2006年 第32卷 第2期
2006, 32(2): 161-169.
摘要:
The maximum principle for fully coupled forward-backward stochastic control system in the global form is proved, under the assumption that the forward diffusion coefficient does not contain the control variable, but the control domain is not necessarily convex.
The maximum principle for fully coupled forward-backward stochastic control system in the global form is proved, under the assumption that the forward diffusion coefficient does not contain the control variable, but the control domain is not necessarily convex.
2006, 32(2): 170-178.
摘要:
The problem of track control is studied for a class of strict-feedback stochastic nonlinear systems in which unknown virtual control gain function is the main feature. First, the so-called stochastic LaSalle theory is extended to some extent, and accordingly, the results of global ultimate boundedness for stochastic nonlinear systems are developed. Next, a new design scheme of fuzzy adaptive control is proposed. The advantage of it is that it does not require priori knowledge of virtual control gain function sign, which is usually demanded in many designs. At the same time, the track performance of closed-loop systems is improved by adaptive modifying the estimated error upper bound. By theoretical analysis, the signals of closed-loop systems are globally ultimately bounded in probability and the track error converges to a small residual set around the origin in 4th-power expectation.
The problem of track control is studied for a class of strict-feedback stochastic nonlinear systems in which unknown virtual control gain function is the main feature. First, the so-called stochastic LaSalle theory is extended to some extent, and accordingly, the results of global ultimate boundedness for stochastic nonlinear systems are developed. Next, a new design scheme of fuzzy adaptive control is proposed. The advantage of it is that it does not require priori knowledge of virtual control gain function sign, which is usually demanded in many designs. At the same time, the track performance of closed-loop systems is improved by adaptive modifying the estimated error upper bound. By theoretical analysis, the signals of closed-loop systems are globally ultimately bounded in probability and the track error converges to a small residual set around the origin in 4th-power expectation.
2006, 32(2): 179-185.
摘要:
The admissibility analysis and robust admissible control problem of the uncertain discrete- time switched linear singular (SLS) systems for arbitrary switching laws are investigated. Based on linear matrix inequalities, some sufficient conditions are given for: A) the existence of generalized common Lyapunov solution and the admissibility of the SLS systems for arbitrary switching laws, B) the existence of static output feedback control laws ensuring the admissibility of the closed-loop SLS systems for arbitrary switching laws and norm-bounded uncertainties.
The admissibility analysis and robust admissible control problem of the uncertain discrete- time switched linear singular (SLS) systems for arbitrary switching laws are investigated. Based on linear matrix inequalities, some sufficient conditions are given for: A) the existence of generalized common Lyapunov solution and the admissibility of the SLS systems for arbitrary switching laws, B) the existence of static output feedback control laws ensuring the admissibility of the closed-loop SLS systems for arbitrary switching laws and norm-bounded uncertainties.
2006, 32(2): 186-192.
摘要:
The robust exponential stability of a class of discrete time impulsive switched systems with structure perturbations is studied. Based on the average dwell time concept and by dividing the total activation time into the time with stable subsystems and the time with unstable subsystems, it is shown that if the average dwell time and the activation time ratio are properly large, the given switched system is robustly exponentially stable with a desired stability degree. Compared with the traditional Lyapunov methods, our layout is more clear and easy to carry out. Simulation results validate the correctness and effectiveness of the proposed algorithm.
The robust exponential stability of a class of discrete time impulsive switched systems with structure perturbations is studied. Based on the average dwell time concept and by dividing the total activation time into the time with stable subsystems and the time with unstable subsystems, it is shown that if the average dwell time and the activation time ratio are properly large, the given switched system is robustly exponentially stable with a desired stability degree. Compared with the traditional Lyapunov methods, our layout is more clear and easy to carry out. Simulation results validate the correctness and effectiveness of the proposed algorithm.
2006, 32(2): 193-199.
摘要:
This paper combines image processing with 3D magnetic tracking method to develop a scalpel for haptic simulation in surgical cutting. First, a cutting parameter acquisition setup is presented and the performance is validated from soft tissue cutting. Then, based on the acquired input-output data pairs, a method for fuzzy system modeling is presented, that is, after partitioning each input space equally and giving the premises and the total number of fuzzy rules, the consequent parameters and the fuzzy membership functions (MF) of the input variables are learned and optimized via a neurofuzzy modeling technique. Finally, a haptic scalpel implemented with the established cutting model is described. Preliminary results show the feasibility of the haptic display system for real-time interaction.
