Abstract: Based on the fact that the power plant ball mill load is hard to predict effectively, an improved support vector machine modeling method based on just-in-time learning is proposed form improving the online self-adaptive ability of the prediction model. Firstly, the instrumental variables are obtained by using grey relational analysis method to optimize the process parameters. Secondly, in the modeling framework of just-in-time learning, the parameters of the SVM prediction model are optimized by utilizing multi-population hybrid optimization algorithm. Finally, the simulation experiments are carried out based on the actual operation data. Simulation results show that compared with the standard BP neural network and the standard SVM prediction model, although the computing cost is increased, the proposed prediction model has better prediction performance and can satisfy the real-time requirements for the ball mill load in the coal pulverizing system.

Key words: ball mill load, online adaptive, just-in-time learning, improved support vector machine, multi-population hybrid optimization algorithm

摘要： 针对电厂球磨机负荷难以进行有效预测的问题，从提高预测模型在线自适应能力的角度出发，提出一种基于即时学习策略的改进SVM建模方法。利用灰色关联分析方法对过程参数进行优化筛选，获得辅助变量；在即时学习策略建模框架下，采用多种群混合优化算法进行SVM预测模型参数的优化选取；基于电厂实际运行数据进行了仿真研究。仿真实验表明，与标准BP神经网络和SVM建模方法的比较，该算法具有更好的预测性能，虽然计算开销有所增加，但能够满足制粉系统球磨机负荷检测的实时性要求。

关键词: 球磨机负荷, 在线自适应, 即时学习, 改进支持向量机, 多种群混合优化算法