Robust adaptive backstepping neural networks control for spacecraft rendezvous and docking with input saturation

This paper presents a robust adaptive neural networks control strategy for spacecraft rendezvous and docking with the coupled position and attitude dynamics under input saturation. Backstepping technique is applied to design a relative attitude controller and a relative position controller, respectively. The dynamics uncertainties are approximated by radial basis function neural networks (RBFNNs). A novel switching controller consists of an adaptive neural networks controller dominating in its active region combined with an extra robust controller to avoid invalidation of the RBFNNs destroying stability of the system outside the neural active region. An auxiliary signal is introduced to compensate the input saturation with anti-windup technique, and a command filter is employed to approximate derivative of the virtual control in the backstepping procedure. Globally uniformly ultimately bounded of the relative states is proved via Lyapunov theory. Simulation example demonstrates effectiveness of the proposed control scheme.     

Starting characteristics of the hypersonic wind tunnel with the mach number variation

The scramjet engine test facility (SETF) of the Korea Aerospace Research Institute is a blow-down, high enthalpy wind tunnel with a free-jet test cell. The free-jet test sections can reduce choking and test larger models compared with solid wall test sections, but the facility needs an excessive starting pressure ratio and is highly unpredictable because of the free-jet space. An air ejector system was designed to simulate a Mach 3.5 test condition. Quasi one-dimensional and computational analyses of the ejector were performed. The test results of the ejector well agree with the analysis results. The SETF showed a normal-shock efficiency of 58% at the Mach 3.5 condition. The air ejector system was modified to conduct a scramjet intake test with a Mach 6.7 condition. The normal-shock efficiency of the SETF was 40% with the Mach 6.7 condition. There was almost no change in the starting pressure ratio due to blockage.     

Novel prescribed performance neural control of a flexible air-breathing hypersonic vehicle with unknown initial errors

A novel prescribed performance neural controller with unknown initial errors is addressed for the longitudinal dynamic model of a flexible air-breathing hypersonic vehicle (FAHV) subject to parametric uncertainties. Different from traditional prescribed performance control (PPC) requiring that the initial errors have to be known accurately, this paper investigates the tracking control without accurate initial errors via exploiting a new performance function. A combined neural back-stepping and minimal learning parameter (MLP) technology is employed for exploring a prescribed performance controller that provides robust tracking of velocity and altitude reference trajectories. The highlight is that the transient performance of velocity and altitude tracking errors is satisfactory and the computational load of neural approximation is low. Finally, numerical simulation results from a nonlinear FAHV model demonstrate the efficacy of the proposed strategy.     

基于神经网络的电梯导轨校直计算

以电梯导轨校直过程中支点跨距、初始最大挠度为输入参数,校直行程和校直后测点的极差为输出,采用有限元优化设计方法得到的数据作为样本,建立了神经网络模型并使用样本对其进行了训练,利用随机数据对训练后的神经网络进行了测试.结果表明,将经过充分训练的神经网络应用于电梯导轨校直行程计算可以得到精确的计算结果,同时神经网络也可以反映校直后的导轨弯曲形式,为电梯导轨校直专家系统提供了一定参考.     

Data-driven model reference control of MIMO vertical tank systems with model-free VRFT and Q-Learning

This paper proposes a combined Virtual Reference Feedback Tuning–Q-learning model-free control approach, which tunes nonlinear static state feedback controllers to achieve output model reference tracking in an optimal control framework. The novel iterative Batch Fitted Q-learning strategy uses two neural networks to represent the value function (critic) and the controller (actor), and it is referred to as a mixed Virtual Reference Feedback Tuning–Batch Fitted Q-learning approach. Learning convergence of the Q-learning schemes generally depends, among other settings, on the efficient exploration of the state-action space. Handcrafting test signals for efficient exploration is difficult even for input-output stable unknown processes. Virtual Reference Feedback Tuning can ensure an initial stabilizing controller to be learned from few input-output data and it can be next used to collect substantially more input-state data in a controlled mode, in a constrained environment, by compensating the process dynamics. This data is used to learn significantly superior nonlinear state feedback neural networks controllers for model reference tracking, using the proposed Batch Fitted Q-learning iterative tuning strategy, motivating the original combination of the two techniques. The mixed Virtual Reference Feedback Tuning–Batch Fitted Q-learning approach is experimentally validated for water level control of a multi input-multi output nonlinear constrained coupled two-tank system. Discussions on the observed control behavior are offered.     

