<p>Piping is an increasingly critical failure mechanism in dike systems, posing significant risks to dike infrastructure safety. Therefore, examining the evolutionary characteristics of piping is critically important. Most existing methods have mainly focused on analyzing a single physical field signal, making it difficult to track the piping process accurately. This study employed physical model experiments to simulate the piping process in a dual-layer dike, focusing on the dynamic response of multi-physical field signals at different stages. The process was visually observed and divided into five stages: initial stability, clay layer rupture, seepage, piping initiation, and piping occurrence. Throughout the experiment, pore pressure, temperature, and electrical signals were continuously monitored using sensors. The results reveal distinct response patterns for each signal. The pore pressure signal initially remains stable, then suddenly decreases, followed by a period of stability, and finally abrupt variation, with simultaneous responses observed across multiple measurement points. The temperature signal initially remains stable, then increases, followed by a gradual decrease at each measurement point, with no response to the migration of sand particles. The electrical signal initially remains stable, then abruptly drops, sharply increases at each measurement point, and eventually undergoes a dramatic change, with more pronounced responses observed at measurement points near the pipe tip. These results contribute to a deeper understanding of the piping process and offer theoretical support for identifying its evolutionary stages in dike systems.</p>