Robust exponential stability analysis of a larger class of discrete-time recurrent neural networks

The robust exponential stability of a larger class of discrete-time recurrent neural networks (RNNs) is explored in this paper. A novel neural network model, named standard neural network model (SNNM), is introduced to provide a general framework for stability analysis of RNNs. Most of the existing RNNs can be transformed into SNNMs to be analyzed in a unified way. Applying Lyapunov stability theory method and S-Procedure technique, two useful criteria of robust exponential stability for the discrete-time SNNMs are derived. The conditions presented are formulated as linear matrix inequalities (LMIs) to be easily solved using existing efficient convex optimization techniques. An example is presented to demonstrate the transformation procedure and the effectiveness of the results.

Adaptive learning with guaranteed stability for discrete-time recurrent neural networks

To avoid unstable learning, a stable adaptive learning algorithm was proposed for discrete-time recurrent neural networks. Unlike the dynamic gradient methods, such as the backpropagation through time and the real time recurrent learning, the weights of the recurrent neural networks were updated online in terms of Lyapunov stability theory in the proposed learning algorithm, so the learning stability was guaranteed. With the inversion of the activation function of the recurrent neural networks, the proposed learning algorithm can be easily implemented for solving varying nonlinear adaptive learning problems and fast convergence of the adaptive learning process can be achieved. Simulation experiments in pattern recognition show that only 5 iterations are needed for the storage of a 15×15 binary image pattern and only 9 iterations are needed for the perfect realization of an analog vector by an equilibrium state with the proposed learning algorithm.

Robust Sliding Mode Control for Nonlinear Discrete-Time Delayed Systems Based on Neural Network

This paper presents a robust sliding mode controller for a class of unknown nonlinear discrete-time systems in the presence of fixed time delay. A neural-network approximation and the Lyapunov-Krasovskii functional theory into the sliding-mode technique is used and a neural-network based sliding mode control scheme is proposed. Because of the novality of Chebyshev Neural Networks (CNNs), that it requires much less computation time as compare to multi layer neural network (MLNN), is preferred to approximate the unknown system functions. By means of linear matrix inequalities, a sufficient condition is derived to ensure the asymptotic stability such that the sliding mode dynamics is restricted to the defined sliding surface. The proposed sliding mode control technique guarantees the system state trajectory to the designed sliding surface. Finally, simulation results illustrate the main characteristics and performance of the proposed approach.

Improved results on passivity analysis of discrete-time stochastic neural networks with time-varying delay

The problem of passivity analysis for a class of discrete-time stochastic neural networks (DSNNs) with time-varying interval delay was investigated. The delay-dependent sufficient criteria were derived in terms of linear matrix inequalities (LMIs). The results are shown to be generalization of some previous results and are less conservative than the existing works. Meanwhile, the computational complexity of the obtained stability conditions is reduced because less variables are involved. A numerical example is given to show the effectiveness and the benefits of the proposed method.

Attractors and the attraction basins of discrete-time cellular neural networks

The dynamic behavior of discrete-time cellular neural networks(DTCNN), which is strict with zero threshold value, is mainly studied in asynchronous mode and in synchronous mode. In general, a k-attractor of DTCNN is not a convergent point. But in this paper, it is proved that a k-attractor is a convergent point if the strict DTCNN satisfies some conditions. The attraction basin of the strict DTCNN is studied, one example is given to illustrate the previous conclusions to be wrong, and several results are presented. The obtained results on k-attractor and attraction basin not only correct the previous results, but also provide a theoretical foundation of performance analysis and new applications of the DTCNN.

Stability analysis of extended discrete-time BAMneural networks based on LMI approach

We propose a new approach for analyzing the global asymptotic stability of the extended discrete-time bidirectional associative memory （BAM） neural networks. By using the Euler rule, we discretize the continuous-time BAM neural networks as the extended discrete-time BAM neural networks with non-threshold activation functions. Here we present some conditions under which the neural networks have unique equilibrium points. To judge the global asymptotic stability of the equilibrium points, we introduce a new neural network model - standard neural network model （SNNM）. For the SNNMs, we derive the sufficient conditions for the global asymptotic stability of the equilibrium points, which are formulated as some linear matrix inequalities （LMIs）. We transform the discrete-time BAM into the SNNM and apply the general result about the SNNM to the determination of global asymptotic stability of the discrete-time BAM. The approach proposed extends the known stability results, has lower conservativeness, can be verified easily, and can also be applied to other forms of recurrent neural networks.

