三角旋回优化支持向量机预测出清价及价格钉  

TGA-SVM to the Forecast of Clearing Price and Price Spike in Power Market

作　　者：左幸[1] 马光文[2] 

机构地区：[1]二滩水电开发有限责任公司,四川成都610051 [2]四川大学能源发展研究中心,四川成都610065

出　　处：《水力发电》2009年第7期5-8,13,共5页Water Power

基　　金：国家自然科学基金资助项目(50539140)

摘　　要：针对传统支持向量机参数较难选择的问题,提出一种新的全局优化方法——三角旋回算法(Triangle Gyration Algorithm,TGA)来优化支持向量机的参数,并且建立了三角旋回支持向量机数学模型。三角旋回算法具有结构简单、鲁棒性强和快速收敛的特点。算法的寻优过程采用历史最优目标函数值进行指导,利用三角变换进行迭代使其能够快速收敛到全局最优。将其应用于电力市场出清价及价格钉的预测实例研究,与传统的支持向量机预测结果比较,三角旋回支持向量机具有更高的预测精度。Due to the difficulty of selecting the parameters of support vector machine （SVM）, this paper presents a new global optimization method, Triangle Gyration Algorithm （TGA）, to optimize the parameters of SVM and establishes TGA- SVM model. TGA embodies the characteristics of simple structure, good robust and fast convergence. The individual of TGA in every generation is guided by the best objective function value and transformed by the trigonometric function, which ensure the whole progress could quickly converge at the global optimal points. Comparing to the results from traditional support vector machine for forecasting market clearing price and price spike in power market, the TGA-SVM manifests the more accurate forecasting results.

关 键 词：市场出清价 价格钉 预测 三角旋回算法 支持向量机 

分 类 号：F713.53[经济管理—市场营销] F407.2[经济管理—产业经济]