本發(fā)明涉及地質(zhì)災(zāi)害預(yù)測技術(shù)領(lǐng)域，具體涉及一種基于機器學(xué)習(xí)獲得地質(zhì)災(zāi)害預(yù)警關(guān)鍵參數(shù)預(yù)測值的方法，包括：S1、梳理地質(zhì)災(zāi)害預(yù)警系統(tǒng)中目標地質(zhì)災(zāi)害隱患點的監(jiān)測參數(shù)；S2、獲取目標地質(zhì)災(zāi)害隱患點監(jiān)測參數(shù)的歷史數(shù)據(jù)；S3、把所有監(jiān)測參數(shù)分成第一類監(jiān)測參數(shù)與第二類監(jiān)測參數(shù)；S4、基于目標地質(zhì)災(zāi)害隱患點監(jiān)測參數(shù)的歷史數(shù)據(jù)創(chuàng)建用于機器學(xué)習(xí)的數(shù)據(jù)集，將第一類監(jiān)測參數(shù)的歷史數(shù)據(jù)或?qū)崟r數(shù)據(jù)作為輸入，將第二類監(jiān)測參數(shù)的實時數(shù)據(jù)作為輸出，通過機器學(xué)習(xí)算法進行訓(xùn)練和預(yù)測，得到地質(zhì)災(zāi)害預(yù)警關(guān)鍵參數(shù)預(yù)測值。本發(fā)明解決了難以降低地質(zhì)災(zāi)害預(yù)警系統(tǒng)的故障率的技術(shù)問題。