An optimal event-triggered fault detection method based on diagnostic observer is proposed. Firstly, a low-order diagnostic observer parameter matrix calculation method combining a numerical approach and an algebraic approach is developed to satisfy Luenberger conditions in residual generation for the remote monitoring system which is triggered by adaptive mixed event-triggered mechanism. Then, the H∞/H∞ performance index is constructed by evaluating the influence of event-triggered transmission error, disturbance and fault on residual signal, and the optimal post-filter is designed by using the co-inner-outer factorization technology to achieve the optimal trade-off between fault sensitivity and disturbance robustness. Further, in order to reduce the conservatism of the traditional preset residual threshold method, according to the boundedness of disturbance and initial system status, the residual evaluation function and the corresponding fault decision logic are designed based on zonotope to realize fault detection. Finally, the effectiveness of the proposed method is verified by simulation of the multi-vehicle tracking system.