一种基于最小二乘支持向量机的预测控制算法

针对工业过程中普遍存在的非线性被控对象，提出一种基于最小二乘支持向量机建模的预测控制算法．首先，用具有RBF核函数的LS-SVM离线建立被控对象的非线性模型；然后，在系统运行过程中，将离线模型在每一个采样周期关于当前采样点进行线性化，并用广义预测算法实现对被控系统的预测控制．仿真结果表明了该算法的有效性和优越性．     

Application of Feedback Linearization Method in Airplane Automatic Landing Control System

Summarizes the I/O feedback linearization about MIMO system, and applies it to nonlinear control equation of airplane. And also designs the tracing control laws for airplane longitudinal automatic landing control system.     

无刷直流电机状态观测器的设计

从介绍扩展卡尔曼滤波算法原理人手，建立了以不导通相反电势为基础的无位置传感器无刷直流电机的状态观测模型，估算出电机转子位置和速度，最后对模型进行了仿真并给出了仿真结果。     

达芬-谐波振子的改进解析逼近解

研究达芬-谐波振子的解析逼近解。所谓达芬-谐波振子是指当位移远小于1时，系统可化为三次非线性振子，而当位移远大于1时，该系统则化为线性谐波振子。通过将变形后的控制方程的线性化与谐波平衡法组合起来，我们建立了达芬-谐波振子频率及周期解的改进解板逼近。改进的解析逼近在振幅的全部取值范围内，包括振幅趋于零及无穷的情况，都有令人非常满意的精度。     

Curve interpolation with variable feedrate for surface requirement

Finish machining of a curved surface is often carried out by an NC system with curve interpolation in the field. This function, called a NURBS interpolation, adopts a feedrate optimizing strategy based on both the geometrical information of the curved path and dynamic properties such as the curvature of the curve, the allowable acceleration and the time constant. However, in the case of a finish cut using a ball-end mill, the curve interpolator needs to take the machining process into account for improved surface roughness, while reducing the polishing time. This surface roughness on high-speed machining is theoretically defined by the feed per tooth and the pickfeed at the given radius of the tool. In this study, the effect of low machinability at the bottom of a tool on surface roughness is also considered. A curve interpolation algorithm is proposed for generating particular feedrate commands that are able to control the roughness of a curved surface. The simulation of the machined surface by the proposed algorithm was carried out, and experimental results are presented. A feedrate scheme that depends on the inclination angle has important potential application in part finishes consistent with prescribed surface roughness. The results show that the proposed algorithm is potentially useful for roughness-controlled machining of curved surface products.     

五自由度无轴承异步电机α-阶逆系统解耦控制

A 5-degrees-of-freedom bearingless induction motor is a multi-variable, nonlinear and strong-coupled system. In order to achieve rotor suspension and operation steadily, it is necessary to realize dynamic decoupling control among torque and suspension forces. In the paper, a method based on α-th order inverse system theory is used to study dynamic decoupling control. Firstly, the working principles of a 3-degrses-of-frsedom magnetic bearing and a 2-degrees-of-freedom bearingiess induction motor are analyzed,the radial-axial force equations of 3-degrees-of-freedom magnetic bearing, the electromagnetic torque equation and radial force equations of the 2-degrees-of-freedom bearingless induction motor are given, and then the state equatious of the 5-degrees-of-freedom bearingless induction motor are set up. Secondly, the feasibility of decoupling control based on dynamic inverse theory is discussed in detail, and the state feedback linearization method is used to decouple and linearize the system. Finally, linear control system techniques are applied to these linearization subsystems to synthesize and simulate. The simulation results have shown that this kind of control strategy can realize dynamic decoupling control among torque and suspeusion forces of the 5-degrees-of-freedom bearingless induction motor, and that the control system has good dynamic and static performance.     

热轧带钢结疤缺陷产生原因及控制

针对某热轧带钢厂带钢表面结疤缺陷的外形特征及分布规律,分析了该缺陷产生的原因。通过对精轧侧导板开度余量及短行程数值设置固化、优化电气控制程序确保剪前导板线性化、精轧侧导板上安装压力传感器等措施,有效降低了结疤缺陷发生率。     

BUCK变换器带恒功率负载反馈精确线性化控制

对BUCK交换器带恒功率负载系统的非线性控制进行了研究.根据恒功率负载的非线性特性,推导出BUCK变换器带恒功率负载的精确线性化反馈控制规律,设计了系统非线性控制器.用SABER软件对控制律进行了仿真,结果表明反馈精确线性化技术可以实现对BUCK变换器带恒功率负载的控制,在电源及负载大范围变化时,保证系统的稳定运行,具有大信号稳定特性.     

球棒系统的近似线性化及Matlab实现

以典型多变量非线性系统一球棒系统为研究对象,用微分几何理论中的状态反馈的近似线性化方法设计该非线性系统的控制器.无论小球在棒的什么位置,棒的角度如何,该控制方法都能使小球稳定在棒的中心位置.利用数学工具Mat-lab仿真此过程,给出了Matlab具体实现模块,结果表明状态反馈方法的有效性.     

一种前馈神经网的快速算法

前馈神经网已经被大量用于非线性信号处理． 经典反向传播算法是一种标准的前馈网络学习算法,但是,对许多应用,反向传播算法的收 敛速度却很慢．本文根据对网络的非线性单元进行线性化而提出一种新的算法,该算法在非 线性信号处理中在精度和收敛速度方面都优于传统的反向传播算法．