• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>FPST-vol.13; American Society of Mechanical Engineers(ASME) International Mechanical Engineering Congress and Exposition; 20061105-10; Chicago,IL(US) >FEASIBILITY STUDY ON THE USE OF DYNAMIC NEURAL NETWORKS (DNN's) FOR MODELING A VARIABLE DISPLACMENT LOAD SENSING PUMP
【24h】

FEASIBILITY STUDY ON THE USE OF DYNAMIC NEURAL NETWORKS (DNN's) FOR MODELING A VARIABLE DISPLACMENT LOAD SENSING PUMP

机译：使用动态神经网络（DNN）建模可变位移载荷传感泵的可行性研究

获取原文
获取原文并翻译 | 示例
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

The feasibility of using a particular form of neural networks, defined as Dynamic Neural Units (DNU's), to model a pump in a load sensing system is investigated in this paper. Because of the highly complex structure of the pump, its compensators and controlling elements, simulation of load-sensing pump systems pose many challenges to researchers. Several models of pumps, compensators and valves have been developed and published in the literature but they are overly simplified or are in an extremely complex form. One modeling approach which can capture the nonlinear dynamic properties of the pump yet still retain reasonable simplicity in its basic form is to use neural network technology. Previous studies have shown some limited success in using feed forward neurons with dynamic properties being introduced using time delays. A problem referred to a error accumulation has prevented these neural based models from being practical dynamic representations of load sensing systems. Based on the topology of the biological neural systems several new structures, Dynamic Neural Units (DNU's) have been developed. Only one DNU is necessary to capture or represent some of the dynamics of a plant, which a static (feed forward) neuron cannot do. The main advantage of the dynamic neuron is that it reduces the network dimension and the amount of computational requirement and has the potential to avoid this error accumulation problem. The use of Dynamic Neural Networks with Dynamic Neural Units in simulating a variable displacement pump is presented in this paper. Only the pump portion of the load sensing pump system is considered due to problems of interacting operating points. A DNU structure and a DNN(which is comprised of DNU's) are introduced. The simulation results establishes the feasibility of using a Dynamic Neural Networks with DNU's to model a simulated nonlinear hydraulic system such as a load sensing pump .
机译：本文研究了使用特定形式的神经网络（称为动态神经单元（DNU））对负载传感系统中的泵进行建模的可行性。由于泵，其补偿器和控制元件的高度复杂，负载传感泵系统的仿真给研究人员带来了许多挑战。泵，补偿器和阀门的几种型号已经开发并发表在文献中，但是它们过于简化或形式极为复杂。一种可以捕获泵的非线性动态特性但仍保留其基本形式合理简单性的建模方法是使用神经网络技术。先前的研究表明，使用前馈神经元具有有限的成功，其动态特性是通过时间延迟引入的。称为错误累积的问题阻止了这些基于神经的模型成为负载传感系统的实际动态表示。基于生物神经系统的拓扑结构，已经开发了几种新的结构，动态神经单元（DNU）。捕获或表示植物的某些动力学只需要一个DNU，而静态（前馈）神经元则无法做到。动态神经元的主要优点是它减少了网络规模和计算需求量，并有可能避免这种错误累积问题。本文提出了带有动态神经单元的动态神经网络在模拟变量泵中的应用。由于工作点相互作用的问题，仅考虑了负载传感泵系统的泵部分。介绍了DNU结构和DNN（由DNU组成）。仿真结果确定了使用带有DNU的动态神经网络对诸如负载传感泵之类的非线性液压系统进行建模的可行性。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号