基于多目标模糊控制的电梯群控策略

目前绝大多数电梯群控系统只根据一个控制目标（最小候梯时间）来调度电梯。该文提出了基于多目标模糊控制的电梯群控策略,不仅考虑了候梯时间,还考虑了乘梯时间和电梯的能量损耗,在此基础上,对建立的电梯群控模糊控制数学模型的相关参数进行处理,取得了良好的效果,为电梯群控策略的研究提供了一种新思路。     

High-order tracking differentiator based adaptive neural control of a flexible air-breathing hypersonic vehicle subject to actuators constraints

In this paper, an adaptive neural controller is exploited for a constrained flexible air-breathing hypersonic vehicle (FAHV) based on high-order tracking differentiator (HTD). By utilizing functional decomposition methodology, the dynamic model is reasonably decomposed into the respective velocity subsystem and altitude subsystem. For the velocity subsystem, a dynamic inversion based neural controller is constructed. By introducing the HTD to adaptively estimate the newly defined states generated in the process of model transformation, a novel neural based altitude controller that is quite simpler than the ones derived from back-stepping is addressed based on the normal output-feedback form instead of the strict-feedback formulation. Based on minimal-learning parameter scheme, only two neural networks with two adaptive parameters are needed for neural approximation. Especially, a novel auxiliary system is explored to deal with the problem of control inputs constraints. Finally, simulation results are presented to test the effectiveness of the proposed control strategy in the presence of system uncertainties and actuators constraints.     

基于单片机的电梯优化控制单元的研究

电梯已成为人们生活中不可或缺的工具,但传统的电梯控制策略对人性化和节能方面的考虑不够充分,造成了电梯使用中的时间和电能的浪费。文中提出了一种省时、节电的电梯优化控制策略,通过增加电梯内和电梯外的指令取消功能、测试梯内人数并在梯外显示功能、满载测试与直驶功能等,减少不必要的电梯启停次数,实现电梯运行的优化控制,从而达到节电、省时的目的。根据这一策略,研制出了电梯用单片机控制单元。实验测试结果表明：该控制单元可减少不必要的启停次数,节省乘梯人的时间和电梯运行的能耗,实现人性化运行,具有良好的应用价值。     

功率分流式液压机械无级恒转速驱动器设计

在功率分流的基础上,采用机械传动和液压传动并联的形式设计一种应用于大功率风力发电的液压机械无级恒转速驱动器,建立其机械结构模型和液压调速系统模型,进行转速特性分析,并推导出了其动态传动比。采取非对称饱和增量式PID控制策略,使得传动比随着系统输入的变化而自动无级的进行调节,从而保证驱动系统恒转速输出,系统输出带动永磁同步发电机进行恒速恒频发电。利用仿真软件Adams和AMESim对其进行联合仿真,结果表明,液压机械无级恒转速驱动器可以实现很好的调节,同时也具有很高的效率,且始终保持在90%以上。同时为液压机械无级恒转速驱动器的工程应用提供了一定的理论参考。     

基于神经网络预测控制的节能电梯能量管理

随着电梯的日益增多,电梯节能问题引起越来越多的关注。针对节能型电梯中超级电容的储能管理问题,提出一种基于神经网络预测控制策略的新方法来动态预测电梯运行中所需的能量。首先根据电梯当前停层、目的停层以及载荷信息,建立反向传播神经网络(BPNN)模型并通过样本训练确定各神经元的权值和偏置值,然后利用该模型在线预测电梯每个行程吸收或回馈的能量,据此调节超级电容的平衡电压实现提前储能/泄能,补偿电梯运行过程中所需的尖峰功率。此外,根据电梯载荷和行程信息动态调整超级电容的平衡电压,可以充分利用超级电容的能量存储空间,在某些行程下除了补偿尖峰功率,还能够补偿一部分额定功率,优化网侧整流器的功率容量。最后通过MATLAB/Simulink搭建的仿真平台和实验样机验证了本方法的可行性。     

