An orthometric hyperparallel spatial directional expansion filtering based fault diagnosis method is proposed for discrete systems with unknow but bounded noises. Firstly, the orthometric hyperparallel space is defined by the extremum of the parallelotope at the adjacent time based on the spatial structure of traditional parallelotope, to wrap the upper and lower bounds of the feasible parameter set and guarante the monotonic convergence of the parameter boundary value. Then, whether the system is faulty is checked by detecting the intersection of the parallelotopic space and the strip space, and then the orthometric hyperparallel space is extended directionally at the time of fault occurrence, and the fault isolation is realized according to the intersection of the extended orthometric hyperparallel space and the strip space. Finally, the fault identification is completed by the shrinkage property of the orthometric hyperparallel space in the iterative process. The simulations are given to verify the feasibility and effectiveness of the proposed algorithm.