基于最小二乘支持向量机的时用水量预测模型

Hourly water demand forecast model based on least squares support vector machine

针对神经网络存在结构较难确定,训练易陷入局部最优以及容易过学习等问题,提出将最小二乘支持向量机用于预测时用水量.最小二乘支持向量机(LSSVM)基于结构风险最小化,能够较好地协调经验风险最小化和学习机器VC维之间的关系,并且LSSVM在支持向量机(SVM)的基础上,通过将价值函数改为最小二乘价值函数以及用等式约束代替不等式约束,将求解的二次规划问题转变为一组等式方程,采用径向基核函数,得到LSSVM模型的待定参数比标准支持向量机少,仅为2个.根据时用水序列具有周期性和趋势性的特点,建立了基于最小二乘支持向量机的时用水量模型.实例分析表明,与基于BP网络的时用水量模型相比,基于最小二乘支持向量机的时用水量模型具有更强的预测能力.

As traditional neural network suffers from the problems like the existence of many local minima and the choice of the number of hidden units, and overfiting least squares support vector machine（LSSVM） is proposed to predict the hourly water demand. LSSVM employs the idea of structural risk minimization and VC dimension of the learning machine. In LSSVM, the SVM problem formulation is modified by introducing a least squares cost function and equality instead of inequality constraints, and the solution follows directly from solving a set of linear equations instead of quadratic programming. In the case of radial basis function（RBF） kernel, LSSVM has only two additional tuning parameters ,which is less than for standard SVM. According to the periodicity and trend of water demand series, an hourly water demand forecast model based on LSSVM is developed. Case study shows that LSSVM-based hourly water demand forecast model has better generalization ability than BP neural network-based forecast model.