摘要
针对数据挖掘分类和响应特征分析存在的难度。提出利用支持向量机算法改进时间序列算法,以提高数据自动化挖掘能力,利用分段聚合近似方法用于降低预处理电力系统负载数据的维数。实验结果表明,当数据样本数量为5000时,标准偏差(σ)内0.25准确性达到最高0.98;测试集数据组的平均数据挖掘误差为7.2%,训练集数据组的平均数据挖掘误差分别为13.81%和13.55%。当迭代次数为10次时,改进时间序列算法精度为0.68,较深度学习算法与神经网络算法分别提高17.8%,8.46%。在迭代次数为100次时,深度学习算法与改进时间序列算法的挖掘精度均为1.0,神经网络算法精度为0.96;改进时间序列算法具有较高的数据挖掘精度。
In view of the difficulty of data mining classification and response feature analysis,the support vector machine algorithm was proposed to improve the time series algorithm to optimize the ability of automatic data mining,and the segmented aggregation approximation method was used to reduce the dimensionality of preprocessing power system load data.The experimental results indicated that when the number of data samples was 5000,σ=0.25 accuracy reached a maximum of 0.98.The average data mining error of the test set dataset was 7.2%,while the average data mining error of the training set dataset was 13.81%and 13.55%,respectively.When the number of iterations was 100,the mining accuracy of the deep learning algorithm and the improved time series algorithm was 1,while the accuracy of the neural network algorithm was 0.96,and the improved time series algorithm has a high data mining accuracy.
作者
房娟
彭嘉宁
FANG Juan;PENG Jianing(State Grid Ningxia Electric Power Co.,LTD.,Yinchuan 640001,China)
出处
《粘接》
CAS
2024年第7期193-196,共4页
Adhesion
关键词
时间序列
支持向量机
电力系统
准确性
精度
time series
support vector machine
power system
accuracy
accuracy