The real-time hydrological data such as water level and discharge of the main canal of the Middle Route of the South-to-North Water Transfers Project are affected by external disturbances, measurement system errors and other factors. The ill-conditioned hydrological data will cause the calculation distortion of the scheduling model, and even lead to the failure of the calculation.Aimed at the logical errors in the upstream and downstream flow data space and the jump of the time series of water level data, the water balance model based on particle swarm optimization and the exponential weighted moving average model were established respectively, and the pathological water regime data was cleaned horizontally and vertically in space and time. Taken the channel section between the Yellow River controlling gate and the Zhanghe River controlling gate as a typical research interval, the flow inversion point was automatically identified by the model. The flow data of 12 controlling gates and 26 water diversion points involved in the channel section were uniformly corrected to realize the rationality of upstream and downstream logic. At the same time, the Yanhe controlling gate in the research section was selected as the representative. Under the basic stable state of operation within 48 hours, the water level data sequence in front of the gate every 2 hours was analyzed, and the jump data was automatically identified and reasonably corrected.The results showed that the model established could automatically identify the pathological water regime data and carried out intelligent cleaning. The processed data was able to better meet the needs of water transfer scheduling analysis and decision-making. So the model has the value of popularization and application.