This paper illustrates the benefits of a self-tuning PID strategy applied to a proton exchange membrane fuel cell system. Controller parameters are updated on-line, at each sampling time, based on an instantaneous lin...This paper illustrates the benefits of a self-tuning PID strategy applied to a proton exchange membrane fuel cell system. Controller parameters are updated on-line, at each sampling time, based on an instantaneous linearization of an artificial neural network model of the process and a General Minimum Variance control law. The self-tuning PID scheme allows managing nonlinear behaviors of the system while avoiding heavy computations. The applicability, efficiency and robustness of the proposed control strategy are experimentally confirmed using varying control scenarios. In this aim, the original built-in controller is overridden and the self-tuning PID controller is implemented externally and executed on-line. Experimental results show good performance in setpoint tracking accuracy and robustness against plant/model mismatch. The proposed strategy appears to be a promising alternative to heavy computation nonlinear control strategies and not optimal linear control strategies.展开更多
建立了交流稳压电源主电路数学模型并分析其闭环稳压控制原理。由于装置具有较强的非线性和变结构、变参数特性,采用经典PID控制器很难获得理想的控制效果。将人工神经网络与传统PID控制器相结合,构成一种不依赖于被控对象精确数学模型...建立了交流稳压电源主电路数学模型并分析其闭环稳压控制原理。由于装置具有较强的非线性和变结构、变参数特性,采用经典PID控制器很难获得理想的控制效果。将人工神经网络与传统PID控制器相结合,构成一种不依赖于被控对象精确数学模型的神经网络PID控制器。为了提高神经网络的收敛速度,采用Levenberg-Marquardt算法计算连接权值更新量,并对当前解施加一个以一定概率保留的随机扰动,加快迭代过程跳出局部极小点。对装置主电路和改进神经网络PID控制器进行仿真,结果表明:系统动态响应快,鲁棒性强,调节平滑,具有较好的控制效果。最后,制造并测试了额定电压660 V、容量400 k VA的实验样机,对理论研究进行了实验验证。展开更多
A multilayered perceptrons＇ neural network technique has been applied in the particle identification at BESIII. The networks are trained in each sub-detector level. The NN output of sub-detectors can be sent to a seq...A multilayered perceptrons＇ neural network technique has been applied in the particle identification at BESIII. The networks are trained in each sub-detector level. The NN output of sub-detectors can be sent to a sequential network or be constructed as PDFs for a likelihood. Good muon-ID, electron-ID and hadron-ID are obtained from the networks by using the simulated Monte Carlo samples.展开更多
文摘This paper illustrates the benefits of a self-tuning PID strategy applied to a proton exchange membrane fuel cell system. Controller parameters are updated on-line, at each sampling time, based on an instantaneous linearization of an artificial neural network model of the process and a General Minimum Variance control law. The self-tuning PID scheme allows managing nonlinear behaviors of the system while avoiding heavy computations. The applicability, efficiency and robustness of the proposed control strategy are experimentally confirmed using varying control scenarios. In this aim, the original built-in controller is overridden and the self-tuning PID controller is implemented externally and executed on-line. Experimental results show good performance in setpoint tracking accuracy and robustness against plant/model mismatch. The proposed strategy appears to be a promising alternative to heavy computation nonlinear control strategies and not optimal linear control strategies.
文摘建立了交流稳压电源主电路数学模型并分析其闭环稳压控制原理。由于装置具有较强的非线性和变结构、变参数特性,采用经典PID控制器很难获得理想的控制效果。将人工神经网络与传统PID控制器相结合,构成一种不依赖于被控对象精确数学模型的神经网络PID控制器。为了提高神经网络的收敛速度,采用Levenberg-Marquardt算法计算连接权值更新量,并对当前解施加一个以一定概率保留的随机扰动,加快迭代过程跳出局部极小点。对装置主电路和改进神经网络PID控制器进行仿真,结果表明:系统动态响应快,鲁棒性强,调节平滑,具有较好的控制效果。最后,制造并测试了额定电压660 V、容量400 k VA的实验样机,对理论研究进行了实验验证。
基金the National High Technology Research and Development Program of China (863 Program) (No.2006AA11Z208)National Program on Key Basic Research Project (973 Program)(No.2006CB705506)
基金Supported by CAS Knowledge Innovation Project(U-602(IHEP),U-34(IHEP))National Natural Science Foundation of China (10491300,10605030)100 Talents Program of CAS(U-54,U-25)
文摘A multilayered perceptrons＇ neural network technique has been applied in the particle identification at BESIII. The networks are trained in each sub-detector level. The NN output of sub-detectors can be sent to a sequential network or be constructed as PDFs for a likelihood. Good muon-ID, electron-ID and hadron-ID are obtained from the networks by using the simulated Monte Carlo samples.
