新型智能操作票系统应用研究  

Application Research of New Intelligent Operation Ticket System

作　　者：宋哲 冷磊磊 SONG Zhe;LENG Lei-lei(Beijing Kedong Electric Power Control System Co.,Ltd.,Beijing 100193,China)

机构地区：[1]北京科东电力控制系统有限责任公司,北京100193

出　　处：《电力学报》2022年第1期84-90,共7页Journal of Electric Power

摘　　要：现有的操作票生成方式已经难以满足电网安全运行的要求,其耗时较长、无法实时开票等缺陷严重影响了电网的运行安全及效率,因此利用智能化方式实现操作票的生成变得尤为必要。为了缩短操作票生成所耗时间,提高调度员的工作效率,对操作票现有的生成方式进行了分析研究,找出操作票现有生成方式的缺陷。为了弥补操作票现有生成方式的缺陷且降低人为因素对操作票生成的影响,分别将多种智能推理算法及深度学习算法进行对比。对比结果显示BiLSTM+CRF算法的语义解析完全匹配率最高、语义解析平均时间最短,故选择该算法作为深度学习算法;根据对50条操作任务进行对比试验的结果,选取推理时间、操作项完成率、操作步骤正确率均满足实际要求的专家系统推理方案作为智能推理算法。在此基础上搭建的新型智能操作票系统,该系统在智能推理及深度学习算法的基础上,融合模拟预演、逆向出票及防误校验功能,提高操作票的生成质量,可实现智能化方式生成操作票。将新型智能操作票系统投入现场测试运行,分别对深度学习算法和推理算法进行了效果验证。在对20条4类(共80条)检修申请单进行试验测试的结果显示,在保持较高正确率的基础上,操作票生成耗时约90 s,较其他两个系统大幅缩短了成票时间。由此可得,新型智能操作票系统对提升调度人员的工作效率、提高调度质量及电网安全稳定运行具有重要意义。The existing operation ticket generation method is hard to meet the requirements of safe operation of power grid.Its defects such as long time-consuming and unable to issue tickets in real time etc.affect the operation safety and efficiency of power grid.Therefore,it is particularly necessary to use intelligent method to realize the generation of operation ticket.In order to shorten the time of operation ticket generation and improve the work efficiency of dispatchers,the existing generation methods of operation ticket are analyzed and studied to find out the defects of the existing generation methods of operation ticket.In order to make up for the defects of the existing generation methods of operation ticket and reduce the impact of human factors on the generation of operation ticket,a variety of intelligent reasoning algorithms and deep learning algorithms are compared.The comparison results show that BiLSTM+CRF algorithm has the highest semantic parsing perfect matching rate and the shortest average semantic parsing time,so this algorithm is selected as the deep learning algorithm.According to the result of the comparative experiment of 50 operation tasks,the inference scheme of expert system which can meet the actual requirements of inference time,operation item completion rate and operation step accuracy rate is selected as the intelligent inference algorithm.On this basis,a new intelligent operation ticket system is built.On the basis of intelligent reasoning and deep learning algorithm,the new intelligent operation ticket system integrates the functions of simulation rehearsal,reverse ticket issuance and error proof verification to improve the generation quality of operation ticket and realize the intelligent generation of operation ticket.The new intelligent operation ticket system is put into field test,and the effect of deep learning algorithm and inference algorithm is verified respectively.The test results of 20 type 4 maintenance application orders(total 80 items)showed that on the basis of maintain

关 键 词：新型智能操作票系统 深度学习 智能推理 模拟预演 逆向成票 

分 类 号：TM73[电气工程—电力系统及自动化]