基于核偏最小二乘的电厂热力参数预测与估计

Parameter Prediction and Estimation of Turbine Generator Based on Kernel Partial Least Squares

为了解决机组运行过程中参数失效和优化过程中参数计算的问题，提出了一种基于核偏最小二乘方法的热力参数预测和估计方法。首先用正常数据建立机组参数的预测和估计模型，确定各变量之间的回归关系，然后将其用于参数的在线预测与估计。其基本思想是通过非线性核函数将数据映射到高维特征空间，然后在高维特征空间中进行偏最小二乘回归运算。该方法可以有效地捕捉变量间的非线性关系，参数预测和估计效果明显好于偏最小二乘法和主元回归方法等线性回归方法。某1000Mw发电机组烟气含氧量历史特征数据集仿真试验及实际应用比对实验证明了该方法的有效性。

In order to solve the problem of the failure of measure parameters and online optimal running in generator units, a novel parameter prediction and estimation method based on kernel partial least squares （KPLS） was proposed. The prediction and estimation mathematic model was established firstly and online estimation of parameters was performed. The basic idea was first mapped data from original space into feature space and then performs PLS regression in the feature space. The proposed method can effectively capture the nonlinear relationship among process variables and has better estimation performance than PLS and other linear approaches. Simulation results of some 1 000 MW turbine generator＇s data prove that the method is effective.