Soil salinization and freshwater scarcity are widely recognized as persistent constraints on sustainable agricultural intensification, especially in arid and semi-arid regions. This study presents a targeted approach by developing an integrated intelligent irrigation-drainage platform that combines Internet of Things (IoT)-based multi-layer environmental sensing, Long Short-Term Memory (LSTM) deep learning for predictive modeling, and digital twin-assisted management. A closed-loop irrigation-drainage infrastructure was concurrently implemented, incorporating multi-source water reuse pipelines, subsurface drainage networks, and photovoltaicpowered units for brackish water purification. Field deployment in saline-affected agricultural plots in northwestern China enabled high-frequency monitoring of soil moisture, temperature, and salinity dynamics under complex soil stratification conditions. The Long Short-Term Memory (LSTM) model demonstrated high predictive accuracy (R2 = 0.97 for soil electrical conductivity and R2 = 0.92 for water content), while feature importance analysis identified soil moisture, temperature, and groundwater depth were the dominant factors driving water-salt interactions. Coordinated irrigation and drainage scheduling significantly reduced soil salinity across permeable soil profiles, while closed-loop water reuse strategies enhanced water-use efficiency without increasing salinity risks. This study highlights the potential of intelligent irrigation-drainage systems integrating real-time monitoring, predictive modeling, and closed-loop water reuse to enhance salt leaching efficiency, optimize water utilization, and stabilize crop production across heterogeneous saline agroecosystems.