难治性癫痫无创闭环神经调控的多尺度优化自抗扰方法研究

For human being, epilepsy seriously affects not only physical and mental health, but also the quality of life. By realizing intercommunication between brain and computers in a noninvasive way, closed-loop neuromodulation is able to perceive neural activities and adjust control signals to suppress epilepsy in real time. By comparison with side effects of antiepileptic drugs and the invasion of surgery, noninvasive closed-loop neuromodulation is capable of achieving better effect and slighter side effects. However, most of the reported approaches are model-based, which can not guarantee lasting and stable effect of neuromodulation. In order to address such problem, active disturbance rejection control, which depends on less information of system model and is robust enough to disturbance, will be exploited. Additionally, in view of the characteristics of neuromodulation, a multi-scale optimization active disturbance rejection control approach will be proposed to address the intractable epilepsy in a noninvasive closed-loop neuromodulation way. In this research project, we will focus on the design of a varying scale extended state observer, so as to optimize the disturbance estimation and noise attenuation. Then, shaping technique of control synthesis will be investigated to optimize the system response. Moreover, based on integral of time-multiplied absolute-value of error (ITAE) criterion, quadratic optimization technique of control parameters will be considered to improve the neuromodulation. Such approach will keep the neuromodulaton be sustained and stable. The project will introduce a new idea to the closed-loop neuromodulation. Results of the project will improve the closed-loop neuromodulation effectively. The research will also provide a theoretical and technical basis for the development of some related areas, such as control and optimization in complex systems, nerves and brain science.

癫痫严重影响人类身心健康和生活质量。与药物的副作用、手术的有创性相比，无创闭环神经调控，以非侵入方式建立脑与计算机的双向通讯，实时感知神经活动、调整调控信号、抑制癫痫，效果好、副作用小。然而，现有调控算法因依赖模型而无法保证调控效果的持续、稳定。对此，项目采用对模型依赖少、受扰动影响小的自抗扰控制，并结合神经调控的特点，提出难治性癫痫无创闭环神经调控的多尺度优化自抗扰控制方法，通过1)研究变尺度扩张状态观测模型，优化扰动估计与噪声抑制；2)研究控制结构整形，优化调控响应；3)研究时间与误差绝对值乘积积分指标最佳的调控参数二次优化，提高调控效果；进而保证调控效果的持续、稳定。项目将为闭环神经调控研究带来新思路，成果将有效提高闭环神经调控效果，为复杂系统的控制与优化、神经与脑科学等相关学科的发展奠定理论和技术基础。

癫痫严重影响人类身心健康和生活质量，癫痫闭环神经调控依据患者实时状态提供有效调控信号，是近年来癫痫治疗领域重要且最具潜力的研究方向。本课题以癫痫闭环神经调控为目标，通过实时感知神经活动、调整调控信号、抑制癫痫，获得了期望的闭环调控效果。针对现有调控算法因依赖模型而无法保证调控效果持续、稳定的问题，采用对模型依赖少、受扰动影响小的自抗扰控制，结合神经调控的特点，研究了1)变尺度扩张状态观测模型，优化扰动估计与噪声抑制；2)控制结构整形，优化调控响应；3)时间与误差绝对值乘积积分指标最佳以及神经调控综合指数最佳的调控参数二次优化，提高调控效果；进而保证调控效果的持续、稳定。项目提出了难治性癫痫无创闭环神经调控的多尺度优化自抗扰控制方法。通过项目研究，所得癫痫闭环调控策略为闭环神经调控研究带来了新的思路，成果有效提高了闭环神经调控效果，为复杂系统的控制与优化、神经与脑科学等相关学科的发展奠定了理论和技术基础。