Abstract: Natural gas pipeline network system has a large and complex structure, with faults occurring frequently during the operation. However, existing monitoring and control system of pipelines is equipped with a single form of alarm, unable to give timely warning in the potential occurrence period of system faults. Besides, it is not fully informationized and intellectualized. Therefore, a state recognition model of pipeline production data based on the control chart theory and random forest algorithm was established. Specifically, a control chart model was built by combining the operation data of natural gas pipeline and the control chart theory based on the 6 types of model on pipeline fault parameters. Then, the high-precision intelligent recognition for different data models was realized with the random forest algorithm. The results of application on site show that the operation data status recognition model based on the control chart theory and random forest algorithm has high accuracy and short time consuming, capable of recognizing the real-time status of operation data and thus providing accurate warning. Generally, the new model is applicable to the single variable fault warning of natural gas pipeline during operation and could provide technical support to the safe operation of pipelines.