应用序列最小优化算法的火电厂协调系统的预测

针对支持向量机二次规划(QP)算法处理大规模数据时计算复杂度高的问题,介绍了适宜处理大规模数据回归问题的序列最小优化(SMO)算法,并在该算法的基础上进行了改进,使运算速度得到进一步的提高。同时,将SMO算法及其改进算法(I-SMO)用于火电厂协调系统的预测,并同QP算法进行了比较。仿真结果表明,I-SMO算法比QP算法具有更高的预测精度和更快的运算速度,并且比SMO算法在计算速度方面又有较大的提高。图6表2参9

Prediction of Coordination Systems in Fossil Fired Power Plants by Using Sequential Minimal Optimization

With an eye on the problem of highly complicated calculations,required for treating the large amount of data with Support Vector Machine's Quadratic Programming(QP),the sequential minimal optimization(SMO) algorithm,which is suitable for dealing with the large amounts of data by regression,is being introduced with improvements for further increasing speed of calculation.SMO algorithm and its improved version I-SMO is now being used for prediction in power plant coordination systems,and simultaneously compared with the QP method.Simulation results show that the prediction accuracy and calculating speed of I-SMO is superior to that of QP algorithm,thus representing also a progress of SMO in respect to speed of calculation.Figs 6,tables 2 and refs 9.