Water is one of the most effective agents that weaken the physio-mechanical properties of rock, trigger significant construction delays, endanger the construction operation, and lead to rock failure. Therefore, an early failure point (EFP) prediction of rock under such conditions is imperative for the robust and reliable implementation of underground engineering. In this research, an EFP of sandstone with different water contents was predicted based on infrared radiation and complex energy evolution during loading. The ratios of elastic to dissipation energy (K_ED) and elastic to total energy (K_ET) were proposed to predict EFP. The results show that K_ED and K_ET give EFP at the same time for sandstone with water contents 0%, 0.991%, 2.136%, and 3.109%, and the average time of EFP ahead 208 s, 250 s, 265.8 s, and 276.9 s than rock failure, respectively. Furthermore, the proposed K_ED and K_ET were predicted using an artificial neural network (ANN). The ANN models’ efficacy was evaluated using the performance coefficient (R²) and root-means-square error (RMSE). The findings revealed high R² and low RMSE for K_ED and K_ET of sandstone with different water contents. The research findings can be used effectively to monitor disasters for the safe and efficient execution of engineering projects.

水是削弱岩石物理力学特性、引发重大施工延误、危及施工作业并导致岩石破坏的最有效介质之一。因此，在这种条件下对岩石进行早期破坏点 (EFP) 预测对于地下工程的稳健和可靠实施至关重要。在这项研究中，基于红外辐射和加载过程中复杂的能量演化，预测了不同含水量砂岩的 EFP。提出了弹性与耗散能 (K _ED ) 和弹性与总能量 (K _ET )的比率来预测 EFP。结果表明，K _ED和 K _ET对含水量为0%、0.991%、2.136%和3.109%的砂岩同时给出EFP，EFP的平均时间分别比岩石破坏提前208 s、250 s、265.8 s和276.9 s。此外，提出的 K _ED和 K _ET是使用人工神经网络 (ANN) 进行预测的。使用性能系数 (R ² ) 和均方根误差 (RMSE)评估 ANN 模型的功效。结果表明，不同含水量砂岩的K _ED和K _ET具有高R ²和低RMSE 。研究成果可以有效地用于监测灾害，以安全高效地执行工程项目。