论文部分内容阅读
基于神经网络方法和状态预测补偿法,提出应用于含时滞空间结构振动控制的新方法。运用单个神经网络代替由两个神经网络组成的控制系统,以减少时滞产生环节和系统时滞。构造了基于神经网络的多步预测时滞补偿策略。研究表明:该文时滞补偿策略可有效运用于受不确定性外界荷载作用的多自由度大跨空间结构的风振控制,且不依赖于实际中难以获得的被控系统的精确数学模型;经时滞补偿后的结构振动控制效果优于未经时滞补偿的控制效果。
Based on neural network method and state prediction compensation method, a new method applied to vibration control of space structures with time-delay is proposed. A single neural network is used instead of the control system composed of two neural networks to reduce the time lag and system delay. A time-delay compensation strategy based on neural network is constructed. The research shows that the proposed compensation strategy can be effectively applied to the wind-induced vibration control of multi-degree-of-freedom long-span space structures subjected to uncertain external loads and does not depend on the accurate mathematical model of the controlled system that is difficult to obtain in practice. The structure vibration control effect compensated by time delay is better than the control effect without time delay compensation.