论文部分内容阅读
阐述一种基于特殊梯度下降——短暂反向传播并通过自学习来获取被控对象的方法，是一种近似最优方式的控制策略，它能够处理以部分形式描述的被控对象，如差分方程、神经网络以及模糊模型等，这就增强了控制器的功能。
A method based on special gradient descent-ephemeral reverse propagation and acquiring controlled objects through self-learning is presented. It is an approximation optimal control strategy that can handle controlled objects described in partial form, such as differential Equation, neural network and fuzzy model, which enhances the controller’s function.