摘要
针对电力输电工程情况复杂多变、影响因素多、经济技术指标难以准确预测的问题,提出了一种基于多因素特征分析的输电工程静态投资预测模型。采用Pearson相关系数分析影响静态投资的主要因素并进行标准化处理,基于梯度提升迭代决策树(GBDT),以每轮实际值和预测值的残差最小化为目标不断训练和优化预测模型。通过对实际输电工程数据进行建模仿真的测试结果表明,该模型能够实现静态投资的合理预测,绝对平均误差率(MAPE)为4.39%,验证了所提模型的普适性及准确性。
Aiming at the problems of complex and changeable power transmission projects,many influencing factors,and difficult to accurately predict economic and technical indicators,this paper proposes a static investment prediction model for power transmission projects based on multi⁃factor characteristic analysis.The Pearson correlation coefficient is used to analyze the main factors affecting static investment and carry out standardized processing.Based on the Gradient Boosting iterative Decision Tree(GBDT),the prediction model is continuously trained and optimized with the goal of minimizing the residual between the actual value and the predicted value in each round.The test results of modeling and simulation on actual transmission engineering data show that the model can realize a reasonable prediction of static investment.The Mean Absolute Percentage Error rate(MAPE)is 4.39%,which verifies the universality and accuracy of the proposed model.
作者
刘伟
LIU Wei(Xinjiang Tianfu Group Co.,Ltd.,Shihezi 832000,China)
出处
《电子设计工程》
2021年第19期79-83,共5页
Electronic Design Engineering
基金
国家电网科技合作项目(TFHT20201479)。
关键词
输电工程
多因素特征
静态投资预测
GBDT
transmission engineering
multi factor characteristics
static investment forecast
GBDT
