基于梯度提升决策树的锅炉燃烧优化控制策略研究

Research on optimal control strategy of boiler combustion based on gradient boosting decision tree

开展锅炉燃烧的优化控制策略研究对于提高资源利用率具有重要意义。文中基于梯度提升决策树算法,结合锅炉燃烧模型设计了一套优化控制策略。该策略在反应粒子模型的基础上,建立了反应器和粒子的质量与能量平衡方程,并生成了反应过程的完整流程模型,从而解决了非线性强耦合的锅炉建模问题。同时通过引入梯度提升决策树,设计实现了基于LightGBM的锅炉优化控制策略,从而提高了锅炉的燃烧效率。实验结果证明了将所提策略部署于模型中的优越性,且相比于无优化策略的情况,所提方法可将燃烧效率提高26%。

The research on optimal control strategy of boiler combustion is of great significance to improve resource utilization.Based on the gradient lifting decision tree algorithm and the boiler combustion model,a set of optimal control strategy is designed in this paper.Based on the reaction particle model,the mass and energy balance equations between the reactor and particles are established,and the complete process model of the reaction process is generated,which solves the problem of nonlinear and strong coupling boiler modeling.By introducing the gradient lifting decision tree,the boiler optimal control strategy based on LightGBM is designed and realized,so as to improve the combustion efficiency of the boiler.The experimental results verify the superiority of the proposed strategy after deployment in the model.Compared with the case without optimization strategy,the combustion efficiency can be improved by 26%.