反应溅射的PID神经网络控制仿真.pdfVIP

真空科学与技术学报 第29卷第4期 364 CHINESEJOURNALOFVACUUMSCⅡNCEANDTECHNOLOGY 2009年7．8月 反应溅射的PID神经网络控制仿真 王振伟1牟宗信1*刘升光2 (1．大连理工大学三束材料改性实验室大连116024；2．大连理工大学物理与光电工程学院大连116024) Simulationof DifferentiationNeural ProportionIntegration Network ControlofReactive Sputtering WangZhenweil，MuZongxinH，ⅡuShengguan92 andElectronBeams，Dalian 116024，Ch／na； (1．1aboratoryofMaterialsModificationbylaser，Ion UniversityofTechno／ogy，Da／／an 2．School and ofPhysicsOptoe／ectron／ctechao／ogyofDUT，Da／／an116024，Ctdna) Abstract1hereactive the differentiationneural sputtering，controlledby proportionintegration was onthebasis ofthe in flowrateon simulated of model．1he S．Berg’s effect，thatis，theimpact changesgas hysteresis dramaticvariationsin rateand studiedtoachievea film gaspartial steadygrowth pressure，depositiontargetvoltage，wag andtomakethebestuseofthereactive simulatedresultsshowthatthePIDNNcontrollerdoesa insta- gas．r111e goodjob thereactive a fast and resistance bilizing sputtering埘tIlquite convergence，rapidresponsestrong few曲哪s，including tointerference． Reactive effect，PIDNN model Keywords sputtering，Hysteresiscontrol，S．Berg’s 摘要在反应溅射工艺中，反应气体流苗和沉积速率、反应气体气压、靶电压之问存在迟滞关系。为了保证反应溅射稳 的快速收敛特点和神经网络的非线性逼近能力，适合对非线性反应溅射系统的控制。本文用PID神经网络(PIDNN)控制器对 反应溅射系统进行精