This paper combines image processing with 3D magnetic tracking method to develop a scalpel for haptic simulation in surgical cutting. First, a cutting parameter acquisition setup is presented and the performance is validated from soft tissue cutting. Then, based on the acquired input-output data pairs, a method for fuzzy system modeling is presented, that is, after partitioning each input space equally and giving the premises and the total number of fuzzy rules, the consequent parameters and the fuzzy membership functions (MF) of the input variables are learned and optimized via a neurofuzzy modeling technique. Finally, a haptic scalpel implemented with the established cutting model is described. Preliminary results show the feasibility of the haptic display system for real-time interaction.
2006, 32(2): 200-206.
摘要:
This paper aims at two problems which exist in most of repairable spare part demand models at present: the exponential distribution as the basic assumption and one typical distribution corresponding to a model. A general repairable spare part demand model built on quasi birth-and- death process is developed. This model assumes that both the operational time of the unit and the maintenance time of the unit follow the continuous time phase type distributions. The first passage time distribution to be out of spares, the first mean time to be out of spares, and an algorithm to get the minimal amount of spares under certain restrictions are obtained. At the end of this paper, a numerical example is given.
This paper aims at two problems which exist in most of repairable spare part demand models at present: the exponential distribution as the basic assumption and one typical distribution corresponding to a model. A general repairable spare part demand model built on quasi birth-and- death process is developed. This model assumes that both the operational time of the unit and the maintenance time of the unit follow the continuous time phase type distributions. The first passage time distribution to be out of spares, the first mean time to be out of spares, and an algorithm to get the minimal amount of spares under certain restrictions are obtained. At the end of this paper, a numerical example is given.
2006, 32(2): 207-212.
摘要:
A robust stabilization problem is considered for time delay nonlinear discrete-time systems based on T-S fuzzy model. A necessary and sufficient condition for the existence of such controllers is given through Lyapunov stability theorem. And it is further shown that this condition is equivalent to the solvability of a certain linear matrix inequality, which can be solved easily by using the LMI toolbox of Matlab. At last, an illustrative example of truck-trailer is presented to show the feasibility and effectiveness of the proposed method.
A robust stabilization problem is considered for time delay nonlinear discrete-time systems based on T-S fuzzy model. A necessary and sufficient condition for the existence of such controllers is given through Lyapunov stability theorem. And it is further shown that this condition is equivalent to the solvability of a certain linear matrix inequality, which can be solved easily by using the LMI toolbox of Matlab. At last, an illustrative example of truck-trailer is presented to show the feasibility and effectiveness of the proposed method.
2006, 32(2): 213-218.
摘要:
This paper discusses the problem of the H∞ filtering for discrete time 2-D singular Roesser models (2-D SRM). The purpose is to design an observer-based 2-D singular filter such that the error system is acceptable, jump modes free and stable, and satisfies a pre-specified H∞ performance level. By general Riccati inequality and bilinear matrix inequalities (BMI), a sufficient condition for the solvability of the observer-based H∞ filtering problem for 2-D SRM is given. A numerical example is provided to demonstrate the applicability of the proposed approach. Key words 2-D singular systems, jump modes, general Riccati inequality, bilinear matrix i
This paper discusses the problem of the H∞ filtering for discrete time 2-D singular Roesser models (2-D SRM). The purpose is to design an observer-based 2-D singular filter such that the error system is acceptable, jump modes free and stable, and satisfies a pre-specified H∞ performance level. By general Riccati inequality and bilinear matrix inequalities (BMI), a sufficient condition for the solvability of the observer-based H∞ filtering problem for 2-D SRM is given. A numerical example is provided to demonstrate the applicability of the proposed approach. Key words 2-D singular systems, jump modes, general Riccati inequality, bilinear matrix i
2006, 32(2): 219-227.
摘要:
研究了基于比例控制分量非线性方法合成的新型非线性PID控制器(简记为:NPID-PCA).该控制器将比例控制分量表示为误差信号的非线性函数,并通过三个独立非线性函数构造合成PID控制器。由于比例控制分量具有最简的非线性特征,因此该类控制器在设计上更为简洁,非线性在控制作用方面的物理意义更为明确.采用样条函数[7]生成非线性作用曲线,用户可以通过图形化交互方式直观地实现非线性设计.通过与文献中的控制实例比较,本文提出的新型非线性PID控制器可以实现更为优良的控制性能指标,包括对带有时延或死区的被控对象.
