摘要
针对挖泥船泥浆管道输送流速控制的大惯性、大时滞、参数时变和建模困难等特点,提出一种单神经元自适应预估控制方案.该方案利用神经网络的自学习能力,对系统结构、参数、不确定性和非线性进行学习,结合Marsik和Streic提出的无辨识自适应控制算法对控制参数进行在线修正,在控制方案中加入Smith预估器,利用搜索寻优的方法对时变的时滞进行在线优化,提高了预估算法的鲁棒性和适应能力.通过现场实验证明了本控制方法的有效性,在疏浚施工环境变化,时滞较大的条件下仍然能够使泥浆流速基本保持稳定,具有较强的抗干扰能力和良好的跟踪性能.
A new single neuron self-adaptive predictive scheme is introduced for controlling the slurry transportation- rate in the dredging pipeline. To deal with such a process of high inertia, long time-delay, time-varying parameters and the difficulty in modeling, this scheme makes use of the self-learning capability of the neuron network to study the control sys- tem structure, parameters, uncertainties and nonlinear characteristics; combines with the identification-free self-adaptive control algorithm proposed by Marsik and Strejc to carry out the on-line adjustment of control variables; incorporates the Smith predictor and employs the optimization searching algorithm for on-line optimizing the time-delay parameter to enhance the robustness and adaptability of the predictive algorithm. Field experiments are also carried out to test the performance of the proposed control scheme. Results show that the control performance is satisfactory in various dredg- ing environments, even the time-delay is significant; the control system compensates external disturbances and exhibits desirable tracking ability.
出处
《控制理论与应用》
EI
CAS
CSCD
北大核心
2009年第3期309-312,共4页
Control Theory & Applications
基金
浙江省湖州市科技攻关计划资助项目(2004GG45).
关键词
疏浚
泥浆管道输送
流速控制
自适应预估控制
dredging
slurry pipeline transport
transportation-rate control
self-adaptive predictive control
