工业过程迭代动态规划算法研究

《工业过程迭代动态规划算法研究》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。

1、AbstractSteady‘stateduringthecomplexindustrialprocessisoftentemporaryandrelative，becauseofthevariousfluctuations，disturbancesandconditionchangesinpractice．Whilethetraditionalmethodsinchemistryindustrialprocessaredifficulttodealwiththoseproblemscausedbydynamicchanges

2、，dynamicoptimizationhasbecomeveryurgentinacademicandindustryresearchduetoitsexcellentperformanceindealingwithdynamicchanges．Inthispaper,someexplorationsonthebasisofthecurrentresearchstatusaredevelopedonthreeaspects：theimprovementsforsolvingdifferentialequationsdurin

3、giterativedynamicprogrammingprocess，stateconstraint-firststrategyforboundaryfixedproblemsbasedoniterativedynamicprogrammingandallimprovedslackvariableapproacheffectivelyfordynamicprocesseswithinequalitystateconstraints．Aftertheverificationonaseriesofclassicbenchmark

4、sofindustrialprocessdynamicoptimization，theaboveworkplayavisibleroleinthesolutionaccuracy,convergencespeedorstabilityeffect．Tobespecific，themainworkandcontributionsaleasfollows：(1)Sincethesolvingprocessofiterativedynamicprogramminginvolvesavarietyofdifferentialequat

5、ions，thespeedofsolvingequationwouldlargelyaffecttheoveralloperationefficiency．Inthispaper,severalimprovementsareproposedtoattempttoraisetherateofequationrunning，includingimprovedfourth-orderRunge·Kuttaalgorithm，Eulermethodandmulti-stepAdamsalgorithm．Conclusionsshowt

6、hatthefourth··orderRunge--Kuttacanachievethebalancedeffectbetweentheaccuracyandrunningspeed，while，theimprovedEuleralgorithmcanreachthebetterresultsbutwithlongertime，andAdamsalgorithmwouldinmoredegreegettheexcellentoptimizationaccuracyandefficiency．(2)Aimingatthedyna

7、micoptimizationproblemwithfinalvalueconstraints，astateconstraint—firststrategybasedoniterativedynamicprogrammingisproposed，where，theperformancegoalandconstraintsgoalaredividedintoseparatetwoindividualoptimizationproblems．Aftertheconstraintsgoalisachieved，thefeasible

8、solutionsaresubstitutedintotheperformancefunctiontosearchthebestcontrolvariable．Thentheoptimalcontroltrajectoriesandoptimalstatestrajector