摘要
针对现有电力数据利用不全面、负荷预测精度不理想等问题,文中提出了一种结合遗传算法(GA)和长短期记忆网络(LSTM)的电力数据分析方法。采用GA算法优化LSTM网络、选择最佳窗口大小及神经元数目等结构因素。利用优化后的LSTM网络进行电力数据分析,实现负荷可靠预测。基于Python仿真平台,以平均绝对误差(MAE)和均方根误差(RMSE)为评价指标对所提方法进行实验论证。结果表明,所提方法能够实现3类用户的分类并准确预测电力负荷,其MAE与RMSE值分别为88.32和120.01,均优于其他对比方法。
Aiming at the problems of incomplete utilization of existing power data and unsatisfactory load forecasting accuracy,a power data analysis method combining Genetic Algorithm(GA)and Long ShortTerm Memory network(LSTM)is proposed.GA algorithm is used to optimize LSTM network,and the optimal window size,number of neurons and other structural factors are selected.The optimized LSTM network is used for power data analysis to achieve reliable load forecasting.Based on the python simulation platform,the Mean Absolute Error(MAE)and Root Mean Square Error(RMSE)are used as evaluation indexes to demonstrate the proposed method.The results show that the proposed method can realize the classification of three types of users and accurately predict the power load,and its MAE and RMSE values are 88.32 and 120.01 respectively,which are better than other comparison methods.
作者
赵越
ZHAO Yue(China Eletric Power Research Institute Co,Ltd,Beijing 100192,China)
出处
《电子设计工程》
2021年第12期161-165,共5页
Electronic Design Engineering
基金
国家电网公司总部科技项目(5400-202018421A-0-0-00)。
关键词
电力数据分析
负荷预测
遗传算法
长短期记忆网络
平均绝对误差
均方根误差
power data analysis
load forecasting
genetic algorithm
long short-term memory network
mean absolute error
root mean square error
