算法收敛

• The following is what i have done in this paper : 1 . the paper systematically summed up the fundamental knowledge of fuzzy set theory , and introduced fuzzy clustering algorithms in detail , and analyzed the defects of low convergence speed and sensitivity to the initialization of fuzzy clustering algorithms , and proposed a modified fuzzy clustering algorithm based on ga
  对模糊理论的基本内容进行了系统的总结和介绍，并详细介绍了模糊聚类算法，分析了模糊聚类算法收敛速度慢且对初始化很敏感的原因，引入了遗传算法，提出了一种改进的模糊聚类算法。
• The problem is then fed to a nonlinear programming routine to solve for all parameters . the flight dynamics model with yawing motion of uh - 60a is given , scale factors of state variables and control variables are applied based on the consideration of algorithm convergence . then , the optimization of trajectories is expressed as optimal control problem with a performance index constrained by helicopter performance limits , faa regulation and so on
  首先,建立带偏航运动的直升机点质量模型,然后对状态变量和控制变量进行无量纲处理,因为状态变量和控制变量量值相差较大,这使得优化算法收敛效果变差,所以需要选择适当的缩放因子,最后把直升机单台发动机停车后的飞行轨迹表示成一个最优控制问题,即在考虑直升机性能限制和far规范后,指定一个优化性能指标,选择合适的路径约束,然后再解最优控制问题。
• The numerical simulation of the cabin indicates : the normal k - model is adaptive to simulate the air flow ; the simplen algorithm can satisfy the requirement of calculation for its rapid convergence velocity ; the second order accurate supg in advection discretization scheme can attain more precise result
  模拟结果得出如下结论：标准k -模型对系统流动模拟是适合的； simplen算法收敛更快，满足运算的需要；采用二阶supg的对流项分离方案可以获得更高的计算精度。
• The main contributions of this dissertation are summarized as follow : ( 1 ) an ilc approach combining feedforward with current feedback is developed based on optimal feedback control and the gradient method . a sufficient condition that guarantees the convergences is given for linear system . the procedures of designing the algorithm can employ lqr , h2 or h approaches to improve the convergence rate of learning in iterations
  本文的主要成果有： 1 、在开闭环综合迭代学习控制结构的基础上，分析了利用梯度下降法设计前馈迭代学习控制器时，为保证算法的收敛性，闭环控制系统应该满足的充分条件，并依据提高算法收敛速率的优化条件，给出了基于lqr 、 h _ 2和h等优化控制技术的迭代学习控制算法的设计方法。
• It will eliminate the handicap of slow speed and elevate the correction of recognizing flow type by displacing the traditional arithmetic . it will provide a new method of creating model and measurement to the special parameters in some difficult system . this article is also the development and exploration of two - phase flow measurement and information fusion
  利用信息融合代替传统的数学建模方法，这将克服以往算法收敛速度慢等缺陷，并能提高流型识别的准确率，这将为复杂系统中难测参数的建模和检测提供新途径，本论文也是对两相流检测和信息融合的一个发展和探索。
• Based on a brief introduction of the theory and development of v90 modem , this thesis puts the emphasis on the analysis and research for the key technology of v90 modem , discusses the applications of the adaptive filter algorithms with v90 modern . least mean square ( lms ) algorithms are discussed in detail when they are used for v90 modem . with analysis of the relationships of snr , step size and convergent speed of lms algorithms , the paper presents the simulated results of adaptive equalization and echo cancellation with lms algorithms
  本文概要介绍了modem技术的发展及v90modem的原理，针对v90modem的技术特点和关键技术，深入讨论了自适应滤波算法在v90modem中的应用，研究了最小均方( lms )算法用于v90modem的一些重要问题，分析了信噪比、步长等参数与lms算法收敛速度的关系，给出了lms算法用于自适应均衡、回声对消的分析和模拟仿真结果，针对viterbi译码技术，分析了信噪比与误码率的关系。
• Finally we study a modified algorithm , which applies the fast fixed - point algorithm for complex signals to blind signals eparation in frequency domain . it ’ s a one - unit algorithm . to prevent converging to the same signal , the deflation algorithm is used to separate signals one by one
  4 .结合上述卷积混合信号的频域分离算法,研究了一个改进单源算法,将复数快速固定点算法应用到频域加速收敛,为了防止算法收敛于同一信号,利用一种抽气技术,这样可一次分离出一个信号,实现卷积混合信号分离。
• This algorithm has conquered the deficiencies of the traditional optimization methods that require calculating the derivative of the objective function and that are easily trapped in the local optimal areas . but the shortcoming of ga is its poor search efficiency . the thesis suggests integrating the gradient - based algorithm sqp into ga , and thus combining the global search ability of ga and fast convergence ability of sqp , which results in two hybrid algorithms ga - sqp - i and ga - sqp - ii
  该方法克服了传统优化方法需要计算目标函数导数且容易陷入局部最优的缺陷，但遗传算法本身具有搜索效率低的缺陷，本文提出将其与传统梯度算法sqp相结合，发挥遗传算法的全局搜索能力和梯度算法收敛快的优势，开发了两种混合算法ga - sqp - i和ga - sqp - 。
• This thesis suggests a process considered minimizes the population size as similar individuals occur in the fitter members of the population , which helps reduce the execution times for ga by removing the redundancy associated with the saturation effect found in the later generation . this thesis uses a method that adds dynamic penalty terms to the fitness function according to the optimal degree of solutions , so as to create a gradient toward a feasible suboptimal or even optimal solutions . on the basis of the difference of the biggest and the smallest of fitness of individual , modifying the fitness function in order to convergence is a satisfaction
  动态调节种群大小，去掉遗传算法在迭代后期搜索产生的过多相似个体，达到减少计算时间的目的；按照解的优劣程度给适应度函数增加一个在ga搜索过程中动态改变的可变罚函数，给搜索最优解创造一个梯度，使遗传算法收敛到可行的较优解或最优解；根据适应度值最大和最小个体的差修正适应度函数，使适应度函数值适中不容易造成收敛太快、局部收敛或根本不收敛而变成随机搜索；为了避免“近亲繁殖”采用竞争择优的交叉操作；利用并行遗传算法的思想，提出一种自适应多子种群进化策略；提出人口汰新政策来解决类似甚至相同的个体的情况发生。