改进协同微粒群优化的模糊神经网络控制系统设计

针对协同微粒群算法不能保证收敛到局部或全局最优值的问题,提出一种改进协同微粒群算法(ICPSO),并证明了该算法能以概率1收敛干全局最优解.应用ICPSO建立一类非线性对象的神经网络辨识模型,并对系统的模糊神经网络自适应控制器的参数进行了离线和在线优化.仿真结果表明,ICPSO能提高系统的建模精度,增强模型的泛化能力,而且由ICPSO训练的控制器可以达到良好的控制效果.     

广义双线性系统的终端滑模控制

研究广义双线性系统的终端滑模变结构控制问题.基于Lyapunov稳定性理论,运用Lyapunov函数方法,给出广义双系统的终端滑动模超出面,设计相应的终端滑模变结构控制器,使得闭环系统渐进稳定,实现滑动模运动,保证系统状态在有限时间内到达平衡点,得到广义双线性系统全局稳定的充分条件.所给的可行性算例,说明这一方法的有效性与可行性.     

一种新的基于遗忘因子的递推子空间辨识算法

针对工业系统中广泛存在的时变特性, 提出一种新的递推子空间辨识算法, 实现对系统状态空间模型的在线递推估计. 为更好地跟踪系统时变特性, 研究基于遗忘因子的输入输出数据矩阵构造机制, 以提高递推算法的收敛速度; 针对算法中奇异值分解的求解问题, 将梯度型算法引入基于遗忘因子的状态子空间跟踪中, 实现对广义能观测矩阵的估计, 避免了子空间近似带来的估计有偏性; 该算法计算简单有效, 且对初值具有更高的鲁棒性; 最后给出该递推算法的性能分析, 理论证明其收敛性, 并通过仿真实例验证算法的有效性.     

一种量子神经网络模型学习算法及应用

提出一种量子神经网络模型及学习算法. 首先基于生物神经元信息处理机制和量子计算原理构造出一种量子神经元, 该神经元由加权、聚合、活化、激励四部分组成. 然后由量子神经元构造出三层量子神经网络模型, 其输入和输出为实值向量, 权值和活性值为量子比特. 基于梯度下降法构造了该模型的超线性收敛学习算法. 通过模式识别和函数逼近两种仿真结果表明该模型及算法是有效的.     

一种求解Shubert函数优化问题的演化算法

综合国内外演化计算研究现状,基于热力学中的自由能极小化原理, 设计了一个全新的热力学演化算法,并通过对于Shubert函数优化问题求解的数值试验,测试了热力学演化算法的优良性能,实验结果表明了热力学演化算法求出的解比一般演化算法求出的解更加接近于全局最优。     

A method for numerical solution of a multidimensional convection-diffusion problem

We propose a modification of the additive splitting algorithm to solve the convection-diffusion problem using an efficient finite-difference scheme. The modification decreases the number of data exchanges and their amount during the numerical solution of a system of multidimensional equations. Approximation, stability, and convergence are considered. Translated from Kibernetika i Sistemnyi Analiz, No. 1, pp. 100–107, January–February 2009.     

A new SQP algorithm based on semidefinite programming

The paper gives a new filter-SQP algorithm of the semidefinite programming through combining the sequence nonlinear algorithm of solving the semidefinite programming, low rank conversion, and filter method, which, under certain conditions, proves that the algorithm is global convergence.     

The Holy Alignment: Geodesic and Astronomical Fundamentals for Calculating the Adjustment of Medieval Naves

It is commonly believed that the longitudinal axes of churches extend exactly in an east-west direction. However, thorough investigations have shown that this is not always correct; rather, both southern and northern deviations of up to about 25° can occur. The angular deviation between the church axis and true east is called the Holy Alignment. This present study presents the possibility that the nave is oriented towards the direction of the sun-rising point on the name day of the patron saint of the church. If several saints share the patronage, the Holy Alignment equals the algebraic sum of the angular distances for each saint. The orientation of a nave can be analyzed by means of common mathematical relations used in geodesy, astronomy and gnomonics. In order to perform such an analysis, it is necessary to know the history of the patronage of the church; the Gauss-Krüger coordinates of the ground plan; and characteristic astronomical quantities at the time the church was built. A calculated example with the saints Andrew, James and Philip for the year 980 illustrates the analysis. An appendix deals with the influence of atmospheric refraction on the apparent altitude of the sun near the horizon.     

Approximation processes for resolvent operators

We introduce general sequences of linear operators obtained from classical approximation processes which are useful in the approximation of the resolvent operators of the generators of suitable C ₀-semigroups. The main aim is the representation of the resolvent operators in terms of classical approximation operators. Work performed under the auspices of PRIN 2006–07 “Kolmogorov equations” (coordinator G. Da Prato)     

The learning convergence of CMAC in cyclic learning

In this paper we discuss the learning convergence of the cerebellar model articulation controller (CMAC) in cyclic learning.We prove the following results.First,if the training samples are noiseless,the training algorithm converges if and only if the learning rate is chosen from (0,2).Second,when the training samples have noises,the learning algorithm will converge with a probability of one if the learning rate is dynamically decreased.Third,in the case with noises,with a small but fixed learning rate ε the mean square error of the weight sequences generated by the CMAC learning algorithm will be bounded by O(ε).Some simulation experiments are carried out to test these results.