摘要
传统概预算定额电力工程造价方法在没有工程细节时,存在估计偏差大且无法利用历史数据的问题。文中根据人工神经网络(ANN)的机器学习机制提出一种基于ANN的电力工程造价预测模型,将历史数据样本归一化作为输入,通过ANN算法对网络进行训练,采用训练后的网络来对工程造价进行估算。文中进行了多个电力工程造价的预测,得到的预测结果与实际造价相差小于5%,满足经验误差要求。
The traditional electric power engineering method for preliminary budget and ration has big estimation error in the absence of engineering details, and can＇t make use of the historical data, so an ANN-based cost forecasting model of elec- tric power engineering is proposed according to the machine learning mechanism of artificial neural network （ANN）. The histori- cal data samples are normalized, and taken as the input of the model. The network is trained with ANN algorithm to estimate the engineering cost. A number of electric power engineering costs are predicted in this paper. The difference of the predicted re- sult and actual cost is less than 5%, which can meet the requirement of empirical error.
出处
《现代电子技术》
北大核心
2017年第24期166-168,共3页
Modern Electronics Technique
基金
国家自然科学基金(61372071)
关键词
电力工程
预测模型
人工神经网络
机器学习
electric power engineering
prediction model
artificial neural network
machine learning
