Monitoring, management, and operation to ensure the safety and efficiency of the water reservoir system is a very important issue for hydroelectric plants, which affects the economic efficiency, survival of hydropower plants, and social safety. In this research, sensor networks and machine learning were combined to monitor and predict problems related to dam systems and water levels of hydroelectric reservoirs. A monitoring interface is designed to display the main parameters of the reservoir, such as water level, temperature, and amplitude or size of expansion joints between concrete blocks of the dam body, which is useful in reservoir systems. The research combined the use of the data acquired at the reservoir field, which are closely related to the expansion gaps and the water levels at present or in history. The research was successful in proposing a monitoring interface and early prediction model of the expansion gap amplitude between concrete blocks of the dam and the water level of the reservoir with high accuracy and reliability.