Neural Networks, Youla Parameterization and Reconfigurable Control Systems

精确自动化将研究一种新的控制体系结构，其中使用神经网络为控制器设置Youla参数的系数。通过这样做，可以实现与神经控制相关的所有“通过学习设计”的好处，同时保证最终系统的稳定性。此外，只要指定了新的参考输入或性能标准，所建议的网络就会自动更新Youla参数。具体研究目标包括：-开发适用于拟议系统的神经网络架构和训练方法，-制定每次运行该系统时更新网络训练的方法；-将控制器实现为一个自动可重构系统，并将所提出的神经控制器应用于飞行控制系统和机器人。

Accurate Automation will investigate a new control architecture in which a neural network is used to set the coefficients of the Youla parameter for the controller. By so doing one achieves all of the "design by learning" benefits associated with neural control while simultaneously guaranteeing that the resultant system will be stable. Moreover, the proposed network automatically updates the Youla parameter whenever a new reference input or performance criteria is specified. Specific research objectives include the: - development of neural network architectures and training methods applicable to the proposed system, - formulation of methods for updating the network training each time this system is operated, - implementation the controller as an auto- reconfigurable system, and - application of the proposed neural controller to flight control systems and robotics.