Fault detection, identification and monitoring play a primary role in systems engineering. This also holds for advanced processes, such as robotic systems, with highest demands on reliability and safety. A method is presented for fast fault detection and location, due to the data deweighting in the identification procedure of the monitored process parameters and the iterative on-line calculation of the statistics of the fault detection scheme. This method has many advantages for microcomputer applications and guarantees a very early process fault detection. Application of the method to the fast fault detection of the de motor actuators of a robotic system is described.