摘要
结合BP神经网络和灰色理论两种单项预测模型算法,提出组合优化预测模型算法,实现对变压器油中溶解气体浓度更为精确的预测。该组合模型算法机理是根据预测误差平方和最小化的原则,首先计算各单项预测模型的权重,然后将各单项模型的权重进行加权综合计算,建立组合最优预测模型。以变压器中溶解的H_2为例验证了该组合算法汲取了两种单项算法的优点,不仅使各单项预测算法的预报误差降低,也有效提高了预测模型的预报性能。
Having forecasting model algorithm of BP neural network and grey theory based to propose a prediction model of combinatorial optimization was implemented to realize more accurate prediction of the dissolved gas concentration in transformer oil. The mechanism of this combined forecasting algorithm is to have the square sum of prediction errors minimized to calculate the weight of each prediction model,and then to carry out weighted calculation of the weight of all prediction models so as to establish an optimal combined forecasting model. Taking hydrogen in transformer as an example,the advantages of this optimal algorithm were verified to indicate that this optimal combined algorithm can reduce prediction error of each individual prediction algorithm and it effectively improves the forecasting performance.
作者
李忠明
LI Zhong-ming(Liaoning Petrochemical Vocational and Technical Colleg)
出处
《化工自动化及仪表》
CAS
2018年第8期607-610,639,共5页
Control and Instruments in Chemical Industry
关键词
组合预测模型
变压器油
溶解气
浓度
BP神经网络
灰色理论
预报误差
预报性能
combined prediction model
transtormer oil
dissolved gas
concentration
BP neural network,grey theory
prediction error
forecasting performance