The dynamic relaxation form finding method aided with advanced recurrent neural network

How to establish a self‐equilibrium configuration is vital for further kinematics and dynamics analyses of tensegrity mechanism.In this study,for investigating tensegrity form‐finding problems,a concise and efficient dynamic relaxation‐noise tolerant zeroing neural network(DR‐NTZNN)form‐finding algorithm is established through analysing the physical properties of tensegrity structures.In addition,the non‐linear constrained opti-misation problem which transformed from the form‐finding problem is solved by a sequential quadratic programming algorithm.Moreover,the noise may produce in the form‐finding process that includes the round‐off errors which are brought by the approximate matrix and restart point calculating course,disturbance caused by external force and manufacturing error when constructing a tensegrity structure.Hence,for the purpose of suppressing the noise,a noise tolerant zeroing neural network is presented to solve the search direction,which can endow the anti‐noise capability to the form‐finding model and enhance the calculation capability.Besides,the dynamic relaxation method is contributed to seek the nodal coordinates rapidly when the search direction is acquired.The numerical results show the form‐finding model has a huge capability for high‐dimensional free form cable‐strut mechanisms with complicated topology.Eventually,comparing with other existing form‐finding methods,the contrast simulations reveal the excellent anti‐noise performance and calculation capacity of DR‐NTZNN form‐finding algorithm.

Neural Dynamics for Cooperative Motion Control of Omnidirectional Mobile Manipulators in the Presence of Noises: A Distributed Approach

This paper presents a distributed scheme with limited communications, aiming to achieve cooperative motion control for multiple omnidirectional mobile manipulators(MOMMs).The proposed scheme extends the existing single-agent motion control to cater to scenarios involving the cooperative operation of MOMMs. Specifically, squeeze-free cooperative load transportation is achieved for the end-effectors of MOMMs by incorporating cooperative repetitive motion planning(CRMP), while guiding each individual to desired poses. Then, the distributed scheme is formulated as a time-varying quadratic programming(QP) and solved online utilizing a noise-tolerant zeroing neural network(NTZNN). Theoretical analysis shows that the NTZNN model converges globally to the optimal solution of QP in the presence of noise. Finally, the effectiveness of the control design is demonstrated by numerical simulations and physical platform experiments.

Anti-noise performance of the pulse coupled neural network applied in discrimination of neutron and gamma-ray

In this study,the anti-noise performance of a pulse-coupled neural network(PCNN)was investigated in the neutron and gamma-ray(n-γ)discrimination field.The experiments were conducted in two groups.In the first group,radiation pulse signals were pre-processed using a Fourier filter to reduce the original noise in the signals,whereas in the second group,the original noise was left untouched to simulate an extremely high-noise scenario.For each part,artificial Gaussian noise with different intensity levels was added to the signals prior to the discrimination process.In the aforementioned conditions,the performance of the PCNN was evaluated and compared with five other commonly used methods of n-γdiscrimination:(1)zero crossing,(2)charge comparison,(3)vector projection,(4)falling edge percentage slope,and(5)frequency gradient analysis.The experimental results showed that the PCNN method significantly outperforms other methods with outstanding FoM-value at all noise levels.Furthermore,the fluctuations in FoM-value of PCNN were significantly better than those obtained via other methods at most noise levels and only slightly worse than those obtained via the charge comparison and zerocrossing methods under extreme noise conditions.Additionally,the changing patterns and fluctuations of the FoMvalue were evaluated under different noise conditions.Hence,based on the results,the parameter selection strategy of the PCNN was presented.In conclusion,the PCNN method is suitable for use in high-noise application scenarios for n-γdiscrimination because of its stability and remarkable discrimination performance.It does not rely on strict parameter settings and can realize satisfactory performance over a wide parameter range.

A New Noise-Tolerant Dual-Neural-Network Scheme for Robust Kinematic Control of Robotic Arms With Unknown Models

Taking advantage of their inherent dexterity,robotic arms are competent in completing many tasks efficiently.As a result of the modeling complexity and kinematic uncertainty of robotic arms,model-free control paradigm has been proposed and investigated extensively.However,robust model-free control of robotic arms in the presence of noise interference remains a problem worth studying.In this paper,we first propose a new kind of zeroing neural network(ZNN),i.e.,integration-enhanced noise-tolerant ZNN(IENT-ZNN)with integration-enhanced noisetolerant capability.Then,a unified dual IENT-ZNN scheme based on the proposed IENT-ZNN is presented for the kinematic control problem of both rigid-link and continuum robotic arms,which improves the performance of robotic arms with the disturbance of noise,without knowing the structural parameters of the robotic arms.The finite-time convergence and robustness of the proposed control scheme are proven by theoretical analysis.Finally,simulation studies and experimental demonstrations verify that the proposed control scheme is feasible in the kinematic control of different robotic arms and can achieve better results in terms of accuracy and robustness.

