This paper proposes a distributed methodology for the detection of actuator faults in a class of linear multi-agent systems in the presence of disturbances. A cascade of fixed-time observers is introduced to give an exact estimate of the global system state for each agent whereby the convergence time is estimated regardless of the initial conditions. Distributed observers are then designed using Linear Matrix Inequalities (LMI), by employing the ℌ and ℌ∞ norms. The proposed method ensures global actuator fault detection, where each agent is capable of detecting not only its faults but also those that occur at any other part of the system. Numerical simulation results are carried out to show the effectiveness of the proposed approach.