脑血流自动调节能力评估关键算法及下肢束带平台研究

发布时间：2018-06-07 00:53

本文选题：脑血流生理调节 + 最小角度准则　；参考：《哈尔滨工业大学》2015年硕士论文

【摘要】：在血压变化正常、可控范围内,大脑内的血流将保持相对恒定,这是由脑血流自动调节机能实现的。调节能力一旦减弱,将引起重大危险性疾病。更重要的是,往往在疾病出现症状之前,调节能力特征参数早已发生改变。临床若能够准确的测量自动调节生理系统,便可以通过优化手段来调控血压,积极有效的改善患者的治疗方式。传统测量方法是建立一个单输入单输出或二输入多输出模型,利用自适应最小二乘辨识算法,通过系统权系数序列评估调节机能的特征参数:系统相位差或调节机能指数。但存在于调控与评估过程中的变异性会对特征参数的辨识产生消极影响,而这是由权系数序列中无关项干扰引起的。模型简单、抗干扰能力差,信号成分复杂、特征不明显都会产生无关项,致使评估结果出现稳态误差和动态波动。同时,以往方法中相位评估结果单一、不具有时域特征,调节机能指数评估为整数类型、精确度较低。因此,传统方法在评估调节机能上效果不理想,得到的辨识曲线误差较大,波动明显。为了减小变异性,一种高效的评估方法显得尤为重要。本课题通过研究、探索,建立了不同于以往的二输入单输出模型,弥补了抗干扰能力不足的缺点;实现高分辨率的高斯滤波,对数据进行关键处理步骤,减弱噪声干扰;应用最小角度准则,减少无关项存在,实现最小角度一次算法与自适应算法,高效控制稳态误差和动态波动,变异性整体上减小。同时,实现相位差希尔伯特算法和调节机能指数非整数算法,将辨识结果最优化。为了采集到最佳特征状态的生理信号,搭建了一个下肢束带平台,完全的诱发调节机能,降低错误辨识的概率。最后进行仿真数据和实验数据分析,并对比传统方法,验证了新评估方法的有效性。
[Abstract]:The blood flow in the brain remains relatively constant in a controlled and normal range of blood pressure, which is achieved by the automatic regulation of cerebral blood flow. Once the regulation ability weakens, will cause the serious dangerous disease. More importantly, regulatory characteristics often change long before symptoms appear. If the automatic physiological system can be measured accurately, the blood pressure can be adjusted by optimizing means, and the treatment mode of patients can be improved actively and effectively. The traditional measurement method is to establish a single input, single output or two input multiple output model. The adaptive least square identification algorithm is used to evaluate the characteristic parameters of the regulating function by the weight coefficient sequence of the system: the system phase difference or the regulating function index. However, the variability in the process of regulation and evaluation can have a negative impact on the identification of characteristic parameters, which is caused by the interference of irrelevant terms in the weight coefficient series. The model is simple, the anti-interference ability is poor, the signal component is complex, and the characteristics are not obvious, which will cause the stable error and dynamic fluctuation of the evaluation results. At the same time, in the previous methods, the phase evaluation results are single and have no time domain characteristics. The adjustment function index evaluation is of integer type, and the accuracy is low. Therefore, the traditional method is not effective in evaluating the regulation function, the error of the identification curve is large and the fluctuation is obvious. In order to reduce variability, an efficient evaluation method is particularly important. Through the research and exploration, the paper establishes a two-input single-output model which is different from the past, which makes up for the deficiency of anti-jamming ability, realizes the high-resolution Gao Si filter, carries on the key processing step to the data, reduces the noise interference; The minimum angle criterion is applied to reduce the existence of independent terms and to realize the minimum angle primary algorithm and adaptive algorithm. The steady-state error and dynamic fluctuation are effectively controlled and the variability is reduced as a whole. At the same time, the phase difference Hilbert algorithm and the non-integer algorithm of adjusting the function index are implemented to optimize the identification results. In order to collect the physiological signal of the best characteristic state, a lower limb banding platform was built to induce the regulation function completely and reduce the probability of error identification. Finally, the simulation data and experimental data are analyzed, and compared with the traditional methods, the effectiveness of the new evaluation method is verified.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TN911.7;R741

【共引文献】