论文部分内容阅读
利用满足Mercer条件的核函数改进线性岭回归算法,提出一种新的直接优化的非线性核岭回归算法。与SVM算法相比,该方法的特点是通过对二次损失函数的优化计算,简化了不确定参数的数量,而且由于采用直接优化计算,能有效减少建模的时间,提高计算效率。仿真实验结果和对航煤干点的软测量应用结果都验证了该算法的有效性。
By using the kernel function improved Mercer condition to improve the linear ridge regression algorithm, a new direct optimized nonlinear kernel ridge regression algorithm is proposed. Compared with the SVM algorithm, this method is characterized by simplifying the number of uncertain parameters by optimizing the quadratic loss function. Moreover, the direct optimization calculation can effectively reduce the modeling time and improve the computational efficiency. The simulation results and the results of the soft measurement of dry-jet pilot verified the effectiveness of the algorithm.