研究了基于比例控制分量非线性方法合成的新型非线性PID控制器(简记为:NPID-PCA).该控制器将比例控制分量表示为误差信号的非线性函数,并通过三个独立非线性函数构造合成PID控制器。由于比例控制分量具有最简的非线性特征,因此该类控制器在设计上更为简洁,非线性在控制作用方面的物理意义更为明确.采用样条函数[7]生成非线性作用曲线,用户可以通过图形化交互方式直观地实现非线性设计.通过与文献中的控制实例比较,本文提出的新型非线性PID控制器可以实现更为优良的控制性能指标,包括对带有时延或死区的被控对象.
2006, 32(2): 228-236.
摘要:
为实现汉语全文词义自动标注，本文采用了一种新的基于无指导机器学习策略的词义标注方法。实验中建立了四个词义排歧模型，并对其测试结果进行了比较.其中实验效果最优的词义排歧模型融合了两种无指导的机器学习策略，并借助依存文法分析手段对上下文特征词进行选择.最终确定的词义标注方法可以使用大规模语料对模型进行训练，较好的解决了数据稀疏问题，并且该方法具有标注正确率高、扩展性能好等优点，适合大规模文本的词义标注工作.
为实现汉语全文词义自动标注，本文采用了一种新的基于无指导机器学习策略的词义标注方法。实验中建立了四个词义排歧模型，并对其测试结果进行了比较.其中实验效果最优的词义排歧模型融合了两种无指导的机器学习策略，并借助依存文法分析手段对上下文特征词进行选择.最终确定的词义标注方法可以使用大规模语料对模型进行训练，较好的解决了数据稀疏问题，并且该方法具有标注正确率高、扩展性能好等优点，适合大规模文本的词义标注工作.
2006, 32(2): 237-245.
摘要:
针对一类带未知输入时滞参数的非线性时滞系统,基于线性矩阵不等式LMI和一种基于"Descriptor form"的Lyapunov-Krasovskii泛函方法,设计了一种带未知时滞估计项的记忆型状态反馈控制器,给出了针对该未知输入时滞参数的自适应控制H∞控制器设计方案,并对闭环系统的稳定性给出了证明.
针对一类带未知输入时滞参数的非线性时滞系统,基于线性矩阵不等式LMI和一种基于"Descriptor form"的Lyapunov-Krasovskii泛函方法,设计了一种带未知时滞估计项的记忆型状态反馈控制器,给出了针对该未知输入时滞参数的自适应控制H∞控制器设计方案,并对闭环系统的稳定性给出了证明.
2006, 32(2): 246-254.
摘要:
本文研究线性终端状态约束下不定随机线性二次最优控制问题.首先利用Lagrange Multiplier 定理得到了存在最优线性状态反馈解的必要条件, 而在加强的条件下也得到了最优控制存在的充分条件. 从某种意义上讲, 以往关于无约束随机线性二次最优控制的一些结果可以看成本文主要定理的推论.
本文研究线性终端状态约束下不定随机线性二次最优控制问题.首先利用Lagrange Multiplier 定理得到了存在最优线性状态反馈解的必要条件, 而在加强的条件下也得到了最优控制存在的充分条件. 从某种意义上讲, 以往关于无约束随机线性二次最优控制的一些结果可以看成本文主要定理的推论.
2006, 32(2): 255-262.
摘要:
基于距离变换的骨架算法往往不能直接用于骨架识别，且骨架的连通性难以保证.本文提出一种新型的骨架算法，由一个初始骨架点开始逐点生长出各骨架分支，同时在骨架生长过程中用离散曲线演化模型消除造成信息冗余的骨架枝，保留视觉上重要的骨架枝，实现了骨架的多尺度控制,实验证明本算法复杂度低，得到的骨架连通性得到保证，能较好地表示图形中视觉重要成分，符合人类视觉习惯，可直接用于图形识别和形状度量.
基于距离变换的骨架算法往往不能直接用于骨架识别，且骨架的连通性难以保证.本文提出一种新型的骨架算法，由一个初始骨架点开始逐点生长出各骨架分支，同时在骨架生长过程中用离散曲线演化模型消除造成信息冗余的骨架枝，保留视觉上重要的骨架枝，实现了骨架的多尺度控制,实验证明本算法复杂度低，得到的骨架连通性得到保证，能较好地表示图形中视觉重要成分，符合人类视觉习惯，可直接用于图形识别和形状度量.
2006, 32(2): 263-270.
