主汽温的改进型神经网络辨识Smith预估器.pdfVIP

主汽温的改进型神经网络辨识Smith预估器孙宇贞．等

主汽温的改进型神经网络辨识Smith预估器

Sm．thPredictorBPNeuralNetwork

ImprovedbyRecognition

forMainSteamControl

Temperature

匆．穹点1’2高楞1彭遵曰，)1，2

摘要：针对超临界机组主汽温对象具有的大惯性、大滞后和非线性等特点，提出了一种基于BP神经网络辨识的改进型Smith预估

方案。采用改进的Smith预估器克服主汽温对象的大滞后特性，并通过神经网络辨识来改善Smith预估控制器对模型精度的依赖，提

高其抗干扰能力。通过对超临界机组主汽温对象在不同工况下的仿真，表明该系统对于对象特性变化有较好的适应能力，在动态品

质、鲁棒性等方面都明显优于常规Smith控制方案。

关键词：模糊控制器Smith预估器超临界机组BP神经网络大滞后主汽温控制辨识鲁棒性

中圈分类号：TH39；TP2文献标志码：ADOI：10．16086／j．enid．issnl000—0380．201607022

atthecharacteristicsoftimeandofthecontrolledofmainsteam

Abstract：Aiming18rgeinertia．1argedelaynonlinearityobjecttemperature-an

basedofmain

schemeofSmithwhichisonBPneuralnetworkisoftime

improvedrecognitionshortcominglarge

predictorproposed．Thedelay

steamisovercomeSmithonmodelofSmithisBP

temperatureobjectbyimprovedprodictor-thedependencyprecisionpredictorimprovedby

ofisatfor

neuralnetworktheanti—disturbanceenhancedthesametime．Theresultofsimulationthemainsteam

recognition-andability

in

underdifferentconditionsshowsthatthecontrolhastotothe

temperatureoperatingsupercriticalpowerplantsystemgoodabilityadapt

variationofistothetraditionalSmithcontrolschemeand

objectcharacteristics；itobviouslysuperiorpredictorin由删cqllalityrobustness．

SmithunitBP

controllerneuralnetworkMainstextmcontrol

Keywords：FuzzypredictorSupercriticalIargetemperature