论文部分内容阅读
提出了两种基于模糊控制的神经网络控制器的设计方案，并将这两种控制器应用于综合火力／飞行系统的耦合控制。两种方法的主要差别在于获取样本的方式不同。方案１是通过对模糊控制方法得到的响应曲线采样获取样本，由Ｂａｃｋ－Ｐｒｏｐａｇａｔｉｏｎ学习算法训练神经网络，得到一组固定权值。神经网络控制器采用这组权值以“联想记忆”的方式工作。方案２则从用模糊控制算法得到的控制查询表中获取样本。因为模糊控制查询表比较大，采样时依据该表构成的相平面图的特点，对采样点数进行了压缩，使所设计的神经网络的规模可以接受，其余的设计步骤与方案１基本相同。仿真结果表明，采用这两种神经网络控制器的控制系统都具有良好控制性能
Two design schemes of neural network controller based on fuzzy control are proposed, and these two controllers are applied to the integrated control of firepower / flight system. The main difference between the two approaches is the way in which samples are taken. Scenario 1 samples the response curves obtained by the fuzzy control method and samples the neural network by the Back-Propagation learning algorithm to obtain a set of fixed weights. The neural network controller uses this set of weights to work in “associative memory.” Scenario 2 obtains samples from the control look-up table obtained from the fuzzy control algorithm. Because the fuzzy control lookup table is relatively large, the sample points are compressed according to the characteristics of the phase plan formed by the table during sampling, so that the size of the designed neural network is acceptable. The rest of the design steps are basically the same as the solution 1. The simulation results show that the control system using these two neural network controllers has good control performance