Tracking control of air-breathing hypersonic vehicles with non-affine dynamics via improved neural back-stepping design

This study considers the design of a new back-stepping control approach for air-breathing hypersonic vehicle (AHV) non-affine models via neural approximation. The AHV's non-affine dynamics is decomposed into velocity subsystem and altitude subsystem to be controlled separately, and robust adaptive tracking control laws are developed using improved back-stepping designs. Neural networks are applied to estimate the unknown non-affine dynamics, which guarantees the addressed controllers with satisfactory robustness against uncertainties. In comparison with the existing control methodologies, the special contributions are that the non-affine issue is handled by constructing two low-pass filters based on model transformations, and virtual controllers are treated as intermediate variables such that they aren't needed for back-stepping designs any more. Lyapunov techniques are employed to show the uniformly ultimately boundedness of all closed-loop signals. Finally, simulation results are presented to verify the tracking performance and superiorities of the investigated control strategy.     

Optimal design of motion control for scan tracking measurement: A CMAC approach

This paper is concerned with the optimal design of motion control for scan tracking measurement using Cerebellar Model Articulation Controller (CMAC) neural networks in order to improve the measuring efficiency while maintaining the measurement accuracy, and to achieve friction compensation. Aiming at the effect of geometric shape and material friction of the model surface on the precision and efficiency of scan tracking measurement, technologies of model surface geometric and friction feature identification and quantification are researched. A novel optimal motion controller for scan tracking measurement is designed and realized, which automatically predicts the surface features (including geometric feature and friction feature) and adjusts the scan tracking velocity in advance. The approach to friction quantification and compensation in measuring process is given specifically. Finally, through Matlab simulation experiments, the realizability and application effect of the studied optimal motion control method for scan tracking measurement are verified, and the measuring efficiency is increased by 123.33%. Simulation results show that the proposed motion controller is an effective way to enhance the measurement efficiency remarkably compared with the traditional control strategy.     

挖掘机器人伺服系统神经网络滑模控制

挖掘机器人伺服系统存在高度非线性、参数不确定和未建模动态等诸多不利因素,提出了一种结合径向基函数(RBF)神经网络的非线性滑模控制器,以提高控制精度和鲁棒性.首先,建立了单联伺服系统的数学模型;其次,采用RBF神经网络对系统的不利因素进行逼近,提出积分滑模面进一步减小稳态误差,同时减少对伺服系统参数的依赖,在此基础上,...     

Optimization study on the isolator length of dual-mode scramjet

The dual-mode scramjet is a prime candidate for air-breathing propulsion engines given its broad speed range and high efficiency. The detailed structure of the pseudo-shock wave in the isolator is worth investigating. Therefore, the present study investigates the effect of isolator length on the pseudo-shock wave. We replicate the experimental work of the reference paper with a two-dimensional numerical model. Pressure distribution on the body surface is applied in the numerical validation. The numerical result is in good agreement with that of the existing experimental work. We simulate the thermal choking condition with a throttling device. As the throat area of throttling device decreases, the back pressure continues increasing and the pseudo-shock wave moves upstream. Avoiding unstart phenomenon is a key objective in the operation of dual-mode scramjet. Therefore, we investigate the critical condition shortly before unstart phenomenon occurs. The pseudo-shock wave structure is described by pressure distribution. Our main objective is to explore the effect of isolator length on the dual-mode scramjet. The optimal isolator length depends on the maximum back pressure and total pressure loss in the critical condition shortly before the unstart phenomenon occurs. The maximum back pressure in the critical condition is higher when the isolator length ranges from 8.7 to 20.7. The total pressure loss in the critical condition decreases at the beginning and then increases when the isolator length ranges from 8.7 to 20.7.

     

重载大惯性液压驱动系统的神经网络近似逆控制

针对重载大惯性液压驱动系统,考虑系统的强非线性、模型不确定性和工作点的变化,设计了系统的神经网络近似逆控制器.该系统逆控制器可以直接从辨识所得的神经网络模型中得到,因而只需要训练一个神经网络.对大惯性重载非线性液压驱动系统的控制仿真研究表明,与传统PID控制器相比,神经网络近似逆控制器具有更好的动态控制性能,对模型不确...     

