选用新型固化材料SH与石灰组合改良黄土,进行无侧限抗压强度正交试验,结果显示,SH与石灰的最优组合是掺量为10%的SH和4%石灰,风干时间28 d。对正交试验结果进行极差和方差分析,通过BP神经网络模型进行强度预测。预测结果与试验结果基本吻合,SH与石灰组合改良的最大相对误差为3.16%,表明该神经网络模型具有较高预测精度。研究结论对进一步研究改良黄土的工程特性有一定的参考价值。 The results show that the optimal combination of SH and lime is SH and 4% lime with the dosage of 10% and the drying time is 28 d. Range and variance analysis of orthogonal test results, strength prediction by BP neural network model. The predicted results are in good agreement with the experimental results. The maximum relative error of SH and lime combination improvement is 3.16%, which shows that the neural network model has higher prediction accuracy. The conclusions of the study have some reference value for further studying the engineering characteristics of loess.