Neural Net Research Based On Fuzzy Algorithm

Intelligence research could improve operation's level of automatization and intelligetion,and it could achieve exact point job so that it advance production continued development.And in modern intelligence research,fuzzy neural net is attention research realm in resent years,it connected with neural net's low layer and computing power and fuzzy logic system's high layer copy thought and deduce abilite,and come off superiority complementary between them. Neural net and fuzzy control are two important intelligence control technology, they can imitate person's intelligence behave,solove notsure non-linear,complicational automatization problem,have very wide applied foreground.The research of neural net and fuzzy control has been 30year's history.In resent years,they not only get the catch man's eye's development and achievement in their each subject realm,but also develop many new realms in the eadge of the two subjects,connevted with neural net and fuzzy control algorithm,make themselves osmosis each other,will bring a new revolution in the realm of information science.Fuzzy control is an typical and prior form of intelligence control,as a branch of intelligence control,it get very fast development in resent years.exceptional for big time-lag.non-linear and others complicative systems that it's hard to build exact math modol,we use fuzzy control theory,through computer come ture realtime control,can get content control effect,and need equipment simply,economic benefit notable.in process of modern industry,fuzzy control is widly used to complicative process or project's control.At present,research of fuzzy neural net has been an notice research way.fuzzy neural net,which is a new net system that connected fuzzy theory and neural net,which is traditional neural net model fuzziness in some degree,it's a distinct feature is,each nodefuzzy neuron) is diffierent from traditional MP neuron's two-value state,could get the interval[0,1],so,each node could express fuzzy concept.fuzzy neural net has the function that integrated neural net and expert system,when we use neuron net solve the problem of fault diagnosis,it has an obvious trand that hope to integrated qualitative knowledge in the neuron net's frame.for this,fuzzy neural net has been a hot point with research.fuzzy neural net is often enter a fuzziness layer in the ordinary feedforward neural net,complete an input subsidiary function's computing.Put neural net into fuzzy control,could ameliorate and improve adapting ability of fuzzy logic derive,connect neural net and fuzzy control could lead to new algorithm and structure's finding and development,make it could improve its adapting ability of keeping fuzzy logic system strongly knowledge's express.As the fast development of modern agricultural and science technique,production facilities and system more and more large-size,complication and automatization,its acquire more and more high with system's reliable and security,agricultural machine intelligence has been a kind of trade in modern times,and we make fuzzy technique and neural net technique osmosis and coopration each other,has been an indubitability development aspect. The work of this article is as follows:(1) The article introduces the BP network model establishment, the training based on the MATLAB neural network toolbox.(2) Analysis the two kind of model study training error performance curve training error, adaptive neural network convergence rate and quickly network performance standards conclusion.(3) Through the BP neural network with the neural network training, the neural network has strong nonlinear characteristics, satisfy the demand of power load forecasting.