Robust Adaptive Critic Neural Network Control of a Class of Nonlinear Dynamic Systems

一类非线性动态系统的鲁棒自适应批评神经网络控制

• 批准号: 0621924
• 负责人: Jagannathan Sarangapani
• 金额: $ 24万
• 依托单位: Missouri University of Science and Technology
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2006
• 资助国家: 美国
• 起止时间: 2006-09-01 至 2011-04-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=0621924&HistoricalAwards=false
• 关键词: Robust; Adaptive; Critic; Neural; Network

Proposal Number: 0621924Proposal Title: Robust Adaptive Critic Neural Network Control of a Class of Nonlinear Dynamic SystemsPI Name: Sarangapani, JagannathanPI Institution: University of Missouri-RollaThe objective of this research is to develop optimal controllers using approximate dynamic programming for nonlinear discrete-time systems with uncertainties and disturbances. Supervised actor-critic neural network architectures will be used to solve the equations resulting from the receding horizon optimal control formulation. Intellectual Merit Intellectual merit of this research lies in realizing unified controller solutions to optimal control problems for uncertain nonlinear discrete-time systems that are important, yet difficult to solve. Outcome of this research is expected to advance the field of control in the area of robust adaptive control. Additionally, expected results will advance the state of the art in the field of neural network control since these investigations are based on new approximate dynamic programming formulations with neural network based solution structures. Broader ImpactsBroader impact of this research will be in the applications areas of fuel flexible engine and spark ignition engine control. Optimal control solutions to such problems could lead to the drastic reductions in engine-out emissions while improving efficiency and adaptability. Outreach activities planned in this proposal include using women and minorities and disseminating results through the newly established NSF Industry/University Cooperative Research Center Site. The project results will be presented through demonstrations to the area high schools in order to create early career interest in science and engineering.

提案编号：0621924提案题目：一类非线性动态系统的鲁棒自适应临界神经网络控制PI姓名： Sarangapani，JagannathanPI机构：密苏里大学罗拉本研究的目的是开发最优控制器，使用近似动态规划的非线性离散时间系统的不确定性和干扰。有监督的行动者-批评者神经网络架构将用于求解由滚动时域最优控制公式产生的方程。智能优点这项研究的智能优点在于实现统一的控制器解决方案的不确定非线性离散时间系统的最优控制问题是重要的，但难以解决。本文的研究成果有望推动控制领域在鲁棒自适应控制领域的发展。此外，预期结果将推进神经网络控制领域的最新技术水平，因为这些研究是基于新的近似动态规划公式和基于神经网络的解决方案结构。更广泛的影响这项研究将在燃料灵活的发动机和火花点火发动机控制的应用领域产生更广泛的影响。这些问题的最优控制解决方案可以导致发动机排放的大幅减少，同时提高效率和适应性。 本提案中计划的推广活动包括利用妇女和少数民族，并通过新成立的国家科学基金会工业/大学合作研究中心网站传播成果。该项目的结果将通过示范介绍给该地区的高中，以创造对科学和工程的早期职业兴趣。