基于Actor-Critic和神经网络的闭环脑机接口控制器设计

在皮层神经元放电活动模型的基础上进行单关节自发运动的研究,从控制理论的角度分析闭环脑机接口的工作原理.使用卡尔曼滤波器和人工神经网络设计系统的解码器替代原系统的脊髓电流,并且比较这两种解码器的优劣.由于在无感知反馈的信号下,解码器的性能下降得比较明显,使用强化学习中Actor-Critic算法结合人工神经网络设计PID控制器,用以产生刺激信号来刺激大脑皮层神经元,使其能够跟踪有感知反馈信号时皮层神经元的放电活动,从而恢复解码器的性能.最后,通过与其他控制算法对比,验证了基于强化学习算法的人工感知反馈信号设计的有效性.     

浅谈脑-机接口技术

脑机接口是近年来发展起来的一种人机接口,它不依赖于大脑的正常输出通路(即外围神经和肌肉组织),就可以实现人脑与外界(计算机或其它外部装置)直接通信的系统。本文概述了基于脑信号的脑机接口的基本结构,常用算法及应用前景。     

基于脑机接口与双激光雷达的移动车导航系统

针对脑机接口信噪比低、准确率差和延时长等问题,提出了基于机器智能辅助的室外移动机器人脑机接口导航方法.设计并实现了一个基于脑机接口与双激光雷达的移动车导航系统.该系统首先采用了基于双激光雷达的改进的角度势场法进行局部路径规划,然后结合脑机接口系统获取的导航意图,经过融合决策给出控制命令,驱动一辆经过机械系统改装的电动汽车.实验表明,该系统能根据环境障碍信息和脑机控制意图实现智能避障与人机协同导航,具有更高的准确性、容错性和鲁棒性.     

参数空间中鲁棒稳定性问题

本文研究了单输入多输出(SIMO)和多输入单输出(MISO)线性定常控制系统在对象 参数扰动下的鲁棒稳定性问题,并对一给定的控制器,系统在标称参数P°下闭环渐稳的情 况,提出了在参数空间中构造比中心位于P°的最大稳定超球还要大的稳定超球的方法,从而 改善了鲁棒稳定性判据.     

感知器算法在运动想象脑电模式识别中的应用

基于脑电(EEG)的脑机接口(BCI)是在人脑和计算机或其它电子设备之间建立不依赖于常规大脑信息输出通路(外周神经和肌肉组织)的全新对外信息交流和控制技术。及时有效地提取和识别与运动想象相关的脑电模式可以帮助严重瘫痪病人控制光标或辅助运动设备以替代其受损的运动功能,建立一种与外界交流沟通的新途径。论文将以EEG(C3,C4)两个通道的mu节律能量作为特征向量,用感知器算法对左右手运动想象脑电模式进行识别,实验结果表明,正确识别率可达87.86%。由于感知器算法计算简单,故可以认为,感知器算法在脑机接口的应用中有较高的实用价值。     

基于LS-SVM的非线性系统自适应输出反馈控制

针对状态不可测的单输入单输出非线性不确定系统,提出一种基于最小二乘支持向量机(LS-SVM)的直接自适应输出反馈控制方法.该方法首先设计一种误差观测器,间接地估计出系统的状态,然后采用最小二乘支持向量机构造自适应控制器,控制器参数的在线调整规律由李亚普诺夫稳定性理论导出.文中严格证明了闭环系统的渐近稳定性,仿真研究表明了此控制方法的可行性和有效性.     

基于模型的闭环系统故障检测的一种新方法

针对在闭环控制系统中,如何取得产生残差的最佳信号的问题,通过对单输入单输出闭环系统各种故障的Matlab仿真研究及结果分析,提出了一种基于模型的闭环系统故障检测的新方法。该方法把控制器的输出残差和系统的输出残差结合起来检测故障,并且考虑到了在实际系统中控制器可能饱和的实际情况。这种检测故障的新方法对于实际系统可能发生的各种故障(如执行器,传感器故障)更为有效,可以提高系统检测故障的性能,减少了系统故障的漏报率。     

磁耦合谐振无线电能传输系统的输出鲁棒控制

磁耦合谐振无线电能传输系统的负载和谐振参数会因为受到外界环境的影响而发生变化, 系统工作频率发生随机漂移, 导致模型参数存在不确定性. 针对参数不确定下的输出鲁棒控制问题, 本文基于H_∞控制理论, 应用Matlab鲁棒控制工具箱设计H_∞控制器, 并基于结构奇异值法分析了闭环系统的鲁棒稳定性与鲁棒性能. 结果表明, 在H_∞控制器的作用下, 实现了闭环摄动系统的输出鲁棒控制, 并为此类高阶非线性不确定闭环系统提供了一种通用的控制器设计方法.     

