基于特征向量提取的核回归建模方法研究

Kernel Regression Modeling Method Based on Feature Vector Selection

针对工业软测量中的非线性数据回归问题,提出一种基于特征向量提取的核回归建模方法。基于核函数非线性变换技术,建立非线性软测量模型—核回归模型。为了减少核回归模型中的优化参数,采用特征向量提取(FVS)算法选择核回归模型的特征向量,最后采用改进的粒子群优化算法估计模型参数。在工业数据上的应用结果说明了方法的有效性。

To the problem of the nonlinear data regression for industrial soft sensor,a kernel regression modeling method based on feature vector selection is presented.From the nonlinear transformation based on kernel function,a nonlinear kernel regression soft sensor model is built.In order to reduce the number of optimization parameters in the kernel regression model,the feature vector selection is applied to select the feature vectors for kernel regression model.The particle swarm optimization algorithm is improve...