简介：Afuzzyneuralnetworkcontrollerforunderwatervehicleshasmanyparametersdifficulttotunemanually.Toreducethenumerousworkandsubjectiveuncertaintiesinmanualadjustments,ahybridparticleswarmoptimization(HPSO)algorithmbasedonimmunetheoryandnonlineardecreasinginertiaweight(NDIW)strategyisproposed.OwingtotherestraintfactorandNDIWstrategy,anHPSOalgorithmcaneffectivelypreventprematureconvergenceandkeepbalancebetweenglobalandlocalsearchingabilities.Meanwhile,thealgorithmmaintainstheabilityofhandlingmultimodalandmultidimensionalproblems.TheHPSOalgorithmhasthefastestconvergencevelocityandfindsthebestsolutionscomparedtoGA,IGA,andbasicPSOalgorithminsimulationexperiments.ExperimentalresultsontheAUVsimulationplatformshowthatHPSO-basedcontrollersperformwellandhavestrongabilitiesagainstcurrentdisturbance.ItcanthusbeconcludedthattheproposedalgorithmisfeasibleforapplicationtoAUVs.