Abstract: To achieve automatic and high-precision real-time monitoring of ice thickness during the ice flood period of the Yellow River, a resistive-capacitive integrated induction type ice thickness monitoring system integrating resistance, capacitance and temperature sensing was designed. This system utilizes the differences in resistivity, dielectric constant and temperature characteristics of air, ice and water to simultaneously collect multimodal signals through a 140 high-density electrode array; it uses low-power MCU and high-precision ADC for signal processing and data fusion; it introduces a multi-sensor data fusion algorithm based on deep learning to build an ice thickness inversion model; and it realizes remote real-time transmission and intelligent analysis of monitoring data through 5G communication module. Laboratory simulation of the ice flood of the Yellow River shows that this system can achieve automatic continuous monitoring of ice thickness, and the average absolute error of the measurement results with the manual measured values is less than 1.5 cm, and the maximum error is no more than 2 cm.