HSMC-SVM的二次逼近快速训练算法

HSMC-SVM是一种直接型高速多类支持向量机,适合用于类别较多的分类场合,但由于SMO算法采用经验方法选择工作集,使得在用SMO算法训练HSMC-SVM时,收敛速度较慢。为提高HSMC-SVM的收敛速度,该文提出用基于二次逼近的可行方向法来训练HSMC-SVM,并使用了样本缩减策略。实验表明,这种方法可以有效提高HSMC-SVM的收敛速度,其收敛速度已经超过了基于libsvm的组合多类支持向量机,完全可以用于分类类别多、样本数量大的分类场合。     

回归支持向量机SMO算法的改进

在Smola 和Sch?觟lkopf的SMO算法中,由于使用了单一的极限值而使得算法的效果没有完全表现出来。使用KKT条件来检验二次规划问题,使用两个极限参量来对回归SMO算法进行改进。通过对比实验,这一改进算法在执行速度上表现出了非常好的性能。     

Nonlinear observer based control for travelling wave nuclear reactors based on the Lyapunov approach during load following operation

Power control of the nuclear reactor is one of the most important subjects in each nuclear power plant. In this paper, a nonlinear controller using sliding mode method which is a robust nonlinear controller is designed to control a Traveling Wave Nuclear Reactor (TWR) power. The reactor core is simulated based on the point kinetics equations and six delayed neutron groups. Considering the limitations of the delayed neutron precursors densities measurement, a sliding mode observer is designed to estimate their values and finally a sliding mode control based on the sliding mode observer is presented to control the reactor core power. The stability analysis is given by means Lyapunov approach, thus the control system is guaranteed to be stable within a large range. Sliding Mode Control (SMC) is one of the robust and nonlinear methods which have several advantages such as robustness against matched external disturbances and parameter uncertainties. Since it has systematic design procedure, it is one of the most powerful solutions to design many practical control systems. The designed control system is evaluated in the presence of disturbances and uncertainties. The results show the robustness and performance of the used control system.     

无刷直流电机转矩观测与电感自适应辨识

为提高无刷直流电机（brushless DC motor,BLDCM）转矩观测结果的准确性,定量分析了传统的反电势滑模观测器（sliding．mode observer,SMO）及相应的转矩观测结果受定子电感参数偏差的影响程度。分析表明,电感参数偏差会在定子电流变化时引起反电势和转矩观测误差。为消除这一影响,利用李雅普诺夫（Lyapunov）稳定性理论,设计了定子电感自适应辨识率,在线辨识得到电感参数,实时调整SMO的系数矩阵和输入矩阵,构成了新型的自适应滑模观测器,提高了观测精度。最后利用RT-LAB实时控制器进行实验,验证了理论分析的正确性。实验结果表明,所提方法能有效辨识出BLDCM的定子电感值,并正确观测出电机转矩。     

基于滑模控制的TPESFTI驱动DFIG系统FCS-MPC

针对双馈异步风力发电机(DFIG)系统中机侧PNC型三电平逆变器的某桥臂发生故障时,提出了一种三相八开关容错逆变器(TPESFTI)控制策略,并给出了其拓扑结构和输出电压模型;考虑TPESFTI中母线电压分压不均衡等问题,研究分析直流侧电容电压不平衡的原因,给出了其相应的不平衡电压动态模型,并利用有限控制集模型预测控制(FCS-MPC)中目标函数的非线性约束项来抑制母线电容分压不均衡对系统的不利影响;同时为了提高FCS-MPC策略的性能,提出了基于滑模控制的FCS-MPC策略。仿真结果表明,该方法不仅能够确保DFIG系统稳定运行,并且能对母线电容电压不平衡进行有效抑制,同时具有良好的动态性能和较强的抗负载能力,进一步验证该控制策略的可靠性和有效性。     

单相脉宽调制整流器传感器故障诊断与容错控制

单相脉宽调制(PWM)整流器通常采用网侧电流与直流侧电压双闭环控制系统结构,一旦其检测系统中的传感器发生故障,将导致反馈值出现偏差,影响系统的控制效果与运行安全。设计一种准确可靠的单相PWM整流器网侧电流与直流侧电压传感器故障诊断与容错控制方案。采用解析冗余的方法实现故障诊断,基于单、双极性两种调制方法的开关函数取值设计了系统滑模观测器与状态估计器;将传感器采集值与观测器估计值作差并归一化建立系统残差,故障诊断单元通过比较残差与阈值定位故障传感器并迅速采取容错控制策略,以观测器的估计值代替故障传感器的测量值实现控制系统重构。通过仿真与实验模拟了三种传感器故障,验证了所提出故障诊断与容错控制方案的有效性与可靠性。     

NaCl溶液的浓度和温度对254SMO和2205不锈钢抗点蚀性的影响

用恒电位法和动电位法研究了254SMO不锈钢和2205双相不锈钢在50 g/L,100 g/L和200g/LNaCl溶液中的临界点蚀温度(CPT)和再钝化温度(CPrT),并绘制了两种材料在不同浓度溶液中的点蚀电位Epit和再钝化电位Eprot随温度的变化图,发现浓度越大,突变区域变缓变宽,材料CPT下降,耐蚀性下降....     

基于SMO的PMSM无位置传感器算法研究

基于滑模变结构观测器(SMO)的基本理论,构造了永磁同步电机(PMSM)矢量控制系统的仿真模型,并对永磁同步电机无位置传感器的转子位置和速度估计算法进行改进,用一个饱和函数代替sign函数,用变截止频率低通滤波器代替固定截止频率低通滤波器,用自适应算法速度观测器代替传统基于电机模型的无位置传感器,实现了无位置传感器对电机转速的准确估算.仿真和实验结果验证了该技术的有效性和合理性.     

基于SMO-SVM的单点金刚笔钝化监测

针对单点金刚笔在砂轮修整过程中易于钝化且难以检测的问题,使用支持向量机建立智能模型。为了得到建立模型所需的样本库,使用小波包分析等方法在线提取修整时声发射信号中的特征信息,并引入钝化平台直径定义钝化临界值。模型本身选用基于串行优化算法的支持向量分类机,使用交叉验证法搭配遗传算法以达到优化模型参数的目的。实验结果表明,该模型在分类精度和计算时间上均优于一般的智能模型,可以有效地监测金刚笔的钝化。     

A New Weighted Least Squares Support Vector Machines and Its Sequential Minimal Optimization Algorithm

Least squares support vector machines （LSSVM） is widely used in pattern recognition and artificial intelligence domain in recent years for its efficiency in classification and regression. The solution of LSSVM is an optimization problem of a Sum squared error （SSE） cost function with only equality constraints and can be solved in a simple linear system. However, its generalization performance is sensitive to noise points and outliers that are often existent in training dataset. In order to endow robustness to LSSVM, a new method for computing weight vector of error is proposed and the substituting of weighted error vector for original error vector in LSSVM gives birth to a new weighted LSSVM. The method gets weight factor by computing distance between sample and its corresponding class center. Sequential minimal optimization （SMO） algorithm is also extended to the new method for its efficient application. Comparison experiments show superiority of the new method in terms of generalization performance, robust property and sparse approximation. Especially, the new method is much faster than the other method for large number of samples.