The problem of distributed attack estimator design is studied for heterogeneous interconnected cyber-physical systems with false data injection (FDI) attacks and sensor faults. Firstly, the system state and sensor fault are augmented into an equivalent generalized state space model, and the distributed attack estimation observer is designed by using the model and the correlation information between subsystems. In the design of the distributed observer, the intermediate variable and the output estimation error feedback term are introduced so that the observer has a flexible parameter matrix structure, which is suitable for heterogeneous systems composed of subsystems with different state dimensions and realizes the simultaneous estimation of attack signals and sensor faults. Secondly, the stability of the dynamic error system is analyzed, and the H∞ performance is employed to suppress the influence of attack signals and external disturbances on the estimation effect. Meanwhile, the solution method of the observer gain matrix is obtained as a linear matrix inequality. Finally, the feasibility of the proposed attack estimation method is verified by numerical simulation and comparative simulation.