同步放电模式的平均场电位。为了验证此控制算法的有效性,又将小世界网络不同步放电模式控制为同步放电模式,实现了神经网络的同步场电位和去同步场电位之间的转换。本文的研究对神经系统疾病治疗的研究提供了思路,对神经系统疾病控制问题的硬件实现提供了理论依据,为神经系统疾病治疗的体外研究,活体研究和临床研究提供了重要的理论价值。
[Abstract]:Epilepsy, as a common nervous system disease that seriously endangers human health, is caused by local lesions of the brain.At present, model-based research is the preferred research method for many scientists devoted to epileptic disease control.The purpose of these model-based theoretical studies is to find a control law that optimizes certain performance indicators. At present, the most attractive is the optimal control of nervous system diseases.The objective of this paper is to implement predictive control on the model of epilepsy disease. Predictive control not only retains the characteristics of optimal performance index in optimal control, but also makes up for the deficiency of global optimization in optimal control by on-line rolling optimization process.The predictive control is closed-loop control, which can improve the effect of open loop stimulation in clinical nervous system diseases.The research in this paper is based on the computational model, which is the fastest, most convenient and effective stage to understand disease.Through predictive control of multi-dimensional conductance chamber model, one-dimensional phase model and neural network model, which can represent the characteristics of epilepsy, the discharge mode of neuron and neural network is controlled. The specific research includes the following: first,In this paper, a two-layer control algorithm, I / O generalized predictive control, is used to control the discharge pattern of epileptic neurons.Two control strategies were applied to the neuron model: control on the cell body and control on the dendrite.The predictive control of the normal and epileptic discharge patterns simulated by the Pinsky-Rinzelberg model is well achieved by the two control strategies.At the same time, the control performance index of this control algorithm is compared with the two control strategies using input and output linearization control alone.Secondly, predictive control is used to control the discharge phase of epileptic neurons in one-dimensional simplified phase model.The phase response curves of PR model are used to obtain the phase models under normal state and epileptic state respectively.The phase control of the phase model in epileptic state is realized based on the input and output generalized predictive control (GPC) so that the phase of the phase model can be traced to the normal state.Finally, predictive control of epileptic status is realized.The relationship between epilepsy and small-world network and mean field potential is studied. The model of Hindmarsh-RoseHR-neuron small-world network is established.The mean field potential of the synchronous discharge mode of the small world network which represents the epileptic state is controlled as the average field potential of the de-synchronous discharge mode of the small world network representing the normal state.In order to verify the effectiveness of this control algorithm, the non-synchronous discharge mode of small-world network is controlled as synchronous discharge mode, and the conversion between synchronous field potential and de-synchronous field potential of neural network is realized.The research in this paper provides a theoretical basis for the research on the treatment of nervous system diseases and the hardware realization of the disease control problems of the nervous system, and provides a theoretical basis for the in vitro study of the treatment of the diseases of the nervous system.In vivo and clinical studies provide important theoretical value.
【学位授予单位】：天津大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：R742.1

【参考文献】