以某300MW机组为例,使用支持向量机理论和遗传算法,利用电站分散控制系统(DCS)获取的热工数据,建立预测锅炉低温过热器受热面清洁状态下的吸热量(清洁吸热量)模型,进而推算出表征积灰污染程度的清洁系数,得到时序变化的积灰曲线,从而对吹灰状况进行监测。其结果表明,以清洁系数表征锅炉低温过热器受热面积灰情况具有可操作性;所建立模型的测试集决定系数、均方根误差和平均绝对误差分别为0.920、36.381和25.457,表明该模型具有良好的预测性能,可以在一定范围内对吹灰信号产生响应;锅炉实际运行工况下应用所建模型表明,给煤量大幅度变化会对受热面的积灰曲线造成一定程度的干扰。该研究为开发大型电站锅炉积灰结渣在线监测系统提供参考和依据。 Taking a 300 MW unit as an example, using the support vector machine theory and genetic algorithm, the thermodynamic data obtained from the distributed control system (DCS) of the power plant is used to predict the endothermic amount (clean endothermic amount) of the low- ) Model, and then calculate the cleaning coefficient that characterizes the fouling degree of fouling, and obtain the fouling curve with time series, so as to monitor the condition of sootblowing. The results show that it is feasible to use the cleaning coefficient to characterize the heating area of ​​boiler low temperature superheater. The determination coefficient, root mean square error and average absolute error of the test set are 0.920, 36.381 and 25.457, respectively, indicating that the model has Good predictive performance can respond to the soot blowing signal within a certain range. The application of the boiler model under actual operating conditions shows that a large change in coal feed quantity will cause a certain degree of disturbance to the ash deposition curve of the heated surface. The research provides a reference and basis for the development of on-line monitoring system of fouling and slagging in large power station boilers.