大规模电力工程数据价值深度挖掘算法设计研究

Research on design of deep mining algorithm for large⁃scale power engineering data value

针对传统电力工程数据处理方法中存在的不可追溯且不易统一管理等问题,文中基于数据挖掘的思想提出了一种电力工程数据价值分析预测模型。该模型采用Boosting算法将多个预测树结构组合形成极端梯度提升树模型,从而实现对非线性数据的深入分析,且经过多次迭代后,可以使训练准确度与学习效果得到显著提升。通过采用改进的双向长短时记忆网络,增强了模型处理时序性数据的能力。同时还使用误差倒数法将两个算法模型相结合,使其具有更高的预测精度。在实验测试中,所提算法的预测结果更贴近实际值,且其MAPE及RMSE测试指标分别为0.201%和0.039%,在所有对比算法中均为最优,可以对电力工程数据价值进行准确的分析和预测。

In view of the shortcomings of traditional power engineering data processing methods,such as non traceability and difficult to unified management,this paper proposes a value analysis and prediction model of power engineering data based on the idea of data mining.The model uses Boosting algorithm to combine multiple prediction tree structures to form an extreme gradient lifting tree model,so as to realize in⁃depth analysis of nonlinear data.After several iterations,the training accuracy and learning effect can be significantly improved.By using bidirectional Long Short⁃Term Memory network,the ability of the model to process time⁃series data is enhanced.At the same time,the two algorithm models are combined by using the error reciprocal method,so that it has higher prediction accuracy.In the experimental test,the prediction result of this algorithm is closer to the actual value,and its MAPE and RMSE test indexes are 0.201%and 0.039%,which are the best among all the comparison algorithms,it can accurately analyze and predict the value of power engineering data.