基于DSP的无刷电机调速系统智能算法设计

基于模糊神经网络控制技术,本文设计了一种无刷直流电机的数字调速控制策略,该控制策略以二维模糊神经网络控制器为核心,结合了模糊控制和神经网络的特点,具有很强的模糊推理和自学习能力.根据该控制策略,设计了基于DSP的硬件电路,并进行了MATLAB的仿真,结果表明,采用模糊神经网络算法能取得令人满意的动静态性能,具有很强的鲁棒性和自适应性,为无刷直流电机的智能控制提供了新的理论依据.     

增益调度自动驾驶仪结构特点与变轨迹飞行控制

构建了增益调度三回路自动驾驶系统来实现导弹实时变轨迹飞行,并对其结构原理、反馈量获取、控制性能及变轨飞行能力进行了研究.应用状态反馈原理和极点配置法建立了导弹纵向通道飞行控制数学模型;分析了弹体速度、动压和舵效率等状态参量对经典三回路自动驾驶仪控制指令的影响;根据增益调度理论,在自动驾驶仪中引入速度比、动压比及舵效率比等参数修正系统非线性,提出实际飞行过程中反馈量的获取方法.最后,分析了速率、复合稳定回路及过载回路的鲁棒稳定性.飞行试验表明:对大气密度、温度等数据的获取精度分别优于1.7％与3.0％;半物理实验表明:变轨弹道速度变化达到20.6％时,截止频率、加速度稳定增益和相角裕度分别由41.78 rad/s、0.87、78.19°变为41.71 rad/s、0.88和72.5°,相角裕度仍高于基准弹道最小值71.46°;动压变化77.6％对,自动驾驶仪仍保持了较好控制特性.实验结果表明,构建的增益调度三回路自动驾驶仪系统基本满足导弹实时变轨飞行的要求.     

电梯运行振动原因及减振方法探讨

电梯的振动一直是电梯安装、大修改造及运行过程中难以解决和控制的问题.文中主要从机械和电气部分两个方面分析了电梯振动的原因及相应减振对策,探讨了抑制电梯振动的方法,对解决类似以振动问题有较为实际的参考价值.     

基于神经网络逆系统的发酵过程多变量解耦控制

发酵过程是时变、非线性、不确定的多变量耦合系统，高性能的解耦控制一直是追求的目标。将逆系统方法与神经网络相结合，提出了一种基于神经网络逆系统的发酵过程解耦控制方法。根据发酵过程的特点，给出了相应的数学模型，并证明了系统的可逆性，进一步构造神经网络逆系统并与发酵系统串联复合成伪线性系统，再设计线性闭环调节器实现高性能解耦控制。仿真结果表明，提出的解耦控制方法能够适应过程模型的不确定性和参数的时变性，具有较强的鲁棒性，克服了解析逆系统解耦控制方案依赖于过程模型和对模型参数的变化很敏感的缺点。     

汽车防抱死制动系统分级智能控制

在分析车辆制动时轮胎与地面接触力学特性的基础上,提出一种用轮速峰值连线来求解参考车速和参考滑移率的方法。为了解决汽车防抱死制动系统(ABS)在各种条件下复杂的控制问题,设计出一种由运行控制、参数校正和组织协调构成的分级智能控制系统。在运行控制级,给出参考滑移率误差的目标轨迹,建立特征模型、控制模态集和推理规则集,以此设计出基于参考滑移率的仿人智能控制器。在参数校正级,为了弥补只针对参考滑移率控制的不足,用车轮角减速度对仿人智能控制量进行校正。在组织协调级,设计出基于轮减速度和参考滑移率的模糊智能控制器来自动辨别制动时的路面信息,给出四轮制动的协调控制规则。运用Matlab进行汽车ABS的仿人智能控制系统研究,搭建出汽车ABS全车测控系统,参照国际标准,在不同条件下进行道路试验。试验结果表明,相对于逻辑门限控制,ABS分级智能控制具有良好的制动平稳性和自适应性,可提高控制精度,是一种有效的新的ABS控制方法。