提出了一种基于最小二乘支持向量机理论及可视化火焰检测系统的燃煤电站锅炉负荷预测的方法。利用可视化火焰检测系统对300MW燃煤锅炉降负荷过程的燃烧火焰及温度场进行了实时测试。通过测试,利用数字图像处理技术提取了燃烧火焰图像特征参数、由双色法测温原理计算得到了其温度场,采用最小二乘支持向量机建立了锅炉负荷预测模型并进行了校验。结果表明,该预测模型泛化能力强、预测精度高,从而为把燃烧火焰图像及温度场信息作为控制信号,进而引入电站锅炉燃烧控制系统提供了技术支持。
A model to predict power boiler's load was presented by means of Least-square bVM (LS-SVM) and measurement system of flame visualization. The combustion flame of a 300MW boiler and its temperature distribution with its load changed was tested by measurement system of flame visualization. The characteristic parameters were obtained using technology of digital image. The temperature distribution was computed by means of two-colour method. A predicted model of power load was established by LS-SVM. The results indicated the predicted model is more general and accurate. Thus, the predicted model is the powerful tool to put flame image and temperature information in the control system of power boiler combustion.