基于LS-SVM的一类非线性系统直接自适应H∞输出反馈控制

针对一类状态不可测的单输入单输出非线性不确定系统,提出了一种基于最小二乘支持向量机（LS-SVM）的直接自适应H∞输出反馈控制方法．该方法首先设计一种误差观测器,间接地估计出系统的状态;然后利用LS-SVM构造白适应控制器,并给出了LS-SVM权向量和偏移值的在线学习规则,通过引入如控制器减弱外部干扰及LS-SVM近似误差对输出误差的影响,利用李亚普诺夫理论证明了整个闭环系统的稳定性．仿真研究表明了该控制方案的可行性和有效性．     

含静止同步串联补偿器的单机系统非线性综合控制器设计

本文考虑含静止同步串联补偿器(SSSC)的单机无穷大系统、汽门、励磁各变量间的相互作用建立了4阶多变量3输入3输出非线性模型, 在此模型基础上, 运用动态逆系统方法完成解耦和线性化, 构造出3个单输入单输出的伪线性系统, 并采用变结构控制理论分别设计控制器. 根据MATLAB仿真结果验证了该控制方法对发电机功角, 机端电压和传输功率这3个控制指标有很好的控制效果.     

深海自持式智能浮标双闭环模糊PID定深控制

针对深海自持式智能浮标运动模型非线性、强耦合性的特点,提出了一种基于双闭环反馈回路的模糊比例-积分-微分（proportion-integral-derivative,PID）定深控制器.根据浮标的浮力调节机构,分析了浮标的运动过程,建立了非线性运动方程.针对外环深度反馈回路,设计了模糊控制器.基于内环速度反馈回路与模糊控制器,设计了联级模糊PID定深控制器.传统PID定深控制器超调量5.6%,最终在目标深度±30 m范围内震荡,而双闭环模糊PID定深控制器在相同的上升时间内,超调量2.0%,深度误差控制在1.0%以内.存在外界扰动的情况下,通过双闭环模糊PID定深控制器的调节,浮标仍可以稳定在目标深度内.仿真结果表明,所建立的双闭环模糊PID定深控制系统具有良好的控制效果和稳定性.     

Simple control method for integrating processes with long deadtime

Control of integrating processes with long deadtime is a challenging problem. Using original Smith predictor control structure will result in an offset problem during load disturbance. Using the internal model control (IMC) principle to design the control structure will result in large overshoot during servo response. In this work, a simple modified Smith predictor controller design is proposed for this important type of system. The overall control structure has only two physically meaningful tuning parameters. One is used to set the speed of the closed-loop servo response and the other tuning parameter is used to set the speed of the closed-loop load response. Two examples are used to demonstrate the improved closed-loop performance of the proposed controller design.     

基于内模控制的改进型三相PWM整流器

针对三相电压型PWM整流器的双闭环控制系统直流母线电压超调较大、对扰动克服能力不强、鲁棒性差等问题,本文对控制系统进行了改进,在电压外环设计一种基于内模原理的控制器以取代传统PI控制器,并对内模控制器中滤波器部分的参数设计进行分析,最后使用MATLAB/SIMULINK进行仿真研究,并与电压外环采用PI控制器的系统性能进行比较。仿真结果表明,在电压环采用内模控制器,可以有效降低直流母线电压超调,快速抑制扰动,具有良好的鲁棒性,性能优于传统PI调节器。     

An improved state-space model structure and a corresponding predictive functional control design with improved control performance

Conventional state-space model predictive control requires a state estimator/observer to access the state information for feedback controller design. Its drawbacks are the numerical convergence stability of the observer and closed-loop control performance deterioration with activated plant input/output constraints. The recent direct use of measured input and output variables to formulate a non-minimal state-space (NMSS) model overcomes these problems, but the subsequent controller is too sensitive to model mismatch. In this article, an improved structure of NMSS model that incorporates the output-tracking error is first formulated and then a subsequent predictive functional control design is proposed. The proposed controller is tested on both model match and model mismatch cases for comparison with previous controllers. Results show that control performance is improved. In addition, a linear programming method for constraints dealing and a closed form of transfer function representation of the control system are provided for further insight into the proposed method.     

