论文部分内容阅读
为解决传统最小二乘支持向量机(LSSVM)采用交叉验证确定参数耗时较长和粒子群(Particle Swarm Optimization,PSO)优化算法早熟收敛的问题,提出一种基于种群活性PSO算法优化LSSVM参数的方法。利用群活性加速度作为多样性测度,当群活性加速下降时,对粒子的位置和速度分别执行进化和变异操作来改进标准PSO算法,然后分析上海市时用水量序列特点及其影响因素,选取影响程度较大的主要因素,将其作为预测模型的输入变量,建立时用水量预测模型;最后采用改进的PSO算法优化LSSVM参数来预测上海市时用水量。实例分析表明,对比文中其他3种模型输入变量组合,选取的预测模型输入变量能够更有效地提高预测精度;与传统LSSVM方法相比,提出的基于改进PSO-LSSVM的时用水量预测方法计算速度更快,预测精度更高。
In order to solve the problem that the traditional Least Squares Support Vector Machine (LSSVM) uses cross-validation to determine the parameter time-consuming and premature convergence of Particle Swarm Optimization (PSO) optimization algorithm, a PSO algorithm based on population active optimization of LSSVM parameters method. Using group activity acceleration as a measure of diversity, when the population activity declines rapidly, evolutionary and mutation operations are performed on the particle’s position and velocity, respectively, to improve the standard PSO algorithm. Then, the characteristics and influencing factors of water consumption in Shanghai are analyzed, Which is the main factor of larger degree, as the input variables of the forecasting model and the water consumption forecasting model at the time of establishment. Finally, the improved PSO algorithm is used to optimize the parameters of LSSVM to predict the water consumption in Shanghai. The case study shows that compared with the traditional LSSVM method, the proposed PSO-LSSVM-based water consumption prediction method can effectively improve the prediction accuracy compared with the other three input variables. Faster, with better forecasting accuracy.