Reactor is of great significance for ensuring the safe and stable operation of power system. Aiming at the problem of low efficiency in single state monitoring of dry-type reactors, leading to the inability to provide timely feedback on operational faults, this paper studies and designs a multi-state monitoring system for dry-type reactors based on temperature and power loss monitoring. Firstly, the monitoring system structure based on interaction between edge terminal and cloud platform is designed. Then, the measurement and calculation principles of temperature and power loss of dry-type reactor are discussed and a comprehensive state assessment method based on random forest is designed for fault identification. The experiment shows that the proposed method has higher accuracy in multi-state monitoring of dry-type reactors, which can be effectively used for monitoring and evaluating operational faults.