The invention discloses an abnormal/fault positioning detection method based on a binomial tree model. The method comprises the steps of firstly, converting the physical layout of an industrial pipe network into the binomial tree structural model, and storing the detected data into corresponding nodes of a binomial tree; secondly, according to the character of the binomial tree and the relevance between information of the detection nodes, and utilizing the flow conservation property of the data flow of the industrial pipe network for achieving computation of the data flow of nodes of sensors not to be deployed; thirdly, according to the relation between the actual measuring value of the deployed sensors of the detection nodes of the industrial pipe network and the computation value, analyzing the operation condition of the industry, and judging the abnormal/fault types; finally, utilizing the structure characteristics of the binomial tree model for analyzing the abnormal/fault positions of the industrial pipe network and giving an alarm. According to the method, the design is simple, the cost is low, abnormal/fault detection by a traditional artificial tour gauging method is broken through, and the real-time detection on the working condition of the industrial pipe network in a real time mode under the limiting situations of the detection nodes is achieved.