基于人群搜索算法优化参数的支持向量机短期电力负荷预测

Short-term power load forecasting of support vector machine based on parameters optimization of population search algorithm

支持向量机是一种新型机器学习算法,它基于结构风险最小化准则取得较小的实际风险,有效提高了泛化能力,具有理论严密、适应性强、全局优化等特点,在模式识别和回归问题等方面应用广泛。以某地区历史负荷数据为输入,通过人群搜索算法对支持向量的各项参数进行寻优计算,得到最优的参数取值,然后把最优参数代入到SVM预测模型中,得到人群搜索算法的支持向量机(SOA-SVM)模型,利用此模型对某地区未来24小时的负荷进行短期预测。通过算例验证,利用SOA-SVM预测的精度要比BP神经网络和PSO-SVM的精度要高,所以说明用此方法进行短期负荷预测是有效和可行的。

Support Vector Machine（ SVM） is a new kind of machine learning algorithm,which is based on structural risk minimization criterion to obtain smaller actual risk,and effectively improves the generalization ability,as well as has the characteristics of the strict theory,strong adaptability,and global optimization,and widely used in pattern recognition and regression problems. The paper is based on the historical load data as input,and searches algorithm on the parameters of support vector optimization through the seeker optimization algorithm（ SOA） to get the optimal parameter selection,and then puts the optimal parameter generation into the SVM prediction model to get the seeker optimization algorithm of support vector machine（ SOA-SVM） model. By using this model,the short-term load prediction of the next 24 hours in one region is achieved. Through the algorithm example,the SOA- SVM prediction precision accuracy is higher than BP neural network and PSO-SVM,so using this method to conduct short-term load forecasting is feasible and effective.