Zero phase error control based on neural compensation for flight simulator servo system

Using the future desired input value, zero phase error controller enables the overall system＇s frequency response exhibit zero phase shift for all frequencies and a small gain error at low frequency range, and based on this, a new algorithm is presented to design the feedforward controller. However, zero phase error controller is only suitable for certain linear system. To reduce the tracking error and improve robustness, the design of the proposed feedforward controller uses a neural compensation based on diagonal recurrent neural network. Simulation and real-time control results for flight simulator servo system show the effectiveness of the proposed approach.

Distributed Cooperative Learning for Discrete-Time Strict-Feedback Multi Agent Systems Over Directed Graphs

This paper focuses on the distributed cooperative learning(DCL)problem for a class of discrete-time strict-feedback multi-agent systems under directed graphs.Compared with the previous DCL works based on undirected graphs,two main challenges lie in that the Laplacian matrix of directed graphs is nonsymmetric,and the derived weight error systems exist n-step delays.Two novel lemmas are developed in this paper to show the exponential convergence for two kinds of linear time-varying(LTV)systems with different phenomena including the nonsymmetric Laplacian matrix and time delays.Subsequently,an adaptive neural network(NN)control scheme is proposed by establishing a directed communication graph along with n-step delays weight updating law.Then,by using two novel lemmas on the extended exponential convergence of LTV systems,estimated NN weights of all agents are verified to exponentially converge to small neighbourhoods of their common optimal values if directed communication graphs are strongly connected and balanced.The stored NN weights are reused to structure learning controllers for the improved control performance of similar control tasks by the“mod”function and proper time series.A simulation comparison is shown to demonstrate the validity of the proposed DCL method.

Neural network solution for finite-horizon H-infinity constrained optimal control of nonlinear systems

In this paper, neural networks are used to approximately solve the finite-horizon constrained input H-infinity state feedback control problem. The method is based on solving a related Hamilton-Jacobi-Isaacs equation of the corresponding finite-horizon zero-sum game. The game value function is approximated by a neural network with time- varying weights. It is shown that the neural network approximation converges uniformly to the game-value function and the resulting almost optimal constrained feedback controller provides closed-loop stability and bounded L2 gain. The result is an almost optimal H-infinity feedback controller with time-varying coefficients that is solved a priori off-line. The effectiveness of the method is shown on the Rotational/Translational Actuator benchmark nonlinear control problem.

基于神经网络的零射程线闭路制导方法

对于固体发动机飞行器,在动力飞行段一般采用能量匹配闭路制导或者倾角约束的闭路制导方法,这种闭路制导方法的控制精度较高,但其精度受到发动机性能偏差的影响较大,为了减小发动机性能偏差对制导精度的影响,提出了基于神经网络的零射程线闭路制导方法。首先,建立了飞行器动力飞行段运动模型,对零射程线闭路制导进行了分析推导;其次,设计了多输入神经网络算法,确定了输入输出参数,对剩余能量、待增速度以及趋零射程线角度进行训练,将神经网络训练结果与零射程线闭路制导相结合,使得在不同的发动机偏差情况下,反馈不同的趋零射程线角度;最后,通过选择不同的偏差状态进行仿真验证,仿真结果表明该方法能有效地减小发动机偏差对制导精度的影响,具有较强的抗偏差能力,制导精度高。

零维预测燃烧模型建模方法

为了解决神经网络建立的柴油机零维燃烧模型对稳态和动态工况预测能力不稳定问题,本文采用遗传算法对神经网络的初始权值、阈值进行综合优化,提出了遗传算法-神经网络算法。基于TBD620型柴油机,通过稳态和瞬态试验获得运行参数和缸压数据,通过代数分析法结合遗传算法获得对应燃烧参数,最后分别利用遗传算法-神经网络算法和神经网络算法对燃烧模型进行构建并对比辨识结果。结果表明:与神经网络算法相比,遗传算法-神经网络算法构建的零维燃烧模型对应φ_(50)和IMEP预测值平均误差分别降低了43.84%和42.73%,遗传算法具有高效的权值、阈值寻优能力,模型具有更高的预测精度,泛化性更好,适用于柴油机零维燃烧模型研究。

