The invention belongs to the technical field of drilling well fault recognition, and discloses a drilling well fault recognition system. The system includes a symptom characteristic vector construction module, a fuzzy relation matrix establishment module, and a prediction gas drilling failure feature recognition module; the symptom characteristic vector construction module is used for constructinga symptom characteristic vector X=([mu]x1,[mu]x2, ...[mu]x11, [mu]x12); the fuzzy relation matrix establishment module is used for establishing a fuzzy relation matrix R; according to a fuzzy relation Y=X*R, Y is obtained by solving; a fault feature vector Y=([mu]y1, [mu]y2, ...[mu]y6, [mu]y7) is obtained, wherein an accident type corresponding to a largest component of the Y is a recognition result; [mu]xi is a degree of membership of an object with a symptom xi; and the prediction gas drilling failure feature recognition module is used for recognizing a prediction gas drilling failure feature in real time. The system studies a method of applying a modern fault recognition technology and a mathematical method to the drilling well fault recognition by analyzing parameter variation characteristics of drilling well faults, proposes a feasible scheme, is very necessary for improving the accuracy of the drilling well fault recognition, and enables loss brought by a complicated situation to be reduced to a lowest limit.