摘要
为了提高短期负荷预测的精度,提出基于量子差分进化算法(Quantum Differential Evolution,QDE)优化的最小二乘支持向量机(Least Squares-Support Vector Machine,LSSVM)模型。该算法克服了最小二乘支持向量机算法中依据经验选定参数的盲目性。实例验证结果表明,QDE-LSSVM的预测精度要远高于BP神经网络与单纯的最小二乘支持向量机,证明了利用量子差分进化选取最小二乘支持向量机的有效性。该算法更适用于当前中国短期负荷预测的需要。
To improve the accuracy of short-term load forecasting,this paper proposes least squares-support vector machine（ LSSVM） short-term load forecasting approach optimized by quantum differential evolution（ QDE） method. This paper overcomes blindness of least squares support vector machine algorithm that based on the experience of the selected parameters. Examples of verification results show that compared with BP neural network and simple least squares support vector machine algorithm,QDE-LSSVM has higher prediction accuracy which proves the efficiency of quantum-bit encoding in choosing the least squares support vector machine parameters. The proposed model is more suitable for the current needs of China＇ s short-term load.
出处
《电力科学与工程》
2015年第12期16-21,33,共7页
Electric Power Science and Engineering
关键词
短期负荷预测
参数优化
量子差分进化
最小二乘支持向量机
short-term load forecasting
parameter optimization
quantum differential evolution
least squares-support vector machine