基于PID和LQR的四旋翼无人机控制系统研究

为提高四旋翼无人机的飞行稳定性、无人飞行器控制系统的鲁棒性和控制精度,以建立的四旋翼无人机飞行控制系统模型为基础,采用现代控制理论与传统控制论相结合的方法,针对姿态角速率、姿态角分别设计内环LQR(线性二次型调节器)控制器,及外环PID控制的双回路闲环控制器.充分利用PID控制器易于掌握且对模型要求精度低、LQR控制器能改善内回路的动态特性和稳态性能的特点,完成四旋翼无人机的飞行控制.通过实验遴选该双闭环控制器相关参数并进行优化,实验结果表明所设计的双回路控制器控制性能指标良好.     

Composite nonlinear feedback design for discrete-time switching systems with disturbances and input saturation

This paper addresses the problem of robust controller design for a class of discrete-time switching systems with input saturation. To this aim, the composite nonlinear feedback method is extended to design a robust controller with improved performances in terms of the response speed and overshoot in the presence of disturbances and input saturation. The proposed approach is theoretically analysed and its closed-loop stability is proved. Then, the performance of the proposed method is verified using numerical simulations.     

Wind turbine integrated structural and LPV control design for improved closed-loop performance

An iterative redesign algorithm is proposed to integrate the design of the structural parameters and a linear parameter-varying (LPV) controller for a three-bladed horizontal-axis wind turbine. The LPV controller is designed for an eighth-order lumped model of the wind turbine consisting of blades, drive-train and the tower. The lumped model response is matched with detailed open-loop numerical simulations using the Fatigue, Aerodynamics, Structures and Turbulence (FAST) code. The controller is scheduled in real-time based on the mean wind speed to account for the varying system dynamics. The objective is to track the operating trajectory meanwhile minimise the H _∞ performance index from the wind turbulence to the controlled output vector consisting of pitch angle, blade tip deflection, and the generator speed and torque. Sensitivity analysis of the closed-loop performance index with respect to the structural parameters of the system is examined. The integrated design problem is formulated as an iterative sequential controller/structure redesign to obtain the structural parameters and controller matrices corresponding to a local optimal performance index. Each step of the iterative procedure is formulated as a linear matrix inequality (LMI) optimisation problem that can be solved efficiently using available LMI solvers. The evolution of the structural parameters and performance index through the integrated design is illustrated. The FAST closed-loop simulations for two selected designs with the smallest values of the performance index demonstrate the improved performance of the overall system through the integrated structure/control redesign in both minimising the effect of the wind disturbance on the generator output power, and reducing the structural loads on the wind turbine.     

Optimal probability density function control for NARMAX stochastic systems

This paper presents a new control strategy for a class of non-Gaussian stochastic systems so that the output probability density function (PDF) of the system can be made to follow a desired PDF. The system considered is represented by an Nonlinear AutoRegressive and Moving Average with eXogenous (NARMAX) inputs with input channel time-delay and non-Gaussian noise. A multi-step-ahead nonlinear cumulative cost function is used to improve tracking performance. For this purpose, a relationship between the PDFs of all the inputs and the PDFs of multiple-step-ahead output is formulated by constructing an auxiliary multivariate mapping. By minimizing this performance function, a new explicit predictive controller design algorithm is established with less conservatism than some previous results. Furthermore, an improved approach is developed to guarantee the local stability of the closed-loop system by tuning the weighting parameters recursively. Simulations are given to demonstrate the effectiveness of the proposed control algorithm and desired results have been obtained.     

Design of decentralized controllers for MIMO processes

In this paper, the design of decentralized PID controller for interactive and delay time processes is proposed. Direct controller synthesis approach is used to obtain parameters of the multi-loop controllers. A detuning factor for each loop is specified based on the closed-loop time constant. Then appropriate controller settings are determined using the Maclaurin series expansion and the model of the process. The success of the proposed algorithm is based on the detuning parameter thus the selection of the desired closed-loop time constant is an important factor. The proposed method has no limitations regarding systems weak or strong interactions, multiple delays, and its operating behaviors such as monotonic or oscillatory. The proposed algorithm is simpler and improved performance of various processes are achieved as compared to some of the prevalent techniques. Simulation examples demonstrate the usefulness of the proposed method.