针对回声状态网络(Echo state network, ESN)的结构设计问题,提出基于灵敏度分析的模块化回声状态网络修剪算法(Pruning algorithm for modular echo state network, PMESN).该网络由相互独立的子储备池模块构成.首先利用矩阵的奇异值分...针对回声状态网络(Echo state network, ESN)的结构设计问题,提出基于灵敏度分析的模块化回声状态网络修剪算法(Pruning algorithm for modular echo state network, PMESN).该网络由相互独立的子储备池模块构成.首先利用矩阵的奇异值分解(Singular value decomposition, SVD)构造子储备池模块的权值矩阵,并利用分块对角阵原理生成储备池.然后利用子储备池模块输出和相应的输出层权值向量,定义学习残差对于子储备池模块的灵敏度以及网络规模适应度.利用灵敏度大小判断子储备池模块的贡献度,并根据网络规模适应度确定子储备池模块的个数,删除灵敏度低的子模块.在网络的修剪过程中,不需要缩放权值就可以保证网络的回声状态特性.实验结果说明,所提出的算法有效解决了ESN的网络结构设计问题,基本能够确定与样本数据相匹配的网络规模,具有较好的泛化能力和鲁棒性.展开更多
A novel wavelet network based adaptive equalizer (WNBAE) is presented and the structure and stochastic gradient learning algorithm is given. The proposed WNBAE has better performance than that of the conventional lin...A novel wavelet network based adaptive equalizer (WNBAE) is presented and the structure and stochastic gradient learning algorithm is given. The proposed WNBAE has better performance than that of the conventional linear transversal equalizer based on the LMS and the RLS algorithms, as well as that of the decision feedback equalizer based on the RLS algorithm, especially for MQAM digital communication reception systems over the nonlinear channels. In addition, it outperforms the BP neural network based adaptive equalizer slightly. However, it has a slow convergence rate and a high computational complexity. Several simulations are performed to evaluate the behavior of the WNBAE.展开更多
To solve the problem of altitude control of a tilt tri-rotor unmanned aerial vehicle(UAV)in the transition mode,this study presents a grey wolf optimization(GWO)based neural network adaptive control scheme for a tilt ...To solve the problem of altitude control of a tilt tri-rotor unmanned aerial vehicle(UAV)in the transition mode,this study presents a grey wolf optimization(GWO)based neural network adaptive control scheme for a tilt trirotor UAV in the transition mode.Firstly,the nonlinear model of the tilt tri-rotor UAV is established.Secondly,the tilt tri-rotor UAV altitude controller and attitude controller are designed by a neural network adaptive control method,and the GWO algorithm is adopted to optimize the parameters of the neural network and the controllers.Thirdly,two altitude control strategies are designed in the transition mode.Finally,comparative simulations are carried out to demonstrate the effectiveness and robustness of the proposed control scheme.展开更多
In this paper, we propose two novel efficient scheduling schemes with network coding in multi-relay wireless network to maximize the transmission efficiency. The first one uses adaptive forwarding with network coding(...In this paper, we propose two novel efficient scheduling schemes with network coding in multi-relay wireless network to maximize the transmission efficiency. The first one uses adaptive forwarding with network coding(AF-NC), in which each relay adaptively calculates the number of packets having innovative information according to the feedback from the sink. With AF-NC, duplicate packets are not sent, and the total number of time slots needed to complete transmission can be significantly reduced. The second scheme, named adaptive forwarding with network coding and retransmission(AFR-NC), combines AF-NC with automatic repeat request(ARQ) to guarantee reliable end-to-end communication with limited resource occupation. Numerical results show that compared with simple forwarding with network coding(F-NC), AF-NC has close successful delivery rate with dramatically less time slots, while AFR-NC achieves strict reliability with limited resource cost.展开更多
文摘针对回声状态网络(Echo state network, ESN)的结构设计问题,提出基于灵敏度分析的模块化回声状态网络修剪算法(Pruning algorithm for modular echo state network, PMESN).该网络由相互独立的子储备池模块构成.首先利用矩阵的奇异值分解(Singular value decomposition, SVD)构造子储备池模块的权值矩阵,并利用分块对角阵原理生成储备池.然后利用子储备池模块输出和相应的输出层权值向量,定义学习残差对于子储备池模块的灵敏度以及网络规模适应度.利用灵敏度大小判断子储备池模块的贡献度,并根据网络规模适应度确定子储备池模块的个数,删除灵敏度低的子模块.在网络的修剪过程中,不需要缩放权值就可以保证网络的回声状态特性.实验结果说明,所提出的算法有效解决了ESN的网络结构设计问题,基本能够确定与样本数据相匹配的网络规模,具有较好的泛化能力和鲁棒性.
文摘A novel wavelet network based adaptive equalizer (WNBAE) is presented and the structure and stochastic gradient learning algorithm is given. The proposed WNBAE has better performance than that of the conventional linear transversal equalizer based on the LMS and the RLS algorithms, as well as that of the decision feedback equalizer based on the RLS algorithm, especially for MQAM digital communication reception systems over the nonlinear channels. In addition, it outperforms the BP neural network based adaptive equalizer slightly. However, it has a slow convergence rate and a high computational complexity. Several simulations are performed to evaluate the behavior of the WNBAE.
文摘To solve the problem of altitude control of a tilt tri-rotor unmanned aerial vehicle(UAV)in the transition mode,this study presents a grey wolf optimization(GWO)based neural network adaptive control scheme for a tilt trirotor UAV in the transition mode.Firstly,the nonlinear model of the tilt tri-rotor UAV is established.Secondly,the tilt tri-rotor UAV altitude controller and attitude controller are designed by a neural network adaptive control method,and the GWO algorithm is adopted to optimize the parameters of the neural network and the controllers.Thirdly,two altitude control strategies are designed in the transition mode.Finally,comparative simulations are carried out to demonstrate the effectiveness and robustness of the proposed control scheme.
基金the National Natural Science Foundation of China(Nos.61102051,61221001 and 61301117)the National High Technology Research and Development Program(863)of China(Nos.2012AA011701 and 2012AA121601)+1 种基金the Shanghai Jiao Tong University Science and Technology Innovation Foundation(No.AF0300021)the Shanghai Key Laboratory Funding(No.12DZ2272600)
文摘In this paper, we propose two novel efficient scheduling schemes with network coding in multi-relay wireless network to maximize the transmission efficiency. The first one uses adaptive forwarding with network coding(AF-NC), in which each relay adaptively calculates the number of packets having innovative information according to the feedback from the sink. With AF-NC, duplicate packets are not sent, and the total number of time slots needed to complete transmission can be significantly reduced. The second scheme, named adaptive forwarding with network coding and retransmission(AFR-NC), combines AF-NC with automatic repeat request(ARQ) to guarantee reliable end-to-end communication with limited resource occupation. Numerical results show that compared with simple forwarding with network coding(F-NC), AF-NC has close successful delivery rate with dramatically less time slots, while AFR-NC achieves strict reliability with limited resource cost.
