基于加权朴素贝叶斯算法的调度指挥态势感知模块设计

Design of Dispatching Command Situation Awareness Module Based on Weighted Naive Bayes Algorithm

针对调度与现场频繁电话沟通调度操作进程信息,以及无法提供给现货市场较精确的设备停复电时间作为边界条件的问题,将大数据挖掘理念引入了调度指挥领域,在南方电网调度指挥控制系统(Power Dispatch Command Control System, DCCS)的基础上,设计了一种对调度指挥态势进行实时感知的功能模块。该模块可通过DCCS获取调度员在业务准备关键节点的业务环境特征数据,采用加权朴素贝叶斯算法对调度员的实时忙碌程度进行挖掘,结合历史数据对忙碌系数进行匹配,并动态更新预计开展时间,得到结果后可传输至现货系统,作为节点电价计算的边界条件。此外,该模块还具备操作预约、操作通知、智能编排、态势展示等功能。模块上线后,减少了调度与现场有关操作信息的无谓沟通,所构建的数据模型也取得了较好的预测准确率。

In order to solve the problem of frequent telephone communication between dispatchers and field, as well as the failure to provide the spot market with more accurate equipment outage and recovery time as the boundary condition, this paper introduces the concept of big data mining into the field of dispatching command for the first time on the basis DCCS, a function module with real-time situation awareness of dispatching command is designed. The module can obtain the business environment characteristic data of dispatchers at the key nodes of business preparation through DCCS, mine the real-time busy degree of dispatchers by using weighted naive Bayes algorithm, match the busy coefficient with historical data, dynamically update the expected development time, and then transmit the results to the spot system as the boundary conditions of the node price calculation. In addition, the module also has the functions of operation reservation, operation notification, intelligent arrangement, situation display and so on. After the module goes online, it reduces the unnecessary communication between dispatching and field for operation information, and the data model constructed also achieves good prediction accuracy.