行业应用与交流IndustrialApplicationsandCommunications《自动化技术与应用》2023年第42卷第6期TechniquesofAutomation&Applications基于改进神经网络与五防规范的操作序列的自动化生成*邓皓元,喻启俊,刘远超,梅春晓,肖婉清(网湖北省电力有限公司超高压公司,湖北武汉430050)摘要:为了高效、安全地制定相应的操作票,提出了一种基于改进人工神经网络的操作序列自动生成方法。利用改进粒子群(PSO)算法对广义径向基神经网络中相关参数进行寻优,并通过设定编码规则,规范输入输出的编码格式与含义,并据此将操作任务目标数字化后输入,利用五防规则对其生成的预结果进行保留或修改,使之最终生成对应的操作序列,从而辅助变电运行人员完成操作票的制定。最后利用湖北某变电站操作票序列做实例验证,实验表明利用该方法自动生成操作序列较传统的人工方法效率更高、更为安全、具有较强的智能性。关键词:改进粒子群优化;广义RBF神经网络;电气五防;自动生成中图分类号:TP183文献标识码:A文章编号:1003-7241(2023)06-0178-05AutomaticGenerationofOperationSequenceBasedonImprovedNeuralNetworkandFivePreventionSpecificationsDENGHao-yuan,YUQi-jun,LIUYuan-chao,MEIChun-xiao,XIAOWan-qing(StateGridHubeiUltraHighVoltageCompany,Wuhan430050China)Abstract:Inordertoformulatecorrespondingoperationticketsefficientlyandsafely,thispaperproposesanautomaticgenerationmethodofoperationsequencebasedonimprovedartificialneuralnetwork.Itusestheimprovedparticleswarmalgorithmtooptimizetherelevantparametersintheradialbasisfunctionneuralnetwork,andsetsthecodingrulestostandardizethecodingformatandmeaningoftheinputandoutput.Accordingtothis,theoperationtasktargetisdigitizedandinput,andthefivedefenserulesareusedtothepre-resultsgeneratedareretainedormodified,sothatthecorrespondingoperationsequenceisfinallygenerated,soastoassistthesubstationoperatortocompletetheformulationoftheoperationticket.Finally,theoperationticketsequenceofacer-tainsubstationinHubeiisusedforexampleverification.Theexperimentshowsthatusingthismethodtoautomaticallygeneratetheoperationsequenceismoreefficient,saferandmoreintelligentthanthetraditionalmanualmethod.Keywords:ImproveParticleSwarmOptimization;generalized...