基于改进LS-SVM的球磨机制粉出力软测量建模

Soft sensor modeling for mill output of ball pulverizer based on an improved least squares support vector machines

针对火电厂双进双出钢球磨直吹式制粉系统出力较难直接测量的问题,本文在Suykens的最小二乘支持向量机(least squares support vector machine,LS-SVM)稀疏化算法的基础上,提出了一种更好的改进方式,并将改进后的LS-SVM算法对双进双出钢球磨直吹式制粉系统出力建立软测量模型,通过对算法改进前后模型仿真结果的对比分析可知,改进后的LS-SVM算法学习收敛速度更快,误差更小,更加适用于在线学习,并为制粉系统的在线优化控制打下了良好的基础。

According to the problem that mill output is difficult to directly measured, a new improved algorithm was proposed based on the Suykens＇ sparseness method of least squares support vector machines （LS-SVM）. The improved LS-SVM was used to estab- lish the soft sensor model of mill output in direct-fired system with duplex inlet and outlet ball pulverizer. The two models of before and after LS-SVM algorithm improvement was compared with each other. Simulation results show that the learning speed is faster and the error is less of after improvement than before. It is more suitable for study on-line. This will lay a good foundation for the optimal control online of pulverizing systems.