论文部分内容阅读
为获得米量级以上的大口径光栅,光栅拼接技术已被公认为是一种经济可行的方法。而在光栅拼接中,光栅拼接稳定性的控制是核心问题之一,因此,在较高拼接精度的要求条件下(错位误差在纳米量级,角度误差在亚微弧度量级),寻求有效合理的比例积分微分(PID)闭环控制算法显得异常重要。将基于BP神经网络的整定方法应用到PID控制中,将传统的用零初态时过程加入单位阶跃输出的第一个值用计算符号函数值来代替,实现了单神经元自适应PID控制;在此基础上,为了改善系统响应初期的上升时间,采用变化神经元比例系数的值来代替常量K值的学习算法。通过仿真分析表明,提出的控制算法使系统响应具有好的快速性的同时又不会产生大的超调量;实验结果也表明改进的控制算法能保证其光栅拼接误差控制精度。
In order to obtain a large aperture grating of more than a meter in size, the technique of grating splicing has been recognized as an economically feasible method. In the raster splicing, the stability of the raster splicing control is one of the core issues, therefore, in the requirements of high splicing accuracy (misalignment error in the nano-scale, angular error sub-micro-scale magnitude), to seek effective and reasonable The proportional integral derivative (PID) closed-loop control algorithm is very important. The tuning method based on BP neural network is applied to the PID control, the traditional zero initial state process is added to the first value of the unit step output to replace the calculated value of the symbol function to realize the single neuron adaptive PID control On this basis, in order to improve the initial rise time of the system response, the value of changing neuron proportional coefficient is used instead of the learning algorithm of constant K value. The simulation results show that the proposed control algorithm can make the system response fast and without large overshoot. The experimental results also show that the improved control algorithm can ensure the accuracy of the control of the splicing error.