基于SSA-SVM的边坡失稳智能预测及预警模型

Intelligent prediction and alert model of slope instability based on SSA-SVM

针对传统统计学习模型等方法对边坡失稳预测精度低、难度大等问题,在对国内外304个边坡案例中高度、角度、容重、黏聚力、内摩擦角、孔隙压力比和边坡状态等参数进行搜集统计的基础上,建立边坡预测数据库,采用麻雀搜索算法(SSA)优化支持向量机(SVM),构建SSA-SVM边坡失稳智能预测模型,实现对边坡失稳智能预测.采用灰狼优化算法、遗传算法、布谷鸟搜索算法、粒子群优化算法、哈里斯鹰优化算法和鲸鱼优化算法优化SVM,并与SSA-SVM模型进行对比,结果表明:SSA-SVM模型在边坡失稳预测中具有突出优势,其准确率、精确率、F1分数、平均精度分数和AUC值分别达到了90.16%,94.28%,91.43%,96.79%和0.954,高于其他优化模型的相应指标,SSA算法相比其他优化算法具有很强的竞争力.

Aiming at the problems of low accuracy and high difficulty in slope instability prediction using traditional statistical learning models and other methods,the predicted slope database was covered by 304 slope cases from domestic and international,including the parameters such as height,angle,bulk density,cohesion,internal friction angle,pore pressure ratio and slope status,and support vector machine(SVM)was optimized by sparrow search algorithm(SSA)to construct the SSA-SVM slope instability intelligent prediction model,which could intelligently predict slope instability.To compare with the SSA-SVM model,SVM was optimized respectively by gray wolf optimization algorithm,genetic algorithm,cuckoo search algorithm,particle swarm optimization algorithm,Harris hawk optimization algorithm and whale optimization algorithm.Results show that the SSA-SVM model has outstanding advantages in slope instability prediction,and its accuracy,precision,F1-score,average precision score and area under curve(AUC)reach at 90.16%,94.28%,91.43%,96.79%and 0.954,respectively,which are higher than the corresponding indicators of other optimization models.Compared with other optimization algorithms,the SSA algorithm has strong competitiveness.