风电业务流程剩余时间预测与系统应用OA
Wind Power Business Process Remaining Time Prediction and System Application
针对工业场景下运维事件日志复杂以及数据量较少等问题,提出一种基于时间卷积网络的剩余时间预测方法.首先,使用因果卷积对流程实例数据进行建模;其次,对不同长度的轨迹前缀分别进行模型训练,以提高模型的针对性;最后,将剩余时间预测模型用于实际的风电运维业务场景中,开发风电运维业务系统,实现工作票表单数据以及事件日志数据的可视化展示.与传统的剩余时间预测方法相比,该方法的预测效果有显著提升,平均绝对误差值降低了5%左右.开发的风电运维业务系统能够预测未完成工作票开票过程中流程的剩余时间,实现业务流程超时告警.
Aiming at the complexity of industrial scene event logs and the small amount of data in industrial scenarios,a remaining time pre-diction method based on time convolutional network is proposed.Firstly,causal convolution is used to model process instance data.Then,the model is trained on track prefixes of different lengths to improve the pertinence of the model.Finally,the remaining time prediction model is used in the actual wind power operation and maintenance business scenario to develop the wind power operation and maintenance business sys-tem to realize the visual display of work ticket form data and event log data.Compared with the traditional remaining time prediction method,the prediction effect of the proposed method is significantly improved,and the average absolute error is reduced by an average of 5%.The de-veloped wind power wind power operation and maintenance business system can predict the remaining time in the process of invoicing unfin-ished work tickets,and realize business process overtime alarms.
李建元;田玉超;尹昱妍;田甜;刘新锋
中国移动通信集团山东有限公司山东建筑大学 计算机科学与技术学院,山东 济南 250101
计算机与自动化
业务流程分析剩余时间预测时间卷积网络风电业务流程流程可视化
business process analysisremaining time predictiontemporal convolution networkwind power business processprocess vi-sualization
《软件导刊》 2024 (006)
18-24 / 7
国家自然科学基金项目(51975332);山东省重点研发计划(重大科技创新工程)项目(2021CXGC011204)
评论