两方零和马尔科夫博弈策略梯度算法及收敛性分析

为了解决基于策略的强化学习方法在两方零和马尔科夫博弈中学习效率低下的问题,提出同时更新双方玩家策略的近似纳什均衡策略优化算法.将两方零和马尔科夫博弈问题描述为最大最小优化问题,针对参数化策略,给出马尔科夫博弈的策略梯度定理,并通过近似随机策略梯度的推导,为算法实施提供可行性基础.通过比较分析不同的最大最小问题梯度更新方法,发现额外梯度相较于其他方法具有更好的收敛性能.基于这一发现,提出基于额外梯度的近似纳什均衡策略优化算法,并给出算法的收敛性证明.在Oshi-Zumo游戏上,使用表格式softmax参数化策略以及神经网络作为参数化策略,验证不同游戏规模场景下算法的有效性.通过对比实验,验证算法相对于其他方法的收敛性和优越性.

基于LSTM神经网络的机载光纤陀螺温度冲击误差补偿技术

环境温度冲击会降低机载光纤陀螺的性能,从而影响飞行器导航和姿态控制精度。在光纤陀螺误差机理研究基础上,本文提出一种基于长短期记忆(LSTM)神经网络的光纤陀螺温度误差补偿模型。该模型通过LSTM网络对光纤陀螺的零偏和标度因数进行实时预测和校正,提高光纤陀螺的测量精度。试验结果表明,在温度冲击下,LSTM预测模型补偿后的标度因数误差小于30ppm,零偏稳定性比常规的线性拟合补偿模型提高0.0034(°)/h。这意味着输出更准确地反映实际角速度值,陀螺仪的零偏漂移更小,输出更接近于零值。动态试验中转台输入为20(°)/s时,LSTM补偿后陀螺输出稳定在19.999~20.001(°)/s区间内,相较于陀螺原始输出误差降低0.008(°)/s。通过LSTM预测模型补偿,能够在环境变化、外部扰动或传感器故障时,通过陀螺仪提供更可靠的数据支持,维持飞行器的稳定性和安全性。

基于零样本学习的单张SAR图像相干斑滤波方法

相干斑滤波是合成孔径雷达(Synthetic aperture radar,SAR)图像解译重要的预处理步骤。近年来,基于卷积神经网络(Convolutional neural network,CNN)的相干斑滤波方法得到了快速的发展。然而,基于监督学习的滤波方法缺乏无相干斑参考SAR图像作为真值,基于自监督学习的滤波方法大多需要同一场景的多时相SAR图像训练网络,但是这些额外的数据集在实际场景中较难获取。此外,自监督学习方法通常需要较大的训练数据集和较深的网络进行相干斑滤波,导致其计算复杂度较高。因此,本文提出了一种基于零样本学习的单张SAR图像相干斑滤波方法。该方法的核心思想是对待测试的单张SAR图像进行子视分解,选取与待测试SAR图像欧式距离最近的子视图像进行配对,理论上证明了使用配对的子视图像自监督训练网络能达到使用无相干斑参考SAR图像监督训练网络的滤波效果。因此,通过设计自监督损失函数快速训练轻量化相干斑滤波网络,将训练好的网络对待测试SAR图像进行滤波。相较于基于监督学习和自监督学习的相干斑滤波方法,本文所提方法不需要无相干斑参考或多时相SAR图像用于模型训练,也不需要额外训练数据,只需使用任意一个轻量化的CNN即可实现相干斑滤波。在Radarsat-2和ALOS-2实测数据上的实验结果表明,本文所提方法的参数量比对比方法低22倍,能更好的实现对匀质区域相干斑的抑制和图像细节的保护。

基于EMD与机器学习算法的近零能耗建筑负荷预测方法

采用皮尔逊相关系数法分析了不同特征变量与冷热负荷的相关性,确定了预测模型的输入特征变量。采用经验模态分解(EMD)对逐日冷热负荷按频分解,然后采用机器学习算法,即反向传播神经网络(BPNN)、随机森林(RF)和支持向量机(SVM),分别对不同频率的负荷量进行了训练、验证,最后重构得到了近零能耗建筑预测负荷。基于上述方法,以北京市某近零能耗居住建筑为研究对象,比较了不同算法预测结果的精确度。结果表明:采用EMD与RF算法相结合对近零能耗建筑冷热负荷的预测精确度较高。进一步采用穷举搜索法对模型初设参数进行了优化,冷热负荷预测结果精确度提高,冷负荷预测结果的决定系数R2、平均绝对百分比误差MAPE分别为0.996、1.32%,热负荷预测结果的R2、MAPE分别为0.997、0.79%。