Abstract:Therationalconstructionofislandintegratedenergysystemisofgreatsignificancetothecleantransformationofcoastalenergy,anditsoptimaldispatchisaneffectivewaytoachievethebalancebetweenenergysupplyandenergydemandinislands.Basedonthis,anoptimaldispatchmethodforislandintegratedenergysystemconsideringwindpowerpredictionisproposed.Firstly,thesystemmodelincludingadvancedenergyconversionequipmentisbuilt,suchashydrogenenergyequipment,seawatersourceheatpump,seawaterdesalinationdevice,andwavepowergenerationdevice.Secondly,sincemarineweathercanleadtoinstabilityinrenewableenergygeneration,theConvolutionalNeuralNetwork-Bi-directionalLongShort-TermMemory(CNN-Bi-LSTM)modelwithanimportancerankofenvironmentalvariablesisappliedtopredictpowergeneration.Then,tomaintainthebasiclivingconditionsofislands,takingtheelectric-cold-freshwater-hydrogenbalanceastheconstraintsandtakingtheimprovementsofsystemoperationeconomyandrenewableenergyconsumptionrateastheobjectivefunctions,anoptimaldispatchmodeloftheintegratedenergysystemisestablished.Thesimulationresultsoftwotypicaldaysinsummerandwintershowthattheproposedpredictionmodelhashighpredictionaccuracy.Theproposedoptimizationdispatchmethodcanachievethebalancebetweenenergysupplyandenergydemandoftheisland,effectivelyreducethesystemoperatingcost,andimprovetherenewableenergyconsumptionrate.Keywords:islandintegratedenergysystem;wavepowergeneration;windandphotovoltaicpowergenerationprediction;optimaldispatchmodel;renewableenergyconsumptionrate摘要：合理构建海岛综合能源系统对沿海能源清洁化转型意义重大，其优化调度更是实现海岛能源供需平衡的有效途径。为此，提出了一种考虑风光功率预测的海岛综合能源系统优化调度方法。首先，搭建包含氢能设备、海水源热泵、海水淡化装置、波浪能发电装置等新型能源转换设备的系统模型。其次，海上气候多变会导致新能源发电不稳定，故采用含环境变量重要性排序的一维卷积神经网络和双向长短时记忆神经网络（convolutionalneuralnetwork-bi-directionallongshort-termmemory，CNN-Bi-LSTM）联合模型对发电功率进行预测。然后，为维持海岛基本生存条件，以电-...
