智能控制智能控制智能控制 (31).pdf

《智能控制智能控制智能控制 (31).pdf》由会员分享，可在线阅读，更多相关《智能控制智能控制智能控制 (31).pdf（6页珍藏版）》请在得力文库 - 分享文档赚钱的网站上搜索。

1、Parallel Distributed ProcessingTraditional methodANNSerial/Batch processing Parallel processingANN has better fault tolerance and faster overall processing capabilities.This is particularly suitable for real-time control and dynamic control.Nonlinear Mapping There is a complex non-linear relationshi

2、p between the input and output of the system.It is difficult to establish its mathematical model with traditional mathematical methods.ANN can approximate any complex nonlinear mapping with arbitrary precision.The application of neural network with nonlinear mapping function is very broad,covering a

3、lmost all fields.Work with Modern Science&TechnologyProvide guidance in brain sciences.Fuzzy neural network has become an important research direction.ANN features are conducive to hardware implementation.Learning&AdaptionLearning by training:ANN can solve such control processes that are hard to dea

4、l with by mathematical models and descriptive rules.The strong abilities of adaptation and information fusion make thenetworksprocessmassiveanddifferentcontrolinputssimultaneously,which is specially suited for control of complex,large-scale&multivariable systems.SummaryAbilitiesPotential applicationsLearning adaptabilitySelf-organization abilityFunction approximation abilityLarge-scale parallel processing ability Intelligent control system Nonlinear control system The uncertainty of control problem