Abstract The cold plate is a single-fluid heat exchanger, often used as the base of onboard electronic devices to cool it down through forced convection of fluid inside. It is not easy to monitor the status of the cold plate and device, due to the limitation of measurement and computing resource on the aircraft. This paper studies the dynamic characteristic of the cold plate by analyzing its typical structure, and a reduced model is derived and then transformed to state-space representation. Next, the extended Kalman filter is applied for online estimation of heat load and mass flow rate. Finally, different sample rates and measurements of detailed simulation data is fed to the algorithm for validation. Results show that monitoring temperature on the surface of device is enough to provide an accurate estimation of head load. Besides, it is suitable to set the sample time between 0.5s and 1.0s.