基于分布式证据学习算法的脱硫浆液品质监测模型研究

An evidential condition monitoring model for desulfurization slurry

基于分布式计算框架与证据学习算法,对脱硫浆液品质建立了健康品质监测模型,突破了海量脱硫系统运行数据对基于传统机器学习的浆液品质监测方法所带来的计算瓶颈,并利用该模型对江苏某1000 MW电厂的浆液品质进行了在线监测。测试表明,所建立的监测模型能够准确监测出脱硫浆液品质的恶化,与其他3类先进监测方法对比结果说明了所建立模型能够达到最优的报警及时性。将分布式计算框架结合证据理论应用于脱硫浆液品质监测是可行的,为脱硫浆液品质监测提供了一种新方法。

Based on distributed computing framework and evidence learning algorithm,a robust condition monitoring model is established for desulfurization slurry,which overcomes the computational bottlenecks brought by massive operational data of desulfurization systems to traditional machine learning-based slurry condition monitoring methods.This model is utilized to perform online monitoring of slurry condition in a 1000 MW power plant in Jiangsu,China.Test results indicate that the monitoring model established is able to detect the deterioration of desulfurization slurry condition accurately.Through comparing with three other advanced monitoring methods,it is demonstrated that the model established can achieve the optimal alarm timeliness.It is feasible to apply the integration between distributed computing framework and evidence theory in desulfurization slurry condition monitoring,providing a new approach for similar monitoring.