摘要:
多处理器、多模式等调度问题的出现，标志着调度问题正呈现出更加复杂多样、更强调柔性的发展趋势.本文采用嵌套的微结构方法，对现有调度问题进行统一表述，并且能够表述一些传统方法无法表述的复杂问题.针对主从微结构，本文用总柔性度（GFD）作为调度问题柔性的衡量指标，并按照GFD的分类给出了求解调度问题的线性规划方法.定量实验不仅表明适当的柔性能有效优化调度问题的最优解，而且针对车间设备群规模合理性给出了统计结论.
多处理器、多模式等调度问题的出现，标志着调度问题正呈现出更加复杂多样、更强调柔性的发展趋势.本文采用嵌套的微结构方法，对现有调度问题进行统一表述，并且能够表述一些传统方法无法表述的复杂问题.针对主从微结构，本文用总柔性度（GFD）作为调度问题柔性的衡量指标，并按照GFD的分类给出了求解调度问题的线性规划方法.定量实验不仅表明适当的柔性能有效优化调度问题的最优解，而且针对车间设备群规模合理性给出了统计结论.
2006, 32(2): 271-278.
摘要:
讨论奇异系统的不定号LQ问题 (二次型指标中的权矩阵含有负特征值的最优控制问题). 首先指出问题的可解性, 并给出了问题等价转化为奇异系统的奇异LQ问题的充要条件. 然后基于等价的奇异系统奇异LQ问题, 给出问题存在唯一最优控制—轨线对的充分条件. 最后用一个算例说明结论的正确性.
讨论奇异系统的不定号LQ问题 (二次型指标中的权矩阵含有负特征值的最优控制问题). 首先指出问题的可解性, 并给出了问题等价转化为奇异系统的奇异LQ问题的充要条件. 然后基于等价的奇异系统奇异LQ问题, 给出问题存在唯一最优控制—轨线对的充分条件. 最后用一个算例说明结论的正确性.
2006, 32(2): 279-285.
摘要:
精确目标定位和手术路径规划是将机器人用于长骨骨折内固定手术中的关键技术.文中使用了局部法对X光图像进行畸变校正；并提出了一种基于几何模型的对准方法，得到一条能够穿过髓内钉远端孔的路径，从而引导串联机器人锁定髓内钉.实验表明，该导航方法仅需一张X光图像，有足够的定位精度和稳定性，可大大减少手术时间，降低医生所受辐射剂量.
精确目标定位和手术路径规划是将机器人用于长骨骨折内固定手术中的关键技术.文中使用了局部法对X光图像进行畸变校正；并提出了一种基于几何模型的对准方法，得到一条能够穿过髓内钉远端孔的路径，从而引导串联机器人锁定髓内钉.实验表明，该导航方法仅需一张X光图像，有足够的定位精度和稳定性，可大大减少手术时间，降低医生所受辐射剂量.
2006, 32(2): 286-290.
摘要:
Based on the optimal fusion algorithm weighted by matrices in the linear minimum variance (LMV) sense, a distributed full-order optimal fusion Kalman filter (DFFKF) is given for discrete-time stochastic singular systems with multiple sensors, which involves the inverse of a high-dimension matrix to compute matrix weights. To reduce the computational burden, a distributed reduced-order fusion Kalman filter (DRFKF) is presented, which involves in parallel the inverses of two relatively low-dimension matrices to compute matrix weights. A simulation example shows the effectiveness.
Based on the optimal fusion algorithm weighted by matrices in the linear minimum variance (LMV) sense, a distributed full-order optimal fusion Kalman filter (DFFKF) is given for discrete-time stochastic singular systems with multiple sensors, which involves the inverse of a high-dimension matrix to compute matrix weights. To reduce the computational burden, a distributed reduced-order fusion Kalman filter (DRFKF) is presented, which involves in parallel the inverses of two relatively low-dimension matrices to compute matrix weights. A simulation example shows the effectiveness.
2006, 32(2): 291-295.
摘要:
The problem of adaptive robust control is addressed for a class of neutral delay systems. All uncertainties are assumed to be bounded by unknown constants. An improved adaptation law is proposed to estimate the square of these unknown bounds. Then, by making use of the updated values of the squared unknown bounds, an adaptive controller is designed to make the solution of the resultant closed-loop system uniformly ultimately bounded. Furthermore, this method avoids chattering and improves the performance. An example is given to illustrate the effectiveness of this method.
The problem of adaptive robust control is addressed for a class of neutral delay systems. All uncertainties are assumed to be bounded by unknown constants. An improved adaptation law is proposed to estimate the square of these unknown bounds. Then, by making use of the updated values of the squared unknown bounds, an adaptive controller is designed to make the solution of the resultant closed-loop system uniformly ultimately bounded. Furthermore, this method avoids chattering and improves the performance. An example is given to illustrate the effectiveness of this method.
