采用值守作业机器人进行短期电力负荷预测方法与应用  

Short Term Power Load Forecasting Method With the Help of On-duty Robot

作　　者：张伟 张潇 冯磊 郭梓晗 秦轩 梁庆华[2] ZHANG Wei;ZHANG Xiao;FENG Lei;GUO Zihan;QIN Xuan;LIANG Qinghua(State Grid Jiangsu Electric Power Co.,Xuzhou Branch,Xuzhou Jiangsu 221000,China;School of Mechanical Engineering,Shanghai Jiao Tong University,Shanghai 200240,China)

机构地区：[1]国网江苏省电力有限公司、徐州供电公司,江苏徐州221000 [2]上海交通大学机械与动力工程学院,上海200240

出　　处：《机械设计与研究》2022年第6期29-33,共5页Machine Design And Research

基　　金：国家电网公司总部科技项目(5700-202018266A-0-0-00)。

摘　　要：电力负荷的精确预测为资源的优化调度起到重要作用,然而传统的自动化采集系统难免出现电力数据缺失情况而影响后续的数据分析。在值守作业机器人现场实时监测的辅助下,得以保证数据采集的完整性,提出了一种融合卷积神经网络(Convolutional Neural Networks,CNN)以及长短期记忆网络(Long Short-Term Memory,LSTM)负荷预测模型。通过训练值守作业机器人采集的历史数据,构建完整网络模型,该网络模型通过CNN网络对输入的数据提取有效序列特征,经过LSTM网络得到负荷预测结果。预测结果与多种算法进行比较,实验表明融合的网络模型具有更高的预测精度。The accurate prediction of power load plays an important role in the optimal scheduling of resources.However,the lack of power data in the traditional automatic acquisition system inevitably affects the subsequent data analysis.With the help of on-site real-time monitoring,the integrity of data collection can be ensured.In this paper,a load forecasting model based on the fusion of convolutional neural networks(CNN)and long short term memory(LSTM)is proposed.Through training withthe historical data collected by the robot on duty,a complete network model is constructed.The network model extracts the effective sequence features of the input data through the CNN network,and gets the load prediction results through the LSTM network.The experimental results show that the fusion network model has higher prediction accuracy.

关 键 词：值守作业机器人 负荷预测 卷积神经网络 长短期记忆网络 

分 类 号：TP249[自动化与计算机技术—检测技术与自动化装置]