Adaptive Neural Network Architectures For Emission Control of Engines (TSE-03G)

用于发动机排放控制的自适应神经网络架构 (TSE-03G)

• 批准号: 0327877
• 负责人: Jagannathan Sarangapani
• 金额: --
• 依托单位: Missouri University of Science and Technology
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 2003
• 资助国家: 美国
• 起止时间: 2003-09-01 至 2008-08-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=0327877&HistoricalAwards=false
• 关键词: Adaptive; Neural; Network; Architectures; Emission

The goal of this project is to provide the next generation emission controller for engines with adaptation, optimization, and learning so as to achieve low NOx and improved fuel economy. The controller will guarantee performance in the presence of unknown nonlinearities, feedback delays, and is supported by a rigorous design and mathematical framework. Specific objectives to support that goal are:1. Develop a robust adaptive NN control scheme for SI engines that would minimize the effect cyclic dispersion at very lean operation (equivalence ratio =0.7) by selecting appropriate feedback parameters.2. Study and model the complex dynamics in cyclic output for SI engines under the influence of high-levels of EGR. Investigate potential for reducing NOx over 50% below current EGR systems.3. Develop a robust adaptive critic NN EGR control scheme that would minimize the effect of cyclic dispersion and would allow satisfactory performance in low NOx regimes via appropriate feedback. Provide methodology to integrate combustion stability with high levels of EGR in SI engines.4. Simulate and verify the EGR and lean stability controller performance on an experimentally validated model. Demonstrate the controller schemes on a single cylinder engine in the laboratory.5. Investigate the complex dynamics in output for diesel engines under the influence of high levels of EGR with an objective of reducing the NOx over 50% with minimal particulate matter. Provide recommendations about the applicability of the proposed EGR controllers for diesel engines.These projects will be pursued in collaboration with organizations such as Caterpillar, Inc. and Oak Ridge National Laboratory. These areas of research could lead to significant advances in the development of advanced control schemes for non-strict feedback nonlinear systems such as next generation spark ignition, diesel as well as nontraditional engines such as homogeneous charge compression ignition, direct injection spark ignition and hybrid engines (electric power and gasoline).

该项目的目标是为发动机提供具有自适应、优化和学习功能的下一代排放控制器，以实现低NOx和提高燃油经济性。该控制器将保证在未知的非线性，反馈延迟的存在下的性能，并支持一个严格的设计和数学框架。支持这一目标的具体目标是：1。为火花点火发动机开发了一种鲁棒的自适应神经网络控制方案，通过选择合适的反馈参数，使发动机在非常稀薄的工况（当量比=0.7）下循环弥散的影响最小.研究和模拟高EGR水平影响下SI发动机循环输出的复杂动态特性。研究降低NOx的潜力，使其低于当前EGR系统50%以上。开发一种鲁棒的自适应临界神经网络EGR控制方案，该方案将使循环分散的影响最小化，并通过适当的反馈在低NOx状态下实现令人满意的性能。提供了一种方法来整合燃烧稳定性与高水平的废气再循环在SI发动机。在经过实验验证的模型上模拟并验证EGR和稀燃稳定性控制器的性能。在实验室中以单缸发动机为例演示控制器方案.研究柴油发动机在高水平EGR影响下的复杂动力学输出，目标是在最小颗粒物的情况下减少NOx超过50%。就柴油发动机EGR控制器的适用性提供建议。这些项目将与卡特彼勒公司等组织合作进行。和橡树岭国家实验室。这些研究领域可能会导致非严格反馈非线性系统，如下一代火花点火，柴油以及非传统发动机，如均质充量压缩点火，直接喷射火花点火和混合动力发动机（电力和汽油）的先进控制方案的发展取得重大进展。