2006, 32(2): 296-303.
摘要:
提出一种新的蚁群算法，将信息素分成有限个级别，通过级别的更新实现对信息素的更新，并且信息素的更新量独立于目标函数值. 文中采用有限马氏链的理论证明算法可以线性地收敛到全局最优解. 针对TSP问题，通过与MMAS和ACS等蚁群算法的数值实验结果进行比较，表明所提出的算法是有效的、鲁棒的.
提出一种新的蚁群算法，将信息素分成有限个级别，通过级别的更新实现对信息素的更新，并且信息素的更新量独立于目标函数值. 文中采用有限马氏链的理论证明算法可以线性地收敛到全局最优解. 针对TSP问题，通过与MMAS和ACS等蚁群算法的数值实验结果进行比较，表明所提出的算法是有效的、鲁棒的.
2006, 32(2): 304-310.
摘要:
离线鲁棒预测控制综合算法离线确定一个控制律序列，对应一组吸引域，在线根据当前状态的位置选择相应的控制律,该类控制器在线计算量非常小，而可行性和最优性与其它综合算法相比或多或少要差一些,为此，采用标称性能指标而不是“最坏情况”性能指标来改进离线综合算法的可行性和最优性,改进的控制器保持了原有控制器的稳定性以及控制律关于系统状态的连续性.仿真结果说明了采用标称性能指标的优越性。
离线鲁棒预测控制综合算法离线确定一个控制律序列，对应一组吸引域，在线根据当前状态的位置选择相应的控制律,该类控制器在线计算量非常小，而可行性和最优性与其它综合算法相比或多或少要差一些,为此，采用标称性能指标而不是“最坏情况”性能指标来改进离线综合算法的可行性和最优性,改进的控制器保持了原有控制器的稳定性以及控制律关于系统状态的连续性.仿真结果说明了采用标称性能指标的优越性。
2006, 32(2): 311-317.
摘要:
软测量是一种有效的在线测量技术，它在工业生产中受到越来越广泛的重视。作为实施软测量的前提，辅助变量的选择至关重要，它将直接影响软测量的效率. 灰关联分析是一种研究变量间相关性的新方法，它适于求解不确定性系统中小样本数据之间的关联程度. 深入研究了软测量辅助变量选择过程中存在的问题，基于灰关联分析，提出了一致关联度算法，对辅助变量进行选择。并给出了一种实用的冗余变量剔除方法. 同时，结合仿真实例对其有效性进行验证.仿真结果表明该方法是有效可行的，有助于提高软测量的精度.
软测量是一种有效的在线测量技术，它在工业生产中受到越来越广泛的重视。作为实施软测量的前提，辅助变量的选择至关重要，它将直接影响软测量的效率. 灰关联分析是一种研究变量间相关性的新方法，它适于求解不确定性系统中小样本数据之间的关联程度. 深入研究了软测量辅助变量选择过程中存在的问题，基于灰关联分析，提出了一致关联度算法，对辅助变量进行选择。并给出了一种实用的冗余变量剔除方法. 同时，结合仿真实例对其有效性进行验证.仿真结果表明该方法是有效可行的，有助于提高软测量的精度.
2006, 32(2): 318-320.
摘要:
This is a brief review on the recent book: Duality System in Applied Mechanics and Optimal Control, by Zhong Wan-Xie, published by Kluwer Academic Publishers, 2004. The book represents a significant effort to re-establish the historic and deep tie between control and mechanics by striving to connect and integrate concepts, methods, and algorithms in mechanics and control so that a unified framework can be established for both analytical and computational purposes. Clearly, it has demonstrated that the duality system method can be used as a mathematical and systematic foundation to deal with many important concepts and problems in both mechanics and control. This book is not only very useful for research and applications, but also extremely helpful for multidisciplinary curriculum development when students from one field are trying to learning and applying concepts and methods from the other field.
This is a brief review on the recent book: Duality System in Applied Mechanics and Optimal Control, by Zhong Wan-Xie, published by Kluwer Academic Publishers, 2004. The book represents a significant effort to re-establish the historic and deep tie between control and mechanics by striving to connect and integrate concepts, methods, and algorithms in mechanics and control so that a unified framework can be established for both analytical and computational purposes. Clearly, it has demonstrated that the duality system method can be used as a mathematical and systematic foundation to deal with many important concepts and problems in both mechanics and control. This book is not only very useful for research and applications, but also extremely helpful for multidisciplinary curriculum development when students from one field are trying to learning and applying concepts and methods from the other field.
