传统的支持向量回归机(Support Vector Regression,SVR)都没有考虑到样本点处的梯度信息,因此必须通过加大训练样本的数目来提高回归精度,但由此导致的开销在计算机试验中有时是非常巨大的,从而影响了其在计算机试验中的广泛应用。针对传统SVR的不足,提出基于梯度信息的支持向量回归机(Gradient-enhanced SVR,GE-SVR)。提出的模型能在较少样本的情况下达到较为理想的回归精度,因而具有一定的理论意义和实用价值。 Traditional Support Vector Regression (SVR) does not consider the gradient information at the sample point. Therefore, the regression accuracy must be improved by increasing the number of training samples, but the resulting overhead is sometimes in computer experiments Very huge, thus affecting its extensive use in computer experiments. Aiming at the deficiency of traditional SVR, a Gradient-enhanced SVR (GE-SVR) is proposed based on gradient information. The proposed model can achieve better regression accuracy with fewer samples and thus has some theoretical and practical values.