Abstract: Dual-pattern diagnosis requires the system to make correct decisions after the occurrence of faulty pattern and before the occurrence of critical pattern to avoid major losses, which is more practical than traditional fault diagnosis. A polynomial algorithm for verifying the dual-pattern diagnosis of discrete event systems under dynamic observation strategy is proposed to address the high cost and low efficiency of sensor setup in static observations. Firstly, a formal definition of dual-pattern diagnosis of discrete-event systems under dynamic observations is given, and pattern recognizers are constructed to label faulty and critical patterns, respectively. Next, a verifier automaton is constructed through automata composition to verify the dual-pattern diagnosability of the system, and then a sufficient and necessary condition for dual-pattern diagnosability of discrete event systems under dynamic observation is deduced based on this verifier. Conclusions and examples show that the polynomial algorithm proposed in this paper can not only effectively verify the dual-pattern diagnosability of discrete-event systems, but also optimize the efficiency of resource utilization while strengthening the diagnosability of the system by dynamically adjusting the trigger settings of the sensors, and provide an efficient and economical solution for dual-pattern diagnosis of complex systems.